Docs

Full API reference for the Alphio open platform.

Getting started

Quick start

1.Create an API key in API Key. Copy it once — we hash it before storage.
2.Onboard your end users: UpsertVirtualUser → CreateChatSession → /openchat-steam/completions
3.Wire your SSE reader to handle submitted → delta* → done events.

curl https://api.alphio.ai/v1/openapi/users \
  -H "Authorization: Bearer pk_live_••••••••••••••••••••••••••••" \
  -H "Content-Type: application/json" \
  -d '{"partner_user_ref": "u_998877"}'

Base URL & envelope

Base URL: https://api.alphio.ai
Response envelope: { ret, msg, data }
HTTP status: Always 200 (for non-SSE). Inspect ret for the business result.
SSE: Chat completions stream as text/event-stream.

Authentication

Every request carries a Bearer API key. End-user-scoped endpoints additionally need an X-Partner-User header.

Header	Value	When
Authorization	Bearer pk_live_…	Every request
Content-Type	application/json	Every request
X-Partner-User	your_user_id	Only for end-user-scoped endpoints: /openchat-steam/completions + whales/following/ranking
Accept	text/event-stream	SSE chat only

The value of X-Partner-User is your own user ID. The platform maps it to a stable internal_user_id like op_{partner_code}_{nanoid}. The user must first be registered via UpsertVirtualUser or chat returns ret=10404.

Errors & limits

Error codes

HTTP is always 200; check ret

ret	Meaning	Typical cause
0	OK	Success
10400	Bad request	Missing required field / type mismatch / empty message
10401	Unauthenticated	Missing / wrong / expired / revoked Bearer
10403	Forbidden	User disabled / cross-user access
10404	Not found	User not registered (call UpsertVirtualUser) / bad session_id
10429	Rate limited	RPM quota exceeded
10500	Server error	Contact support with occurred_at / chat_id
10501	Insufficient balance	Top up to resume chat

Rate limits & billing

Default ceilings: 600 partner RPM · 60 per-user RPM · 8 concurrent chat streams. Negotiated on a per-plan basis.

Billing: one chat round (submit → done) deducts one billable unit. Interrupted / disconnected sessions still bill (anti-spam). failed events do not bill. Analysis, Whales, and Data endpoints are free of charge.

Virtual users

Manage end-user identities under your partner account. Virtual users are created lazily; the same partner_user_ref always maps to the same internal_user_id.

POST/v1/openapi/users

UpsertVirtualUser

Request parameters

Field	Type	Required	Description
partner_user_ref	string	yes	Your own user ID for this end user.
display_name	string	no	Optional, for audit/debugging.
metadata	string	no	Opaque JSON string, not parsed by the platform.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	VirtualUser	Response payload.
└internal_user_id	string	Platform-assigned ID. Format: "op_{partner_code}_{nanoid12}".
└partner_user_ref	string	Echoed back.
└display_name	string	Echoed back.
└metadata	string	Echoed back (JSON string).
└status	string	"active" / "disabled".
└is_new	bool	true only on first creation.
└created_at	string	ISO8601 UTC.
└last_active_at	string	ISO8601 UTC. Empty if never used.

Example request

curl -X POST https://api.alphio.ai/v1/openapi/users \
  -H "Authorization: Bearer pk_live_••••••••••••••••••••••••••••" \
  -H "Content-Type: application/json" \
  -d '{
    "partner_user_ref": "u_998877",
    "display_name": "Alice"
  }'

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "internal_user_id": "op_intellectia_qUk8CyDetjsK",
    "partner_user_ref": "u_998877",
    "display_name": "Alice",
    "metadata": "",
    "status": "active",
    "is_new": true,
    "created_at": "2026-06-09T08:02:14Z",
    "last_active_at": ""
  }
}

POST/v1/openapi/users/get

GetVirtualUser

Look up a virtual user by partner_user_ref. Response shape matches UpsertVirtualUser except is_new is always false.

Request parameters

Field	Type	Required	Description
partner_user_ref	string	yes	—

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	VirtualUser	Response payload.
└internal_user_id	string	Platform-assigned ID. Format: "op_{partner_code}_{nanoid12}".
└partner_user_ref	string	Echoed back.
└display_name	string	Echoed back.
└metadata	string	Echoed back (JSON string).
└status	string	"active" / "disabled".
└is_new	bool	true only on first creation.
└created_at	string	ISO8601 UTC.
└last_active_at	string	ISO8601 UTC. Empty if never used.

POST/v1/openapi/users/disable

DisableVirtualUser

Disable a virtual user. Their chat requests will return ret=10403 until re-enabled via UpsertVirtualUser.

Request parameters

Field	Type	Required	Description
partner_user_ref	string	yes	—

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	DisableVirtualUserData	Response payload.
└status	string	Always "disabled".

POST/v1/openapi/users/list

ListVirtualUsers

Paginated list of your virtual users.

Request parameters

Field	Type	Required	Description
filter	VirtualUserListFilter	no	Optional filters.
└status	string	no	"active" / "disabled" / "" (all).
└created_after	string	no	ISO8601 UTC.
└created_before	string	no	ISO8601 UTC.
page	int32	no	1-indexed. Default 1.
size	int32	no	Default 20, max 200.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	ListVirtualUsersData	Response payload.
└items	VirtualUser[]	—
└internal_user_id	string	Platform-assigned ID. Format: "op_{partner_code}_{nanoid12}".
└partner_user_ref	string	Echoed back.
└display_name	string	Echoed back.
└metadata	string	Echoed back (JSON string).
└status	string	"active" / "disabled".
└is_new	bool	true only on first creation.
└created_at	string	ISO8601 UTC.
└last_active_at	string	ISO8601 UTC. Empty if never used.
└pagination	Pagination	—
└page	int32	1-indexed.
└size	int32	Items per page.
└total	int64	Total rows matching the filter.
└has_more	bool	Whether more pages exist.

Chat (SSE streaming)

Multi-turn AI chat with built-in skills (technical analysis, market data, news). Response is text/event-stream — every event is emitted as `event: NAME\ndata: JSON\n\n`. Unlike all other endpoints, the SSE stream is NOT wrapped in `{ret, msg, data}`.

POST/openchat-steam/completionsX-Partner-User required

ChatCompletions

Start a chat turn. Returns an SSE stream until done or failed.

Each delta.content is incremental — append to a client buffer. The final done event carries the full content as a sanity check.

Failed events do not bill. Submitted → done is one billable round.

Request parameters

Field	Type	Required	Description
chat_id	string	yes	Your generated unique ID for this turn, e.g. chat-{uuid}. Used for idempotency and tracing.
message	string	yes	The user's question.
metadata	ChatMetadata	yes	—
└session_id	string	yes	From CreateChatSession. Wrong/missing → ret=10404.
└timezone	string	no	IANA timezone, e.g. "Asia/Shanghai". Affects "today/yesterday" interpretation.
└language	string	no	"zh" / "en" / etc. Default: adaptive to user language.

Response fields

Field	Type	Description
event: submitted	SSE	Acknowledgement that the request was accepted.
└chat_id	string	Echoed back.
└status	string	"RUNNING".
└user_id	string	internal_user_id resolved from X-Partner-User.
└session_id	string	Echoed back.
└occurred_at	string	ISO8601 UTC.
event: status	SSE	Intermediate progress.
└stage	string	agent_round_start / tool_call_start / model_call_start / etc.
└message	string	Human-readable status.
└progress	float	0.0 – 1.0.
event: delta	SSE	Streaming answer chunk.
└content	string	Incremental fragment — append to a client buffer.
└status	string	"RUNNING".
event: done	SSE	Stream complete.
└content	string	Full final answer (sanity check against the joined deltas).
└status	string	"SUCCESS".
└progress	float	1.0.
└token_usage	object	Token counters for billing visibility.
event: failed	SSE	Terminal failure (insufficient balance, backend error, etc).
└message	string	Error description.
└status	string	"FAILED".

Example request

curl -N -X POST https://api.alphio.ai/openchat-steam/completions \
  -H "Authorization: Bearer pk_live_••••••••••••••••••••••••••••" \
  -H "X-Partner-User: u_998877" \
  -H "Accept: text/event-stream" \
  -H "Content-Type: application/json" \
  -d '{
    "chat_id": "chat-998877-1719283400",
    "message": "What is AAPL'\''s current RSI?",
    "metadata": {
      "session_id": "sess_ifm92XaV7gqx",
      "timezone": "America/New_York",
      "language": "en"
    }
  }'

Example response

event: submitted
data: {"chat_id":"chat-998877-1719283400","status":"RUNNING",
       "user_id":"op_intellectia_qUk8CyDetjsK","session_id":"sess_ifm92XaV7gqx",
       "occurred_at":"2026-06-09T08:18:24.77Z"}

event: status
data: {"chat_id":"...","stage":"agent_round_start",
       "status":"PROCESSING","progress":0.15,"message":"Understanding question"}

event: status
data: {"chat_id":"...","stage":"tool_call_start",
       "status":"PROCESSING","progress":0.50,"message":"Fetching technical indicators"}

event: delta
data: {"chat_id":"...","status":"RUNNING","content":"AAPL's current RSI(14) is 68,"}

event: delta
data: {"chat_id":"...","status":"RUNNING","content":" approaching the overbought zone (70)..."}

event: done
data: {"chat_id":"...","status":"SUCCESS","progress":1.0,
       "stage":"task_completed","content":"AAPL's current RSI(14) is 68, approaching..."}

Chat sessions & messages

A session groups related chat turns. History within a session is automatically injected as context.

POST/v1/openapi/chat/sessions/create

CreateChatSession

Request parameters

Field	Type	Required	Description
internal_user_id	string	yes	From UpsertVirtualUser.
title	string	no	Display title for your UI.
metadata	string	no	Opaque JSON string.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	ChatSession	Response payload.
└session_id	string	Format: "sess_{nanoid}".
└internal_user_id	string	Owner.
└title	string	—
└metadata	string	JSON string.
└status	string	"active" / "deleted".
└last_message_at	string	ISO8601 UTC. Empty for a freshly created session.
└created_at	string	ISO8601 UTC.
└updated_at	string	ISO8601 UTC.

POST/v1/openapi/chat/sessions/list

ListChatSessions

Request parameters

Field	Type	Required	Description
internal_user_id	string	no	Empty = list across all your virtual users.
status	string	no	"active" / "deleted" / "" (all). Default "active".
page	int32	no	Default 1.
size	int32	no	Default 20, max 200.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	ListChatSessionsData	Response payload.
└items	ChatSession[]	—
└session_id	string	Format: "sess_{nanoid}".
└internal_user_id	string	Owner.
└title	string	—
└metadata	string	JSON string.
└status	string	"active" / "deleted".
└last_message_at	string	ISO8601 UTC. Empty for a freshly created session.
└created_at	string	ISO8601 UTC.
└updated_at	string	ISO8601 UTC.
└page	int32	1-indexed.
└size	int32	Items per page.
└total	int64	Total rows matching the filter.
└has_more	bool	Whether more pages exist.

POST/v1/openapi/chat/sessions/get

GetChatSession

Request parameters

Field	Type	Required	Description
session_id	string	yes	—

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	ChatSession	Response payload.
└session_id	string	Format: "sess_{nanoid}".
└internal_user_id	string	Owner.
└title	string	—
└metadata	string	JSON string.
└status	string	"active" / "deleted".
└last_message_at	string	ISO8601 UTC. Empty for a freshly created session.
└created_at	string	ISO8601 UTC.
└updated_at	string	ISO8601 UTC.

POST/v1/openapi/chat/sessions/update

UpdateChatSession

Edit the title or metadata of an existing session.

Request parameters

Field	Type	Required	Description
session_id	string	yes	—
title	string	no	Empty = unchanged.
metadata	string	no	Empty = unchanged.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	ChatSession	Response payload.
└session_id	string	Format: "sess_{nanoid}".
└internal_user_id	string	Owner.
└title	string	—
└metadata	string	JSON string.
└status	string	"active" / "deleted".
└last_message_at	string	ISO8601 UTC. Empty for a freshly created session.
└created_at	string	ISO8601 UTC.
└updated_at	string	ISO8601 UTC.

POST/v1/openapi/chat/sessions/delete

DeleteChatSession

Soft-delete by default. Message history remains queryable.

Request parameters

Field	Type	Required	Description
session_id	string	yes	—
soft_delete	bool	no	Default true. Set false for permanent deletion.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	DeleteChatSessionData	Response payload.
└status	string	"deleted" / "hard_deleted".

POST/v1/openapi/chat/sessions/messages

ListSessionMessages

Cursor-paginated history. Order is by created_at DESC.

Request parameters

Field	Type	Required	Description
session_id	string	yes	—
before_message_id	string	no	Cursor — pass the message_id of your current oldest message to fetch older ones.
size	int32	no	Default 50, max 200.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	ListSessionMessagesData	Response payload.
└items	ChatMessage[]	—
└message_id	string	Format: "msg_{nanoid}".
└chat_id	string	Matches the SSE chat_id. One chat_id usually produces 2 messages (user + assistant).
└session_id	string	May be empty for session-less chat.
└internal_user_id	string	—
└role	string	"user" / "assistant".
└content	string	Raw Markdown stream. May contain ```json card fences.
└attachments	ChatAttachment[]	—
└url	string	Pre-signed URL. May expire.
└mime	string	—
└name	string	—
└tool_calls	ChatToolCall[]	—
└step_id	string	—
└kind	string	"round" / "tool_call".
└round	int32	—
└tool_name	string	—
└args_json	string	Stringified tool arguments.
└result_summary	string	—
└status	string	"ok" / "error" / "running".
└started_at_ms	int64	Unix milliseconds.
└ended_at_ms	int64	Unix milliseconds.
└sources	ChatSource[]	—
└title	string	—
└url	string	—
└snippet	string	—
└error_code	string	Non-empty on terminal failure.
└error_message	string	—
└doc_id	string	Chatbot's assistant doc_id.
└created_at	string	ISO8601 UTC.
└has_more	bool	—

POST/v1/openapi/chat/messages/get

GetChatMessage

Request parameters

Field	Type	Required	Description
message_id	string	yes	—

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	ChatMessage	Response payload.
└message_id	string	Format: "msg_{nanoid}".
└chat_id	string	Matches the SSE chat_id. One chat_id usually produces 2 messages (user + assistant).
└session_id	string	May be empty for session-less chat.
└internal_user_id	string	—
└role	string	"user" / "assistant".
└content	string	Raw Markdown stream. May contain ```json card fences.
└attachments	ChatAttachment[]	—
└url	string	Pre-signed URL. May expire.
└mime	string	—
└name	string	—
└tool_calls	ChatToolCall[]	—
└step_id	string	—
└kind	string	"round" / "tool_call".
└round	int32	—
└tool_name	string	—
└args_json	string	Stringified tool arguments.
└result_summary	string	—
└status	string	"ok" / "error" / "running".
└started_at_ms	int64	Unix milliseconds.
└ended_at_ms	int64	Unix milliseconds.
└sources	ChatSource[]	—
└title	string	—
└url	string	—
└snippet	string	—
└error_code	string	Non-empty on terminal failure.
└error_message	string	—
└doc_id	string	Chatbot's assistant doc_id.
└created_at	string	ISO8601 UTC.

POST/v1/openapi/chat/messages/rate

RateChatMessage

Up/down vote an assistant reply. Only valid for role=assistant messages.

Request parameters

Field	Type	Required	Description
message_id	string	yes	—
rating	string	yes	"up" or "down".
note	string	no	Optional user note.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	RateChatMessageData	Response payload.
└message_id	string	Echoed back.
└rating	string	Echoed back.
└created_at	string	ISO8601 UTC.

POST/v1/openapi/chat/recommend-questions

ChatRecommendQuestions

Suggested starter questions for your chat entry screen, regenerated daily from a digest of recent market news. Served from cache — this endpoint never invokes an LLM and is not billed. Each call returns a fresh random subset, so repeated calls give your UI variety.

Not billed and no X-Partner-User required.

An empty questions array with ret=0 means the daily batch is not available yet (or expired) — hide the suggestions UI rather than treating it as an error.

Request parameters

Field	Type	Required	Description
lang	string	no	Language tag: "en" / "en-us", "zh" / "zh-cn" / "zh-hans", "zh-tw" / "zh-hant" / "zh-hk". Case-insensitive. Empty or unrecognized falls back to "en".
count	int32	no	Max questions to return. Default 10, max 50.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	ChatRecommendQuestionsData	Response payload.
└questions	string[]	Randomly ordered subset of the day's batch.
└lang	string	Language actually served after normalization/fallback: "en" / "zh-cn" / "zh-tw".
└generated_at	int64	Unix seconds when the batch was generated.

Example request

curl -X POST https://api.alphio.ai/v1/openapi/chat/recommend-questions \
  -H "Authorization: Bearer pk_live_••••••••••••••••••••••••••••" \
  -H "Content-Type: application/json" \
  -d '{
    "lang": "en",
    "count": 3
  }'

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "questions": [
      "How could the latest Fed commentary affect rate-cut expectations this summer?",
      "Is NVDA's pullback after earnings a buying opportunity?",
      "What is driving the rally in gold prices this week?"
    ],
    "lang": "en",
    "generated_at": 1781136000
  }
}

Pivot point

Pivot point / support-resistance signals.

POST/v1/openapi/analysis/pivot-point/info

PivotPointInfo

Returns every 神奇九转 + RSI(6) pivot signal (DIP / ALERT) the algorithm fired within the look-back window. The K-line period is selectable with multiplier / timespan (1–720 minute, hour, day, week, month, quarter, year); both omitted = daily bars over the trailing 1 year, the behaviour that predates the fields. Computed server-side from datafabric kline (US / HK / crypto all supported); cached 5 minutes, 1 minute for intraday periods. Empty `data.list` with `ret=0` means no signal in the window — distinct from a real error. A window longer than the period's max span comes back one page at a time — pass `data.next_cursor` back as `cursor` to walk further back.

Request parameters

Field	Type	Required	Description
code	string	yes	Ticker, e.g. "AAPL", "00700", "BTC". Case-insensitive; uppercased server-side.
asset_type	int32	yes	0=stock, 1=etf, 2=crypto.
locale	string	no	Market the ticker belongs to: "US", "HK", "FOREX", "METAL". Case-insensitive; uppercased server-side. Empty defaults to "US". Crypto (asset_type=2) keeps the field for shape consistency but it's inert.
multiplier	int32	no	Period count, paired with timespan. Allowed values depend on timespan: minute 1–720, hour 1–12, month 1 / 3 (= quarter) / 12 (= year); day, week, quarter and year accept 1 only. Any other combination is a parameter error rather than a silently mis-aggregated series. Default 1.
timespan	int32	no	K-line period unit: 2=minute, 3=hour, 4=day, 5=week, 6=month, 7=quarter, 8=year (1=second is not supported). Omitted together with multiplier = daily. Bar stamping — day and above carry the period's FIRST TRADING DAY (NVDA's August 2026 monthly bar is 2026-08-03, its Q1 and 2026 yearly bars are both 2026-01-02); minute and hour bars are right-edge stamped and anchored to each trading session's open, truncated at the close (HK 60m → 10:30 / 11:30 / 12:00 / 14:00 / 15:00 / 16:00, HK 90m → 11:00 / 12:00 / 14:30 / 16:00, and a 3m series stamps the 09:30 bar 09:33). This anchoring is done by THIS endpoint, which folds hourly widths from 5-minute bars for the purpose; /v1/openapi/data/kline passes timespan straight through to the vendor and returns its CLOCK-aligned hourly stamps (10:00, 11:00) instead — so a 60m bar read from the two endpoints carries different timestamps by design. Crypto and perp serve only the exchange's native widths: 1 / 3 / 5 / 15 / 30 / 60 / 120 / 240 / 360 / 480 / 720 minutes, plus day / week / month. Note the upstream vendor stamps week and above on the CALENDAR period start (August 2026 arrives as 08-01); this endpoint moves them onto the first day the market actually traded, so those stamps can differ from the same bar read through /v1/openapi/data/kline.
from	string	no	Window start, YYYY-MM-DD (UTC, inclusive). Must be set together with to. The algorithm runs on exactly the [from, to] bars with no warmup preload — the signal math is window-sensitive (RSI warmup, TD-9 counts, FILTER suppression), so to reproduce the App chart's markers pass exactly the bar range the chart has loaded. Both empty = a default look-back sized for the period: 1 year for daily bars, days for a 1-minute series, decades for yearly.
to	string	no	Window end, YYYY-MM-DD (UTC, inclusive). Must be set together with from. Max span 5 years for minute / hour / day / week, 50 years for month / quarter / year. A LONGER window is not rejected — it is clamped to the ceiling, keeping `to` and moving `from` forward, and `msg` says which window actually ran (e.g. "ok (requested from 2020-12-07 exceeds the max span for this period; window clamped to 2021-08-27..2026-08-31)"). Windows shorter than 30 bars return "insufficient kline history" — which also means yearly signals only exist for tickers with 30+ years of history. Ignored when `cursor` is set — the cursor decides where the page ends.
cursor	int64	no	Backward paging cursor, unix seconds. One request only computes the bars inside the max span for the period, so a longer history comes back one page at a time: take `data.next_cursor` from the response and pass it here to get the page immediately BEFORE it — the bars with ts strictly less than cursor, one max-span window wide, never earlier than `from`. `to` is ignored when cursor is set; `from` still bounds how far back paging can go. Pages are computed independently yet agree bar-for-bar with one big window: the server preloads ~100 extra bars before each page start to warm RSI / TD-9 / COUNT up and returns only the signals inside the page, so concatenating pages neither drops nor duplicates a signal at a boundary. Omit or 0 = first page.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisPivotPointInfoData	Response payload.
└list	AnalysisPivotAlertInfo[]	Filtered signals in CHRONOLOGICAL ASCENDING order. Empty when no signal fired in the look-back window. For the latest signal, take list[list.length-1].
└date	string	Bar label in the market's own timezone (UTC for crypto / perp): "YYYY-MM-DD HH:MM" for minute and hour periods, "YYYY-MM-DD" for day and above.
└code	string	Echoed ticker, uppercased.
└sign	string	"DIP" (bullish reversal) or "ALERT" (bearish reversal).
└mark	string	"Buy Signal" / "Sell Signal" — human-readable label.
└price	string	Display price with 4-decimal precision. DIP = low × 0.98 (paint below the bar); ALERT = high × 1.03 (paint above).
└ts	int64	Bar timestamp, unix seconds — the same instant date renders.
└next_cursor	int64	Pass it back as `cursor` to get the page before this one. Unix seconds — the ts of the OLDEST bar on this page. 0 means there is no earlier page: paging has reached `from`, fewer than 30 bars remain before it, or this first page already covered the whole requested window.

Example request

# Daily (period omitted — the pre-existing behaviour)
POST https://api.alphio.ai/v1/openapi/analysis/pivot-point/info
{ "code": "AAPL", "asset_type": 0, "locale": "US", "from": "2026-01-01", "to": "2026-07-28" }

# 60-minute bars
POST https://api.alphio.ai/v1/openapi/analysis/pivot-point/info
{ "code": "AAPL", "asset_type": 0, "locale": "US", "multiplier": 60, "timespan": 2 }

# Older signals — feed the previous response's next_cursor back as cursor
POST https://api.alphio.ai/v1/openapi/analysis/pivot-point/info
{ "code": "AAPL", "asset_type": 0, "locale": "US", "multiplier": 1, "timespan": 2, "cursor": 1789394400 }

Example response

// Daily
{
  "ret": 0, "msg": "ok",
  "data": {
    "list": [
      {
        "date": "2026-02-09", "code": "AAPL",
        "sign": "ALERT", "mark": "Sell Signal",
        "price": "286.5460", "ts": 1770613200
      },
      {
        "date": "2026-06-09", "code": "AAPL",
        "sign": "DIP", "mark": "Buy Signal",
        "price": "291.7340", "ts": 1781007137
      }
    ]
  }
}

// 60-minute — note date carries the time of day
{
  "ret": 0, "msg": "ok",
  "data": {
    "list": [
      {
        "date": "2026-08-24 11:30", "code": "AAPL",
        "sign": "ALERT", "mark": "Sell Signal",
        "price": "286.5460", "ts": 1787585400
      },
      {
        "date": "2026-08-27 14:30", "code": "AAPL",
        "sign": "DIP", "mark": "Buy Signal",
        "price": "291.7340", "ts": 1787855400
      }
    ]
  }
}

// 1-minute with cursor — next_cursor is the oldest bar on this page; 0 = no more pages
{
  "ret": 0, "msg": "ok",
  "data": {
    "list": [
      {
        "date": "2026-09-11 10:31", "code": "AAPL",
        "sign": "DIP", "mark": "Buy Signal",
        "price": "289.4100", "ts": 1789137060
      }
    ],
    "next_cursor": 1789051500
  }
}

POST/v1/openapi/analysis/pivot-point/signals

PivotPointSignals

Bulk scan / signal screener.

Request parameters

Field	Type	Required	Description
signal_type	int32	yes	0=all, 1=bottom (dip / bottom-reversal), 2=alert (top warning).
page	int32	no	1-indexed.
page_size	int32	no	Items per page.
asset_type	string	no	Numeric string: "0"=stock, "1"=etf, "2"=crypto. Empty = all.
period	int32	no	0=all, 1=15min, 2=30min, 3=1h, 4=2h, 5=4h. This is the screener's own enum — NOT the timespan / multiplier pair PivotPointInfo takes, and it has no daily or above option.
sort_field	int32	no	0=timestamp, 2=volume (1=market_cap is deprecated).
order	string	no	"asc" (default) / "desc".
first_id	string	no	Forward cursor.
last_id	string	no	Backward cursor.
refresh	bool	no	Force-bust cache.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisPivotPointSignalsData	Response payload.
└signals	AnalysisDayTradingSignal[]	—
└symbol	string	—
└code	string	—
└name	string	Company name.
└logo	string	—
└price	string	—
└change_ratio	string	—
└timestamp	int64	Unix seconds.
└signal_types	int32[]	Active signal labels.
└period	int32	—
└klines	AnalysisKlinePoint[]	OHLCV + technicals.
└open	double	—
└high	double	—
└low	double	—
└close	double	—
└volume	double	—
└timestamp	int64	Unix seconds.
└ma5	double	—
└ma20	double	—
└rsi6	double	—
└macd	double	—
└macd_signal	double	—
└macd_hist	double	—
└asset_type	int32	—
└volume	string	—
└id	string	Stable row ID for cursor pagination.
└prev_close	string	—
└total	int64	Total rows matching the filter.
└new_add	int64	Rows newly added since last query.

Swingmax (US)

Swing-trade signals for US equities.

POST/v1/openapi/analysis/swingmax/signals

SwingmaxSignals

Query open swingmax signals for the given tickers. Returns the SAME rich item shape as /swingmax/stocks (AnalysisSwingmaxStock — with kline, company name/logo, the r1/r2/r3 target ladder, stop loss). Ladder field names match the App swing-max/list payload one-to-one.

Lookup by id since 2026-09-15: pass ids (max 20) instead of tickers to fetch specific signals, including ones that have closed since — the tickers path stays open-only. Every item now carries id; it is assigned at entry and never changes, so store it from any response and pass it back later.

The status field is localized. A "lang" BODY PARAM takes priority when set; otherwise the "lang" REQUEST HEADER is used: omit both for English, "zh" / "zh-cn" for Simplified Chinese, "zh-tw" for Traditional Chinese. Every other field (prices, timestamps) is language-independent.

Request parameters

Field	Type	Required	Description
tickers	AnalysisAsset[]	no	Tickers to query. Send either tickers or ids — one of them is required; both together is InvalidArgument.
└asset_type	int32	no	0=stock, 1=etf, 2=crypto.
└symbol	string	no	Legacy field.
└ticker	string	no	Preferred field.
└locale	string	no	"US" / "HK".
ids	int64[]	no	Signal ids to look up — the id field of a returned item. At most 20; more is InvalidArgument, NOT truncated, so a silently missing 21st id cannot happen. Mutually exclusive with tickers. Unlike tickers, which only sees the open pool, ids also returns positions that have closed since (exit_time_ts > 0), so an id taken from an earlier response can be followed until the trade ends. Unknown ids are skipped, not errored. Added 2026-09-15.
lang	string	no	Localizes the status text. When set it takes PRIORITY over the "lang" request header: "zh" / "zh-cn" = Simplified Chinese, "zh-tw" = Traditional Chinese. Omit/empty to fall back to the header (English by default).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisSwingmaxSignalsData	Response payload.
└signals	AnalysisSwingmaxStock[]	Same item shape as /swingmax/stocks — one entry per matched open ticker.
└id	int64	Signal id. Assigned when the position opens and unchanged for its whole life, including after it closes. Store it and pass it back as ids on /swingmax/signals to follow one trade. Added 2026-09-15. Filled on /swingmax/signals and on /swingmax/stocks for every stype: since 2026-09-16 the closed list (stype≠1) reads the same v2 store as the open pool, so an id taken from any /swingmax/stocks row round-trips through ids on /swingmax/signals. Before that date closed rows carried 0.
└symbol	string	—
└code	string	—
└company_name	string	—
└logo	string	—
└return	double	Realized return since entry, percent. Weighted over the ladder: 40% sold at Target 2 + 30% sold at Target 3 + the remaining 30% at final exit (unsold parts marked to the latest close).
└entry_price	double	—
└exit_price	double	Final-exit fill price for the remaining position. 0 while the position is open.
└entry_time	string	Entry date string from upstream, e.g. "2026-06-02".
└exit_time	string	Exit date string; empty while the position is open.
└change_ratio	double	Not populated since 2026-09-16 (the closed list moved to the v2 store, which has no such column); always 0. Use return instead.
└kline	AnalysisKlinePointV2[]	Snapshot bars with full OHLC + volume — enough to draw candlesticks.
└close	double	—
└timestamp	int64	Unix seconds.
└open	double	—
└high	double	—
└low	double	—
└volume	double	—
└entry_time_ts	int64	Unix seconds.
└exit_time_ts	int64	Unix seconds; 0 while open.
└target_price	string	Legacy alias of r2_price (Target 2), kept for backward compatibility — prefer the r1/r2/r3_price ladder.
└stop_loss	string	Stop-loss price. The whole remaining position exits if it is hit.
└interval	int32	Signal timeframe: 0=1d, 1=15min, 2=30min, 3=1h, 4=2h, 5=4h.
└side	int32	0=long, 1=short.
└market_cap	string	—
└current_price	string	—
└r1_price	string	Target 1 price — the first (lowest) pivot level above entry. A price milestone only: no partial sell is executed at R1.
└r1_signal_ts	int64	Reserved for shape parity with the App swing-max/list; always 0 on this stock feed (only the crypto pipeline fills R1 triggers).
└r2_price	string	Target 2 price. When first reached, the strategy sells 40% of the position. Same value as the legacy target_price.
└r2_signal_ts	int64	Unix seconds when the price first reached Target 2 (40% partial exit executed); 0 if not reached yet.
└r2_sell_price	string	Fill price of the 40% partial exit at Target 2; empty if not reached yet.
└r3_price	string	Target 3 price — the highest ladder level. When first reached, the strategy sells another 30% of the position.
└r3_signal_ts	int64	Unix seconds when the price first reached Target 3 (30% partial exit executed); 0 if not reached yet.
└r3_sell_price	string	Fill price of the 30% partial exit at Target 3; empty if not reached yet.
└status	string	Server-rendered progress narrative (entry → 40% partial exit at Target 2 → 30% at Target 3 → final close) — same text the App shows on swing-max/list. Localized by the "lang" body param (priority) or "lang" request header: omit both for English, "zh" / "zh-cn" = Simplified Chinese, "zh-tw" = Traditional Chinese. Empty on the closed list (stype≠1).

Example request

POST https://api.alphio.ai/v1/openapi/analysis/swingmax/signals
Authorization: Bearer pk_live_••••••••••••••••••••••••••••
{
  "tickers":[
    {"asset_type":0,"ticker":"AAPL","locale":"US"},
    {"asset_type":0,"ticker":"NVDA","locale":"US"}
  ],
  "lang":"zh-tw"
}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "signals":[
      {"symbol":"AAPL","code":"AAPL.O","company_name":"Apple Inc","logo":"https://...",
       "return":3.21,"entry_price":187.2,"exit_price":0,"entry_time":"2026-06-02","exit_time":"",
       "change_ratio":3.21,
       "kline":[{"close":190.5,"timestamp":1717286400,"open":188.9,"high":191.2,"low":188.1,"volume":51234567}],
       "entry_time_ts":1717286400,"exit_time_ts":0,"target_price":"210.5","stop_loss":"177.84",
       "interval":0,"side":0,"market_cap":"2.9T","current_price":"193.1",
       "r1_price":"198.6","r1_signal_ts":0,
       "r2_price":"210.5","r2_signal_ts":1719878400,"r2_sell_price":"211.02",
       "r3_price":"224.9","r3_signal_ts":0,"r3_sell_price":"",
       "status":"Booked partial profits @211.02 on 2026-07-01 09:35 ET and riding runners"}
    ]
  }
}

POST/v1/openapi/analysis/swingmax/stocks

SwingmaxStocks

Browse the supported universe. NOTE: only stype=1 returns the OPEN positions pool; any other value returns the closed list. Since 2026-09-16 both come from the same v2 store: closed rows carry id, the r1/r2/r3 ladder fields and status like open rows, and return is the realized ladder return. Two fields differ from the pre-2026-09-16 closed list: change_ratio is no longer populated (0), and side=1 (short) returns an empty list because the v2 store is long-only. Before 2026-09-16 the closed list came from an older store that had stopped updating in 2025-08 and carried no id.

data.total is the count AFTER keyword filtering and BEFORE paging — divide it by size to get the page count. Paging past the last page returns an empty list with total unchanged, which is the normal way to detect the end, not an error.

Request parameters

Field	Type	Required	Description
stype	int32	yes	GetSwingMaxStocksType. 1=open pool, 2=removed(closed), 3=today, 4=today-open, 5=today-update. Only stype=1 = open pool; 2-5 all return the same closed list (v2 store, rows carry id).
sort_type	int32	no	1-4 match the platform-wide meaning shared with every other list endpoint: 1=latest, 2=oldest, 3=highest return, 4=lowest return. 5-11 are extra orderings only this endpoint supports: 5=latest exit, 6=oldest exit, 7=latest entry, 8=oldest entry, 9=highest market cap, 10=lowest market cap, 11=highest potential.
page	int64	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Default 20, max 20 — larger values are clamped, not rejected. The same ceiling as every other paged list on this API; only the two swingmax /open endpoints are uncapped. Before 2026-09-15 the default was 100 with no clamp, so omitting size failed with a parameter error from upstream.
keyword	string	no	Case-insensitive substring match on ticker, code and company name. Omit to disable filtering. Max 64 characters.
side	int32	no	0=long, 1=short.
trade_status	int32	no	0=not traded, 1=traded, 2=all.
lang	string	no	Localizes the status text. When set it takes PRIORITY over the "lang" request header: "zh" / "zh-cn" = Simplified Chinese, "zh-tw" = Traditional Chinese. Omit/empty to fall back to the header (English by default).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisSwingmaxStocksData	Response payload.
└stocks	AnalysisSwingmaxStock[]	—
└id	int64	Signal id. Assigned when the position opens and unchanged for its whole life, including after it closes. Store it and pass it back as ids on /swingmax/signals to follow one trade. Added 2026-09-15. Filled on /swingmax/signals and on /swingmax/stocks for every stype: since 2026-09-16 the closed list (stype≠1) reads the same v2 store as the open pool, so an id taken from any /swingmax/stocks row round-trips through ids on /swingmax/signals. Before that date closed rows carried 0.
└symbol	string	—
└code	string	—
└company_name	string	—
└logo	string	—
└return	double	Realized return since entry, percent. Weighted over the ladder: 40% sold at Target 2 + 30% sold at Target 3 + the remaining 30% at final exit (unsold parts marked to the latest close).
└entry_price	double	—
└exit_price	double	Final-exit fill price for the remaining position. 0 while the position is open.
└entry_time	string	Entry date string from upstream, e.g. "2026-06-02".
└exit_time	string	Exit date string; empty while the position is open.
└change_ratio	double	Not populated since 2026-09-16 (the closed list moved to the v2 store, which has no such column); always 0. Use return instead.
└kline	AnalysisKlinePointV2[]	Snapshot bars with full OHLC + volume — enough to draw candlesticks.
└close	double	—
└timestamp	int64	Unix seconds.
└open	double	—
└high	double	—
└low	double	—
└volume	double	—
└entry_time_ts	int64	Unix seconds.
└exit_time_ts	int64	Unix seconds; 0 while open.
└target_price	string	Legacy alias of r2_price (Target 2), kept for backward compatibility — prefer the r1/r2/r3_price ladder.
└stop_loss	string	Stop-loss price. The whole remaining position exits if it is hit.
└interval	int32	Signal timeframe: 0=1d, 1=15min, 2=30min, 3=1h, 4=2h, 5=4h.
└side	int32	0=long, 1=short.
└market_cap	string	—
└current_price	string	—
└r1_price	string	Target 1 price — the first (lowest) pivot level above entry. A price milestone only: no partial sell is executed at R1.
└r1_signal_ts	int64	Reserved for shape parity with the App swing-max/list; always 0 on this stock feed (only the crypto pipeline fills R1 triggers).
└r2_price	string	Target 2 price. When first reached, the strategy sells 40% of the position. Same value as the legacy target_price.
└r2_signal_ts	int64	Unix seconds when the price first reached Target 2 (40% partial exit executed); 0 if not reached yet.
└r2_sell_price	string	Fill price of the 40% partial exit at Target 2; empty if not reached yet.
└r3_price	string	Target 3 price — the highest ladder level. When first reached, the strategy sells another 30% of the position.
└r3_signal_ts	int64	Unix seconds when the price first reached Target 3 (30% partial exit executed); 0 if not reached yet.
└r3_sell_price	string	Fill price of the 30% partial exit at Target 3; empty if not reached yet.
└status	string	Server-rendered progress narrative (entry → 40% partial exit at Target 2 → 30% at Target 3 → final close) — same text the App shows on swing-max/list. Localized by the "lang" body param (priority) or "lang" request header: omit both for English, "zh" / "zh-cn" = Simplified Chinese, "zh-tw" = Traditional Chinese. Empty on the closed list (stype≠1).
└total	int32	—

POST/v1/openapi/analysis/swingmax/open

SwingmaxOpen

Returns every currently-open position in a single call by default. Static signals only — no kline, no realtime return, no company name/logo. Cheap, cached ~1 minute.

Request parameters

Field	Type	Required	Description
page	int32	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Omit to return every open position in one call — that is the historical behaviour of this endpoint and it is unchanged. Set it only if you want pages.
sort_type	int32	no	0 = endpoint default, 1 = latest first, 2 = oldest first, 3/4 = by potential_gain (target-based upside, NOT realised return), high/low first. The same number means the same thing on the US and HK endpoints.
keyword	string	no	Case-insensitive substring match on ticker and code — this endpoint carries no company name. Omit to disable filtering. Max 64 characters.
lang	string	no	Localizes server-rendered status text. Takes priority over the lang header.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisSwingmaxOpenData	Response payload.
└positions	AnalysisSwingmaxOpenPosition[]	Every currently-open position.
└id	int64	Signal id — the same number as id on /swingmax/signals and /swingmax/stocks. Take it from this list and pass it back as ids on /swingmax/signals to follow one position until it closes. Added 2026-09-15.
└symbol	string	—
└code	string	—
└buy_price	string	Entry / Buy Target.
└stop_loss	string	—
└target1	string	Sell Target 1 (r2).
└target2	string	Sell Target 2 (r3).
└potential_gain	double	Static: (r3/buy - 1) * 100, percent.
└signal_time	int64	Entry time, unix seconds.
└last_exit_ts	int64	Exit time, unix seconds; 0 when open.
└total	int32	—

Example request

POST https://api.alphio.ai/v1/openapi/analysis/swingmax/open
{}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "positions":[
      {"symbol":"GME","code":"GME.N","buy_price":"21.715","stop_loss":"20.629",
       "target1":"24.252","target2":"28.152","potential_gain":29.6,
       "signal_time":1647388800,"last_exit_ts":0}
    ],
    "total": 1
  }
}

Swingmax (HK)

Swing-trade signals for Hong Kong equities. Same engine and same r2/r3 target ladder as the US endpoints, but a separate item shape: HK carries board_lot and trade_currency, which are meaningless for US equities. Amounts are in trade_currency, not always HKD.

POST/v1/openapi/analysis/swingmax/hk/signals

SwingmaxHKSignals

Current open swingmax positions for the given HK tickers. Mirrors /swingmax/signals on the US side, including the fact that it returns OPEN positions only. Until 2026-09-08 this endpoint also returned closed positions — the status filter was missing, so querying two tickers could return a decade of history alongside the live ones. If you were relying on that to read past trades, switch to /swingmax/hk/stocks with stype=2 (closed); that is the endpoint built for it.

The lang BODY PARAM did nothing on the three HK swingmax endpoints before 2026-09-15 — only the header was honored, so a request with lang in the body and no header came back in English. Fixed; both now behave as documented on all three. Sending the header still works.

tickers is an array of objects since 2026-09-09, matching US /swingmax/signals field for field. The old plain string array ["00700"] still works — the gateway normalizes it to [{"ticker":"00700"}] — but it is a transitional shim for integrations completed before the change; new code should send objects.

The status field is localized. The "lang" body param takes priority when set; otherwise the "lang" request header is used: omit both for English, "zh" / "zh-cn" for Simplified Chinese, "zh-tw" for Traditional Chinese. Every other field (prices, timestamps) is language-independent.

Field names match US swingmax as of 2026-09-09: name is now company_name, buy_date/buy_price are entry_time/entry_price, sell_date/sell_price are exit_time/exit_price, and the r2_triggered/r3_triggered booleans are gone — use r2_signal_ts/r3_signal_ts > 0 instead. The booleans said that a target was hit but never when, so no marker could be placed on a chart. entry_time_ts and exit_time_ts are new. Also on 2026-09-09: profit_rate was renamed to return, and entry_price / exit_price / return became numbers instead of strings — all three now match US swingmax exactly. HK-specific fields (trade_currency, board_lot, realized, exit_reason, hold_days) are unchanged.

Open positions only — a ticker with no live position returns nothing here even if it traded many times before. Use /swingmax/hk/stocks (date range + stype) to browse history.

Bars arrive on this endpoint. They were empty from launch until 2026-09-07 — the field was in the response shape but nothing ever filled it, so callers got an empty array with no error and reasonably concluded HK had no bars. If you cached that conclusion, re-check.

Universe: ordinary shares and REITs only. Warrants, CBBCs, ETFs and LIPs are excluded — HK lists roughly 30k derivative warrants against 3k ordinary shares, so an unfiltered pool would be about 90% derivatives.

Additional filters: average daily turnover at least HKD 20M, close at least HKD 2, market cap at least HKD 3B. The price floor exists because the HK main board has no delisting rule equivalent to the US $1 threshold.

Amounts are in trade_currency, NOT always HKD. Check the field before aggregating across symbols.

Request parameters

Field	Type	Required	Description
tickers	AnalysisAsset[]	no	HK codes to query, 5 digits each — an array of OBJECTS, the same shape as US /swingmax/signals: [{"ticker":"00700"}]. asset_type and locale are accepted but inert here: the path already fixes the market. Send either tickers or ids — one of them is required; both together is InvalidArgument.
└asset_type	int32	no	0=stock, 1=etf, 2=crypto.
└symbol	string	no	Legacy field.
└ticker	string	no	Preferred field.
└locale	string	no	"US" / "HK".
ids	int64[]	no	Signal ids to look up — the id field of a returned item. At most 20; more is InvalidArgument, NOT truncated, so a silently missing 21st id cannot happen. Mutually exclusive with tickers. Unlike tickers, which only sees the open pool, ids also returns positions that have closed since (exit_time_ts > 0), so an id taken from an earlier response can be followed until the trade ends. Unknown ids are skipped, not errored. Added 2026-09-15.
lang	string	no	Localizes the status text. When set it takes priority over the "lang" request header: "zh" / "zh-cn" for Simplified Chinese, "zh-tw" for Traditional Chinese. Omit or leave empty to fall back to the header (English by default).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisSwingmaxHKSignalsData	Response payload.
└signals	AnalysisSwingmaxHKItem[]	One entry per matched ticker.
└id	int64	Signal id — same meaning as US swingmax id: assigned at entry, stable through close. Pass it back as ids on /swingmax/hk/signals to follow one trade. Added 2026-09-15. Also present on /swingmax/hk/open.
└symbol	string	HK code — always 5 digits, e.g. 00700.
└code	string	Symbol with the .HK suffix.
└company_name	string	Company name, in the language of the listing.
└trade_currency	string	HKD / CNY / USD. HK has three trading counters; the same company can list under more than one symbol, so amounts are NOT comparable across symbols without checking this. HK-only field.
└board_lot	int32	Shares per lot. HK trades in lots, not shares — 00700 is 100 shares/lot (~HKD 40k). 0 means the lot size could not be resolved. HK-only field.
└entry_time	string	Signal date, a Hong Kong trading day (YYYY-MM-DD).
└entry_time_ts	int64	Entry time, unix seconds — midnight HKT on entry_time. Always > 0. Use this to place the entry marker; it saves parsing the date string.
└entry_price	double	Buy target (breakout level), in trade_currency. The entry marker goes here. A number, not a string — matching US swingmax.
└stop_loss	string	Support level. Entry minus 15%; moves up to the entry price once r2 triggers.
└r2_price	string	Sell Target 1 (first resistance).
└r2_sell_price	string	Actual r2 fill; empty when not triggered. This is the price to mark, not r2_price.
└r2_signal_ts	int64	r2 trigger time, unix seconds. 0 means not triggered — draw the first target marker only when this is > 0. Same rule as US swingmax.
└r3_price	string	Sell Target 2 (second resistance).
└r3_sell_price	string	Actual r3 fill; empty when not triggered.
└r3_signal_ts	int64	r3 trigger time, unix seconds. 0 means not triggered — draw the second target marker only when this is > 0.
└exit_time	string	Exit date; empty while the position is still open.
└exit_time_ts	int64	Exit time, unix seconds. 0 while the position is open — this one field carries both when it exited and whether it has exited, exactly as in US swingmax. There is no exit_signal_ts field in either market.
└exit_price	double	Final exit price. The exit marker goes here. 0 while the position is open — check exit_time_ts, not this, to tell open from closed.
└return	double	Net return of the WHOLE position, decimal fraction. Same field, same meaning and same type as US swingmax return. It was called profit_rate until 2026-09-09.
└realized	string	Return already banked by the scale-out, decimal fraction. This — not return — is the win indicator: realized > 0 means the trade counted as a win, matching how US swingmax reports its win rate. The two numbers differ materially, so do not mix them. HK-only field.
└exit_reason	string	ALERT = stopped at the original -15% stop (the only losing bucket). BREAKEVEN = r2 was hit, the stop moved up to the entry price, and it later stopped out at cost. R3 = took profit and gave back via the trailing stop. MAX_HOLD = held the full 60 trading days. HK-only field.
└hold_days	int32	Trading days held. HK-only field.
└status	string	open / closed.
└kline	AnalysisKlinePointV2[]	Daily bars with full OHLC + volume — enough to draw candlesticks. Roughly the last 90 calendar days, ending today. Same source and format as the US endpoints. NOT returned by /swingmax/hk/open, which is built for pulling the whole list at once and would inflate several times over with bars.
└close	double	—
└timestamp	int64	Unix seconds.
└open	double	—
└high	double	—
└low	double	—
└volume	double	—

Example request

POST https://api.alphio.ai/v1/openapi/analysis/swingmax/hk/signals
Authorization: Bearer pk_live_••••••••••••••••••••••••••••
{
  "tickers":[
    {"asset_type":0,"ticker":"00700","locale":"HK"},
    {"asset_type":0,"ticker":"09988","locale":"HK"}
  ],
  "lang":"zh-tw"
}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "signals":[
      {"symbol":"00700","code":"00700.HK","company_name":"腾讯控股","trade_currency":"HKD","board_lot":100,
       "entry_time":"2026-03-12","entry_time_ts":1773244800,"entry_price":402.4,"stop_loss":"402.4000000",
       "r2_price":"441.8000000","r2_sell_price":"441.8000000","r2_signal_ts":1775750400,
       "r3_price":"482.8800000","r3_sell_price":"","r3_signal_ts":0,
       "exit_time":"","exit_time_ts":0,"exit_price":0,"return":0,"realized":"0.0391000",
       "exit_reason":"","hold_days":31,"status":"open"}
    ]
  }
}

POST/v1/openapi/analysis/swingmax/hk/stocks

SwingmaxHKStocks

Browse HK swingmax signals over a date range. Mirrors /swingmax/stocks on the US side.

Amounts are in trade_currency, NOT always HKD. Check the field before aggregating across symbols.

Request parameters

Field	Type	Required	Description
stype	int32	no	GetSwingMaxStocksType. 1=open pool, 2=closed, 3=today's new signals, 4=today's still-open. 5 (today-updated) returns InvalidArgument on HK: the HK signal table records only entry_time, with no last-updated column to distinguish it.
side	int32	no	0=long. HK swingmax is long-only, so side=1 (short) returns InvalidArgument rather than an empty list — an empty list reads as "no signals today" and would leave you waiting for data that never comes.
trade_status	int32	no	0 or 2 (all). HK has no paper-trade execution tracking, so trade_status=1 (traded) returns InvalidArgument.
page	int64	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Default 20, capped at 20 — the same ceiling as every other paged list on this API; only the two swingmax /open endpoints are uncapped. Was default 100 / cap 500 before 2026-09-15.
sort_type	int32	no	0 = endpoint default, 1 = latest first, 2 = oldest first, 3/4 = by return (whole-trade net P&L), high/low first. Note this is NOT realized, which is the banked scale-out portion used for win-rate — the two differ materially. The same number means the same thing on the US and HK endpoints.
keyword	string	no	Case-insensitive substring match on ticker, code and company name. Omit to disable filtering. Max 64 characters.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisSwingmaxHKStocksData	Response payload.
└stocks	AnalysisSwingmaxHKItem[]	One entry per signal. This array was called items until 2026-09-09; it is stocks now, matching US /swingmax/stocks. Reading the old key gets you nothing and no error — it looks like HK returned no data.
└id	int64	Signal id — same meaning as US swingmax id: assigned at entry, stable through close. Pass it back as ids on /swingmax/hk/signals to follow one trade. Added 2026-09-15. Also present on /swingmax/hk/open.
└symbol	string	HK code — always 5 digits, e.g. 00700.
└code	string	Symbol with the .HK suffix.
└company_name	string	Company name, in the language of the listing.
└trade_currency	string	HKD / CNY / USD. HK has three trading counters; the same company can list under more than one symbol, so amounts are NOT comparable across symbols without checking this. HK-only field.
└board_lot	int32	Shares per lot. HK trades in lots, not shares — 00700 is 100 shares/lot (~HKD 40k). 0 means the lot size could not be resolved. HK-only field.
└entry_time	string	Signal date, a Hong Kong trading day (YYYY-MM-DD).
└entry_time_ts	int64	Entry time, unix seconds — midnight HKT on entry_time. Always > 0. Use this to place the entry marker; it saves parsing the date string.
└entry_price	double	Buy target (breakout level), in trade_currency. The entry marker goes here. A number, not a string — matching US swingmax.
└stop_loss	string	Support level. Entry minus 15%; moves up to the entry price once r2 triggers.
└r2_price	string	Sell Target 1 (first resistance).
└r2_sell_price	string	Actual r2 fill; empty when not triggered. This is the price to mark, not r2_price.
└r2_signal_ts	int64	r2 trigger time, unix seconds. 0 means not triggered — draw the first target marker only when this is > 0. Same rule as US swingmax.
└r3_price	string	Sell Target 2 (second resistance).
└r3_sell_price	string	Actual r3 fill; empty when not triggered.
└r3_signal_ts	int64	r3 trigger time, unix seconds. 0 means not triggered — draw the second target marker only when this is > 0.
└exit_time	string	Exit date; empty while the position is still open.
└exit_time_ts	int64	Exit time, unix seconds. 0 while the position is open — this one field carries both when it exited and whether it has exited, exactly as in US swingmax. There is no exit_signal_ts field in either market.
└exit_price	double	Final exit price. The exit marker goes here. 0 while the position is open — check exit_time_ts, not this, to tell open from closed.
└return	double	Net return of the WHOLE position, decimal fraction. Same field, same meaning and same type as US swingmax return. It was called profit_rate until 2026-09-09.
└realized	string	Return already banked by the scale-out, decimal fraction. This — not return — is the win indicator: realized > 0 means the trade counted as a win, matching how US swingmax reports its win rate. The two numbers differ materially, so do not mix them. HK-only field.
└exit_reason	string	ALERT = stopped at the original -15% stop (the only losing bucket). BREAKEVEN = r2 was hit, the stop moved up to the entry price, and it later stopped out at cost. R3 = took profit and gave back via the trailing stop. MAX_HOLD = held the full 60 trading days. HK-only field.
└hold_days	int32	Trading days held. HK-only field.
└status	string	open / closed.
└kline	AnalysisKlinePointV2[]	Daily bars with full OHLC + volume — enough to draw candlesticks. Roughly the last 90 calendar days, ending today. Same source and format as the US endpoints. NOT returned by /swingmax/hk/open, which is built for pulling the whole list at once and would inflate several times over with bars.
└close	double	—
└timestamp	int64	Unix seconds.
└open	double	—
└high	double	—
└low	double	—
└volume	double	—
└total	int32	—

Example request

POST https://api.alphio.ai/v1/openapi/analysis/swingmax/hk/stocks
Authorization: Bearer pk_live_••••••••••••••••••••••••••••
{"stype":2,"page":1,"size":50,"sort_type":1}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "stocks":[
      {"symbol":"00700","code":"00700.HK","company_name":"腾讯控股","trade_currency":"HKD","board_lot":100,
       "entry_time":"2026-03-12","entry_time_ts":1773244800,"entry_price":402.4,"stop_loss":"402.4000000",
       "r2_price":"441.8000000","r2_sell_price":"441.8000000","r2_signal_ts":1775750400,
       "r3_price":"482.8800000","r3_sell_price":"","r3_signal_ts":0,
       "exit_time":"2026-04-28","exit_time_ts":1777305600,"exit_price":402.4,
       "return":0.0391,"realized":"0.0391000",
       "exit_reason":"BREAKEVEN","hold_days":31,"status":"closed"}
    ],
    "total": 1
  }
}

POST/v1/openapi/analysis/swingmax/hk/open

SwingmaxHKOpen

Returns every currently-open HK position in a single call by default. Mirrors /swingmax/open on the US side, including the shared paging parameters.

This endpoint changed shape on 2026-09-09 to match US /swingmax/open field for field. It no longer returns the same item as /swingmax/hk/signals and /swingmax/hk/stocks. The same numbers now carry different names depending on which endpoint you call: buy_price is entry_price, target1 is r2_price, target2 is r3_price, signal_time is entry_time_ts, last_exit_ts is exit_time_ts. That split exists in US swingmax and HK now mirrors it, so the two markets' /open responses are interchangeable. Fields with no counterpart here — company_name, return, realized, exit_reason, hold_days, kline — are on /swingmax/hk/signals and /swingmax/hk/stocks.

Amounts are in trade_currency, NOT always HKD. Check the field before aggregating across symbols.

Request parameters

Field	Type	Required	Description
page	int32	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Omit to return every open position in one call — that is the historical behaviour of this endpoint and it is unchanged. Set it only if you want pages.
sort_type	int32	no	0 = endpoint default, 1 = latest first, 2 = oldest first, 3/4 = by return (whole-trade net P&L, not the banked realized portion), high/low first. The same number means the same thing on the US and HK endpoints.
keyword	string	no	Case-insensitive substring match on ticker, code and company name. Omit to disable filtering. Max 64 characters.
lang	string	no	Localizes server-rendered status text. Takes priority over the lang header.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisSwingmaxHKOpenData	Response payload.
└positions	AnalysisSwingmaxHKOpenPosition[]	Every currently-open HK position. Field for field the same shape as US /swingmax/open, including HK-only trade_currency and board_lot at the end.
└id	int64	Signal id — the same number as id on /swingmax/hk/signals and /swingmax/hk/stocks. Take it from this list and pass it back as ids on /swingmax/hk/signals to follow one position until it closes. Added 2026-09-15.
└symbol	string	HK code — always 5 digits, e.g. 00700.
└code	string	Symbol with the .HK suffix.
└buy_price	string	Entry / Buy Target, in trade_currency. Same number that /swingmax/hk/signals calls entry_price — a string here, a double there, because this endpoint mirrors US /swingmax/open.
└stop_loss	string	Support level. Entry minus 15%; moves up to the entry price once target1 triggers.
└target1	string	Sell Target 1 (r2). Same number /swingmax/hk/signals calls r2_price.
└target2	string	Sell Target 2 (r3). Same number /swingmax/hk/signals calls r3_price.
└potential_gain	double	Static: (target2/buy_price - 1) * 100, percent, rounded to 2 decimals. Identical computation to US swingmax, so the two markets' values are directly comparable. Note it measures the distance to target2, the FURTHER target — a position that already ran past target1 still reports the full distance to target2. This is what sort_type 3/4 sorts by on this endpoint.
└signal_time	int64	Entry time, unix seconds. Same number /swingmax/hk/signals calls entry_time_ts.
└last_exit_ts	int64	Exit time, unix seconds; always 0 here since this endpoint returns open positions only.
└trade_currency	string	HKD / CNY / USD. buy_price and both targets are in this currency, so they are NOT comparable across symbols without checking it. HK-only field.
└board_lot	int32	Shares per lot. HK trades in lots, not shares — 00700 is 100 shares/lot. 0 means the lot size could not be resolved. HK-only field.
└total	int32	Rows after filtering, before paging.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/swingmax/hk/open
{}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "positions":[
      {"symbol":"00700","code":"00700.HK","buy_price":"402.4","stop_loss":"342.04",
       "target1":"441.8","target2":"482.88","potential_gain":20,
       "signal_time":1787155200,"last_exit_ts":0,
       "trade_currency":"HKD","board_lot":100}
    ],
    "total": 1
  }
}

DayTrading (US)

Intraday signals for US equities. A different animal from Swingmax: signals fire on 15m/30m bars, a dozen or so a day, and most resolve within hours — Swingmax is a daily-bar swing strategy whose positions live for weeks. Poll accordingly.

POST/v1/openapi/analysis/daytrading/open

DaytradingOpen

Positions from one US trading session, open and closed. Which session: on a trading day before the US open (09:30 ET), and on any non-trading day, it is the previous trading session; from the open onward it is the current one, including after the close. Closed positions carry sell_price, sell_ts and a realized return_rate; open ones carry current_price and a live return_rate. NOTE: between the open and the first signal of the day (intraday signals only start at 10:00 ET), and on days when no signal fires at all, the current session is empty — that is expected, not an outage.

The response array is data.list, as on all three daytrading endpoints. It was called positions on /open and signals on /signals until 2026-09-10. Reading the old key yields nothing and no error — it looks like the endpoint returned no data.

There is deliberately no time-range parameter. The signal set is small (about 2k rows for US in total), so page + total is enough to reach any record — and a hidden default window is a trap: callers who pass nothing get an empty list and conclude the endpoint is broken, which is exactly what happened when the HK history endpoint launched. Sort with sort_type, then walk the pages using total.

Intraday signals fire only in two windows — 10:00-11:00 and 13:00-14:00 ET — and everything still open is force-closed at 15:50 ET. Outside US market hours this endpoint will usually return an empty list, which is correct, not an outage.

Request parameters

Field	Type	Required	Description
page	int32	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Default 20, capped at 20 — the same ceiling as every other paged list on this API. Unlike the two swingmax /open endpoints this one is NOT uncapped: every row carries minute bars, so a full-book response would be several times larger. Was 10 before 2026-09-15.
sort_type	int32	no	0 = endpoint default (by return_rate, high to low — realized for closed positions, live unrealized for open ones), 1 = latest first, 2 = oldest first, 3 = highest return first, 4 = lowest return first. The same number means the same thing on the US and HK endpoints.
keyword	string	no	Case-insensitive substring match on ticker, code and company name. Omit to disable filtering. Max 64 characters.
lang	string	no	Localizes the status text. When set it takes priority over the "lang" request header: "zh" / "zh-cn" for Simplified Chinese, "zh-tw" for Traditional Chinese. Omit or leave empty to fall back to the header (English by default).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisDaytradingOpenData	Response payload.
└list	AnalysisDaytradingPosition[]	Every position of the session this endpoint is showing — open and closed. Field for field the same shape as /daytrading/signals and /daytrading/history.
└id	int64	Signal id — the same number on /open, /signals and /history for the same trade. Store it from any of the three and pass it back as ids on /signals to follow this position.
└symbol	string	Bare code. US: NVDA. HK: five digits, e.g. 00700.
└code	string	Suffixed form. US: NVDA.O. HK: 00700.HK.
└name	string	Company name.
└logo	string	Logo URL; empty when unavailable.
└interval	string	Bar interval behind the entry, 15m or 30m.
└signal_date	string	Signal day on that market's trading calendar, YYYY-MM-DD. Derived from buy_ts in the market's own timezone — an HK 09:30 signal is still the previous day in New York.
└signal_time	string	Human-readable trigger time, YYYY-MM-DD HH:MM:SS in the market's own timezone. Same format as on /daytrading/signals and /daytrading/history.
└buy_ts	int64	Entry time, unix seconds. On 30m signals it is the right-edge timestamp of the bar that triggered the entry, so it matches one klines entry exactly.
└buy_price	string	Entry price.
└sell_ts	int64	Exit time, unix seconds; 0 while the position is still open. This endpoint returns every position of the session it shows, closed ones included, so closed rows carry their exit time. On 30m signals a kline-triggered exit (take-profit or stop) carries that bar's right-edge timestamp; the 15:50 force-close carries wall-clock time.
└sell_price	string	Exit price; empty while the position is still open.
└stop_loss_price	string	Strategy stop, entry × 0.97 — both markets stop at -3%. Without it you cannot reproduce the strategy.
└current_price	string	Latest price.
└return_rate	string	Current return, percent.
└status	string	Human-readable progress text. Same field as on /daytrading/signals and /daytrading/history.
└klines	AnalysisKlinePointV2[]	Minute bars, same point shape as everywhere else in this API. Sorted by timestamp ascending (oldest first) — the same order in both markets, on all three daytrading endpoints, and as swingmax. Bar stamping since 2026-09-17: on 30m signals the bars are right-edge stamped and anchored to the session open, the same convention as /pivot-point/info — the first US bar of the day is 10:00, the first HK afternoon bar 13:30, and buy_ts is one of these timestamps. 15m signals keep the upstream left-edge stamps with the 09:30 bar removed, so that series starts at 09:45.
└total	int32	Rows after filtering, before paging.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/daytrading/open
Authorization: Bearer pk_live_••••••••••••••••••••••••••••
{"page":1,"size":10,"sort_type":1}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "list":[
      {"id":90211,"symbol":"NVDA","code":"NVDA.O","name":"NVIDIA Corp","logo":"https://.../nvda.png",
       "interval":"15m","signal_date":"2025-08-19","signal_time":"2025-08-19 10:00:00",
       "buy_ts":1755612000,"buy_price":"118.42","sell_ts":0,"sell_price":"",
       "stop_loss_price":"114.87","current_price":"119.80",
       "return_rate":"1.17","status":"Entered @118.42 and staying patient",
       "klines":[]}
    ],
    "total": 1
  }
}

POST/v1/openapi/analysis/daytrading/signals

DaytradingSignals

Signal records for the given tickers — what fired and how it resolved, as opposed to /daytrading/open which describes the current book. Same request shape as /swingmax/signals: pass tickers, get every signal for those codes in one response.

Lookup by id since 2026-09-15: pass ids (max 20) instead of tickers. Items have carried id all along; that is the value to pass back. Unknown ids are skipped rather than errored, so an empty list for a single id means that id does not exist.

Field names now match /daytrading/open and /daytrading/history — this endpoint used to expose dip_price / alert_price, which were the signal table's own column names. Only the names changed; the values map one-to-one. All three daytrading endpoints now return the same 17 fields; what differs is the rows, not the shape — this one carries in-flight signals, /history holds settled round-trips only.

Removed on 2026-09-14: page, size, sort_type and keyword. A single code has at most a few dozen signals, so paging had no use here, and sending any of them now fails at the gateway as an unknown field. There is deliberately no time-range parameter either: the response is every signal for the codes you asked for, so a hidden default window cannot hand you an empty list that reads like an outage. To walk the whole table use /daytrading/history (paged); for today's book use /daytrading/open.

Historical win-rate figures are NOT exposed here on purpose: the internal historical analyzer currently evaluates past signals against today's stock universe, which is look-ahead biased. Do not derive strategy performance from a long backfill of this endpoint.

Request parameters

Field	Type	Required	Description
tickers	AnalysisAsset[]	no	An array of OBJECTS, not of strings: [{"ticker":"NVDA"}]. Either tickers or ids is required; both together is InvalidArgument. Matches both the bare code and the suffixed one, case-insensitively. Every signal for those codes comes back in one response, newest buy_ts first.
└asset_type	int32	no	0=stock, 1=etf, 2=crypto.
└symbol	string	no	Legacy field.
└ticker	string	no	Preferred field.
└locale	string	no	"US" / "HK".
ids	int64[]	no	Signal ids to look up — the id field of an item returned by /open, /signals or /history (all three carry the same signal id). At most 20; more is InvalidArgument, NOT truncated. Mutually exclusive with tickers. Selects the same rows the tickers path would (in-flight and resolved signals alike), just by id. Negative ids — HK backtest rows from /history — are resolved too, returning that round-trip in the same item shape (status starts with "Backtest:"). 0 is InvalidArgument; unknown ids are skipped, not errored. Added 2026-09-15.
lang	string	no	Localizes the status text. When set it takes priority over the "lang" request header: "zh" / "zh-cn" for Simplified Chinese, "zh-tw" for Traditional Chinese. Omit or leave empty to fall back to the header (English by default).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisDaytradingSignalsData	Response payload.
└list	AnalysisDaytradingSignalItem[]	One entry per signal.
└id	int64	Signal id — the value to pass back as ids on this endpoint; the same number as id on /open and /history for the same trade.
└symbol	string	Bare code. US: NVDA. HK: five digits, e.g. 00700.
└code	string	Suffixed form. Empty on US — that pipeline has no ticker lookup on this path.
└name	string	Company name. Populated on HK (real name from the HK code table). Empty on US.
└logo	string	Logo URL. Empty on both: HK has no logo source, and the US pipeline is left untouched here.
└interval	string	Bar interval that produced the entry, 15m or 30m.
└signal_date	string	Signal day on that market's trading calendar, YYYY-MM-DD.
└signal_time	string	Human-readable trigger time. HK only.
└buy_ts	int64	Entry time, unix seconds. On 30m signals it is the right-edge timestamp of the bar that triggered the entry, so it matches one klines entry exactly.
└buy_price	string	Entry signal price.
└sell_ts	int64	Exit time, unix seconds. 0 while still open. On 30m signals a kline-triggered exit (take-profit or stop) carries that bar's right-edge timestamp; the 15:50 force-close carries wall-clock time.
└sell_price	string	Exit signal price; empty while still open.
└stop_loss_price	string	Strategy stop (entry × 0.97 — both markets stop at -3%).
└current_price	string	Latest price; empty for settled signals.
└return_rate	string	Percent, not a fraction. Same name and meaning as on /open and /history.
└status	string	Human-readable progress/exit text — the same field as status on /open and /history.
└klines	AnalysisKlinePointV2[]	Minute bars centred on the entry. Sorted by timestamp ascending (oldest first) — the same order in both markets, on all three daytrading endpoints, and as swingmax. Bar stamping since 2026-09-17: on 30m signals the bars are right-edge stamped and anchored to the session open, the same convention as /pivot-point/info — the first US bar of the day is 10:00, the first HK afternoon bar 13:30, and buy_ts is one of these timestamps. 15m signals keep the upstream left-edge stamps with the 09:30 bar removed, so that series starts at 09:45.
└total	int32	Always equal to the length of list — this endpoint is not paged.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/daytrading/signals
Authorization: Bearer pk_live_••••••••••••••••••••••••••••
{"tickers":[{"asset_type":0,"ticker":"NVDA","locale":"US"}]}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "list":[
      {"id":90211,"symbol":"NVDA","code":"","name":"","logo":"","interval":"15m","signal_date":"2026-08-19","signal_time":"",
       "buy_ts":1755612000,"buy_price":"118.42","sell_ts":1755620100,"sell_price":"121.97","stop_loss_price":"114.87","current_price":"",
       "return_rate":"3.00","status":"Booked partial profits @121.97",
       "klines":[{"close":118.9,"timestamp":1755612000,"open":118.4,"high":119.1,"low":118.2,"volume":1234567}]}
    ],
    "total": 1
  }
}

POST/v1/openapi/analysis/daytrading/history

DaytradingHistory

Closed round-trips — one row per completed entry-and-exit, with both prices, both timestamps and the realised return. Deliberately carries no K-line data. Call /v1/openapi/data/kline if you need bars — inlining them here would grow the payload by an order of magnitude for callers who only want the numbers.

Easy to confuse with /daytrading/signals: signals is the event log (what fired, what state it is in now), history is the P&L ledger (did this trade make money). Signals answers "what happened", history answers "how did it end".

When you pass a symbol, the most recent still-open position for that symbol is prepended as a synthetic row with exit_time, exit_price and return_rate all 0 — and it is inserted after paging, so that page can return size + 1 rows and total will not count it. The HK endpoint does not do this.

Filter on exit_time > 0 before computing any statistic. It is the only rule that holds for both markets, and it keeps those return_rate 0 rows from being counted as break-even trades in a win rate.

Request parameters

Field	Type	Required	Description
symbol	string	no	Optional ticker, e.g. NVDA. Omit for all symbols. Passing one enables the synthetic open-position row described in the notes.
page	int32	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Default 20, capped at 20 — the same ceiling as every other paged list on this API; only the two swingmax /open endpoints are uncapped. Was 10 before 2026-09-15.
sort_type	int32	no	0 = endpoint default, 1 = latest first, 2 = oldest first. 3 and 4 (by return) return InvalidArgument here: this endpoint pages upstream, so sorting would only reorder the current page — sort client-side after fetching all pages instead. The same number means the same thing on the US and HK endpoints.
keyword	string	no	Exact ticker match, applied upstream before paging (the trades table has no name column). If symbol is also set, symbol wins — an exact filter should not be overridden by a fuzzy one. Omit to disable filtering. Max 64 characters.
lang	string	no	Localizes the status text. When set it takes priority over the "lang" request header: "zh" / "zh-cn" for Simplified Chinese, "zh-tw" for Traditional Chinese. Omit or leave empty to fall back to the header (English by default).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisDaytradingHistoryData	Response payload.
└list	AnalysisDaytradingHistoryItem[]	One entry per round-trip, newest entry first.
└id	int64	Signal id — the same number as id on /open and /signals for this trade, so an id stored from any of the three endpoints can be passed back as ids on /signals. Before 2026-09-15 this field was the round-trip table's own key, a different keyspace. Negative means a backtest row (HK only — exit_reason starts with "Backtest:"): every backtest row carries its own distinct negative id, so it works as a bookmark key, and passing it back as ids on /daytrading/hk/signals returns that round-trip in the same item shape as a live signal. 0 means a legacy row with no linked signal; 0 is the one value ids rejects.
└symbol	string	Bare code. US: NVDA. HK: five digits, e.g. 00700.
└code	string	Suffixed form. US: NVDA.O. HK: 00700.HK.
└name	string	Company name.
└logo	string	Logo URL; empty when unavailable.
└interval	string	Bar interval behind the entry, 15m or 30m.
└signal_date	string	Signal day on that market's trading calendar, YYYY-MM-DD.
└signal_time	string	Human-readable trigger time.
└buy_ts	int64	Entry time, unix seconds. On 30m signals it is the right-edge timestamp of the bar that triggered the entry, so it matches one klines entry exactly.
└buy_price	string	Entry price.
└sell_ts	int64	Exit time, unix seconds. 0 = not settled yet. On 30m signals a kline-triggered exit (take-profit or stop) carries that bar's right-edge timestamp; the 15:50 force-close carries wall-clock time.
└sell_price	string	Exit price; empty when not settled.
└stop_loss_price	string	Strategy stop price (HK daytrading stops at -3%). This was missing before — without it you cannot reproduce or check the strategy.
└current_price	string	Latest price.
└return_rate	string	Percent, not a fraction: "2.35" means +2.35%. Same name and meaning as on /daytrading/open.
└status	string	Human-readable exit text. Same field as on /daytrading/open.
└klines	AnalysisKlinePointV2[]	Minute bars around the trade — the same series the product cards draw candlesticks from. Previously suppressed by this layer even though upstream always returned them. Sorted by timestamp ascending (oldest first) — the same order in both markets, on all three daytrading endpoints, and as swingmax. Bar stamping since 2026-09-17: on 30m signals the bars are right-edge stamped and anchored to the session open, the same convention as /pivot-point/info — the first US bar of the day is 10:00, the first HK afternoon bar 13:30, and buy_ts is one of these timestamps. 15m signals keep the upstream left-edge stamps with the 09:30 bar removed, so that series starts at 09:45.
└total	int32	Total rows matching the range, for paging. Excludes the synthetic open-position row.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/daytrading/history
Authorization: Bearer pk_live_••••••••••••••••••••••••••••
Content-Type: application/json

{ "symbol": "NVDA", "page": 1, "size": 20 }

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "list": [
      {
        "id": 3521,
        "symbol": "NVDA",
        "code": "NVDA",
        "trade_date": "2026-08-19",
        "entry_time": 1755612000,
        "exit_time": 1755620100,
        "entry_price": "118.4200",
        "exit_price": "121.9700",
        "return_rate": "3.00",
        "interval": "15m",
        "exit_reason": "Booked 3.00% profits @121.97 at 11:35 AM"
      }
    ],
    "total": 1
  }
}

DayTrading (HK)

Intraday signals for HK equities. Same shape as the US feed but a SEPARATE pipeline end to end — its own process, its own signal table, its own switch. One market going down does not affect the other, which is also why the paths differ instead of taking a locale parameter. The strategy parameters are NOT the US ones: take profit at 0.5%, book on a 0.2% pull-back from the high, stop at -3%, and entries only fire while the market (equal-weighted intraday move across the pool) is below -0.4% (it was -0.8% until 2026-09-29; relaxed so more sessions produce entries). Those numbers were calibrated on HK data — the US set (3% / 1%) scored a 46.5% win rate with negative expectancy when replayed on HK bars. Sessions are 09:30-12:00 and 13:00-16:00 HKT; signal windows are 10:00-11:00 and 13:00-14:00, and anything still open is force-closed at 15:50.

POST/v1/openapi/analysis/daytrading/hk/open

DaytradingHKOpen

Positions from one Hong Kong trading session, open and closed. Which session: on a trading day before the HK open (09:30 HKT), and on any non-trading day, it is the previous trading session; from the open onward it is the current one, including after the close. `code` carries the .HK suffix (00700.HK); `symbol` is the bare five-digit number. Closed positions carry sell_price, sell_ts and a realized return_rate; open ones carry current_price and a live return_rate. NOTE: between the open and the first signal of the day (intraday signals only start at 10:00 HKT), and on days when the market-breadth gate never opens, the current session is empty — that is expected, not an outage.

Intraday signals fire only in two windows — 10:00-11:00 and 13:00-14:00 HKT — and everything still open is force-closed at 15:50 HKT. Note the lunch break: the HK session runs 09:30-12:00 and 13:00-16:00, so the second window opens right at the afternoon bell. Outside HK market hours this endpoint will usually return an empty list, which is correct, not an outage.

Request parameters

Field	Type	Required	Description
page	int32	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Default 20, capped at 20 — the same ceiling as every other paged list on this API. Unlike the two swingmax /open endpoints this one is NOT uncapped: every row carries minute bars, so a full-book response would be several times larger. Was 10 before 2026-09-15.
sort_type	int32	no	0 = endpoint default (by return_rate, high to low — realized for closed positions, live unrealized for open ones), 1 = latest first, 2 = oldest first, 3 = highest return first, 4 = lowest return first. The same number means the same thing on the US and HK endpoints.
keyword	string	no	Case-insensitive substring match on ticker, code and company name. Omit to disable filtering. Max 64 characters.
lang	string	no	Localizes the status text. When set it takes priority over the "lang" request header: "zh" / "zh-cn" for Simplified Chinese, "zh-tw" for Traditional Chinese. Omit or leave empty to fall back to the header (English by default).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisDaytradingOpenData	Response payload.
└list	AnalysisDaytradingPosition[]	Every position of the session this endpoint is showing — open and closed. Field for field the same shape as /daytrading/signals and /daytrading/history.
└id	int64	Signal id — the same number on /open, /signals and /history for the same trade. Store it from any of the three and pass it back as ids on /signals to follow this position.
└symbol	string	Bare code. US: NVDA. HK: five digits, e.g. 00700.
└code	string	Suffixed form. US: NVDA.O. HK: 00700.HK.
└name	string	Company name.
└logo	string	Logo URL; empty when unavailable.
└interval	string	Bar interval behind the entry, 15m or 30m.
└signal_date	string	Signal day on that market's trading calendar, YYYY-MM-DD. Derived from buy_ts in the market's own timezone — an HK 09:30 signal is still the previous day in New York.
└signal_time	string	Human-readable trigger time, YYYY-MM-DD HH:MM:SS in the market's own timezone. Same format as on /daytrading/signals and /daytrading/history.
└buy_ts	int64	Entry time, unix seconds. On 30m signals it is the right-edge timestamp of the bar that triggered the entry, so it matches one klines entry exactly.
└buy_price	string	Entry price.
└sell_ts	int64	Exit time, unix seconds; 0 while the position is still open. This endpoint returns every position of the session it shows, closed ones included, so closed rows carry their exit time. On 30m signals a kline-triggered exit (take-profit or stop) carries that bar's right-edge timestamp; the 15:50 force-close carries wall-clock time.
└sell_price	string	Exit price; empty while the position is still open.
└stop_loss_price	string	Strategy stop, entry × 0.97 — both markets stop at -3%. Without it you cannot reproduce the strategy.
└current_price	string	Latest price.
└return_rate	string	Current return, percent.
└status	string	Human-readable progress text. Same field as on /daytrading/signals and /daytrading/history.
└klines	AnalysisKlinePointV2[]	Minute bars, same point shape as everywhere else in this API. Sorted by timestamp ascending (oldest first) — the same order in both markets, on all three daytrading endpoints, and as swingmax. Bar stamping since 2026-09-17: on 30m signals the bars are right-edge stamped and anchored to the session open, the same convention as /pivot-point/info — the first US bar of the day is 10:00, the first HK afternoon bar 13:30, and buy_ts is one of these timestamps. 15m signals keep the upstream left-edge stamps with the 09:30 bar removed, so that series starts at 09:45.
└total	int32	Rows after filtering, before paging.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/daytrading/hk/open
Authorization: Bearer pk_live_••••••••••••••••••••••••••••
{"page":1,"size":10,"sort_type":1}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "list":[
      {"id":13,"symbol":"00522","code":"00522.HK","name":"ASMPT Ltd","logo":"https://.../00522.png",
       "interval":"30m","signal_date":"2026-09-10","signal_time":"2026-09-10 13:41:49",
       "buy_ts":1789018909,"buy_price":"164.40","sell_ts":0,"sell_price":"",
       "stop_loss_price":"159.47","current_price":"165.60",
       "return_rate":"0.73","status":"Entered @164.40 and staying patient",
       "klines":[]}
    ],
    "total": 1
  }
}

POST/v1/openapi/analysis/daytrading/hk/signals

DaytradingHKSignals

Signal records for the given tickers — what fired and how it resolved, as opposed to /daytrading/hk/open which describes the current book. Same request shape as /swingmax/hk/signals: pass tickers, get every signal for those codes in one response.

Removed on 2026-09-14: page, size, sort_type and keyword. A single code has at most a few dozen signals, so paging had no use here, and sending any of them now fails at the gateway as an unknown field. There is deliberately no time-range parameter either: the response is every signal for the codes you asked for, so a hidden default window cannot hand you an empty list that reads like an outage. To walk the whole table use /daytrading/hk/history (paged); for today's book use /daytrading/hk/open.

Request parameters

Field	Type	Required	Description
tickers	AnalysisAsset[]	no	An array of OBJECTS, not of strings: [{"ticker":"00700"}]. Either tickers or ids is required; both together is InvalidArgument. Matches both the bare code and the suffixed one, case-insensitively. Every signal for those codes comes back in one response, newest buy_ts first.
└asset_type	int32	no	0=stock, 1=etf, 2=crypto.
└symbol	string	no	Legacy field.
└ticker	string	no	Preferred field.
└locale	string	no	"US" / "HK".
ids	int64[]	no	Signal ids to look up — the id field of an item returned by /open, /signals or /history (all three carry the same signal id). At most 20; more is InvalidArgument, NOT truncated. Mutually exclusive with tickers. Selects the same rows the tickers path would (in-flight and resolved signals alike), just by id. Negative ids — HK backtest rows from /history — are resolved too, returning that round-trip in the same item shape (status starts with "Backtest:"). 0 is InvalidArgument; unknown ids are skipped, not errored. Added 2026-09-15.
lang	string	no	Localizes the status text. When set it takes priority over the "lang" request header: "zh" / "zh-cn" for Simplified Chinese, "zh-tw" for Traditional Chinese. Omit or leave empty to fall back to the header (English by default).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisDaytradingSignalsData	Response payload.
└list	AnalysisDaytradingSignalItem[]	One entry per signal.
└id	int64	Signal id — the value to pass back as ids on this endpoint; the same number as id on /open and /history for the same trade.
└symbol	string	Bare code. US: NVDA. HK: five digits, e.g. 00700.
└code	string	Suffixed form. Empty on US — that pipeline has no ticker lookup on this path.
└name	string	Company name. Populated on HK (real name from the HK code table). Empty on US.
└logo	string	Logo URL. Empty on both: HK has no logo source, and the US pipeline is left untouched here.
└interval	string	Bar interval that produced the entry, 15m or 30m.
└signal_date	string	Signal day on that market's trading calendar, YYYY-MM-DD.
└signal_time	string	Human-readable trigger time. HK only.
└buy_ts	int64	Entry time, unix seconds. On 30m signals it is the right-edge timestamp of the bar that triggered the entry, so it matches one klines entry exactly.
└buy_price	string	Entry signal price.
└sell_ts	int64	Exit time, unix seconds. 0 while still open. On 30m signals a kline-triggered exit (take-profit or stop) carries that bar's right-edge timestamp; the 15:50 force-close carries wall-clock time.
└sell_price	string	Exit signal price; empty while still open.
└stop_loss_price	string	Strategy stop (entry × 0.97 — both markets stop at -3%).
└current_price	string	Latest price; empty for settled signals.
└return_rate	string	Percent, not a fraction. Same name and meaning as on /open and /history.
└status	string	Human-readable progress/exit text — the same field as status on /open and /history.
└klines	AnalysisKlinePointV2[]	Minute bars centred on the entry. Sorted by timestamp ascending (oldest first) — the same order in both markets, on all three daytrading endpoints, and as swingmax. Bar stamping since 2026-09-17: on 30m signals the bars are right-edge stamped and anchored to the session open, the same convention as /pivot-point/info — the first US bar of the day is 10:00, the first HK afternoon bar 13:30, and buy_ts is one of these timestamps. 15m signals keep the upstream left-edge stamps with the 09:30 bar removed, so that series starts at 09:45.
└total	int32	Always equal to the length of list — this endpoint is not paged.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/daytrading/hk/signals
Authorization: Bearer pk_live_••••••••••••••••••••••••••••
{"tickers":[{"asset_type":0,"ticker":"09988","locale":"HK"}],"lang":"zh-tw"}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "list":[
      {"id":41207,"symbol":"09988","code":"09988.HK","name":"阿里巴巴-W","logo":"","interval":"15m","signal_date":"2026-08-19","signal_time":"2026-08-19 10:15:00",
       "buy_ts":1755569700,"buy_price":"118.40","sell_ts":1755571500,"sell_price":"119.00","stop_loss_price":"114.85","current_price":"",
       "return_rate":"0.51","status":"Booked partial profits @119.00",
       "klines":[{"close":118.6,"timestamp":1755569700,"open":118.4,"high":118.9,"low":118.3,"volume":2310000}]}
    ],
    "total": 1
  }
}

POST/v1/openapi/analysis/daytrading/hk/history

DaytradingHKHistory

This endpoint is currently seeded with BACKTEST rows, not live fills. They are identifiable two ways: dip_id is negative (live signal ids are positive and auto-increment), and exit_reason starts with "Backtest:". Live rows begin appending from the 2026-09-08 session onward, into the same list.

What the backtest is: the production signal generator replayed over 5-minute bars for 414 HK names across 2026-06-10 to 2026-09-02 (63 sessions), with the live parameters — 0.5% take-profit, 0.2% trailing drawdown, -3% stop, entries only while the equal-weighted pool is down more than 0.8% on the day. 394 round-trips, 72.1% of them positive.

Backtest caveats that do not apply to live rows: fills are taken at bar prices rather than at the tick the live engine would have traded, and no transaction cost is deducted. HK round-trip cost is about 0.22%, which takes mean return per trade from +0.30% down to +0.081%. Do not present these figures as realised performance.

Easy to confuse with /daytrading/hk/signals: signals is the event log (what fired, what state it is in now), history is the P&L ledger (did this trade make money). Signals answers "what happened", history answers "how did it end".

exit_time = 0 means the round-trip is not settled yet — either still held, or the exit price was missing. Those rows carry return_rate 0 as well; treat 0/0 as "unknown", not as "broke even".

return_rate is gross in both kinds of row. The live writer derives it from alert_price / dip_price with no cost deducted, and the backtest rows follow the same convention on purpose — mixing gross and net in one column would make the two indistinguishable.

Request parameters

Field	Type	Required	Description
symbol	string	no	Optional. Bare five-digit HK code (00700). Omit for all symbols.
page	int32	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Default 20, capped at 20 — the same ceiling as every other paged list on this API; only the two swingmax /open endpoints are uncapped. Was 10 before 2026-09-15.
sort_type	int32	no	0 = endpoint default, 1 = latest first, 2 = oldest first. 3 and 4 (by return) return InvalidArgument here: this endpoint pages upstream, so sorting would only reorder the current page — sort client-side after fetching all pages instead. The same number means the same thing on the US and HK endpoints.
keyword	string	no	Exact ticker match, applied upstream before paging (the trades table has no name column). If symbol is also set, symbol wins — an exact filter should not be overridden by a fuzzy one. Omit to disable filtering. Max 64 characters.
lang	string	no	Localizes the status text. When set it takes priority over the "lang" request header: "zh" / "zh-cn" for Simplified Chinese, "zh-tw" for Traditional Chinese. Omit or leave empty to fall back to the header (English by default).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisDaytradingHistoryData	Response payload.
└list	AnalysisDaytradingHistoryItem[]	One entry per completed round-trip, newest entry first.
└id	int64	Signal id — the same number as id on /open and /signals for this trade, so an id stored from any of the three endpoints can be passed back as ids on /signals. Before 2026-09-15 this field was the round-trip table's own key, a different keyspace. Negative means a backtest row (HK only — exit_reason starts with "Backtest:"): every backtest row carries its own distinct negative id, so it works as a bookmark key, and passing it back as ids on /daytrading/hk/signals returns that round-trip in the same item shape as a live signal. 0 means a legacy row with no linked signal; 0 is the one value ids rejects.
└symbol	string	Bare code. US: NVDA. HK: five digits, e.g. 00700.
└code	string	Suffixed form. US: NVDA.O. HK: 00700.HK.
└name	string	Company name.
└logo	string	Logo URL; empty when unavailable.
└interval	string	Bar interval behind the entry, 15m or 30m.
└signal_date	string	Signal day on that market's trading calendar, YYYY-MM-DD.
└signal_time	string	Human-readable trigger time.
└buy_ts	int64	Entry time, unix seconds. On 30m signals it is the right-edge timestamp of the bar that triggered the entry, so it matches one klines entry exactly.
└buy_price	string	Entry price.
└sell_ts	int64	Exit time, unix seconds. 0 = not settled yet. On 30m signals a kline-triggered exit (take-profit or stop) carries that bar's right-edge timestamp; the 15:50 force-close carries wall-clock time.
└sell_price	string	Exit price; empty when not settled.
└stop_loss_price	string	Strategy stop price (HK daytrading stops at -3%). This was missing before — without it you cannot reproduce or check the strategy.
└current_price	string	Latest price.
└return_rate	string	Percent, not a fraction: "2.35" means +2.35%. Same name and meaning as on /daytrading/open.
└status	string	Human-readable exit text. Same field as on /daytrading/open.
└klines	AnalysisKlinePointV2[]	Minute bars around the trade — the same series the product cards draw candlesticks from. Previously suppressed by this layer even though upstream always returned them. Sorted by timestamp ascending (oldest first) — the same order in both markets, on all three daytrading endpoints, and as swingmax. Bar stamping since 2026-09-17: on 30m signals the bars are right-edge stamped and anchored to the session open, the same convention as /pivot-point/info — the first US bar of the day is 10:00, the first HK afternoon bar 13:30, and buy_ts is one of these timestamps. 15m signals keep the upstream left-edge stamps with the 09:30 bar removed, so that series starts at 09:45.
└total	int32	Total rows matching the range, for paging.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/daytrading/hk/history
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{ "symbol": "00700", "page": 1, "size": 20 }

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "list": [
      {
        "id": 12,
        "symbol": "00700",
        "code": "00700.HK",
        "trade_date": "2026-09-08",
        "entry_time": 1788838200,
        "exit_time": 1788845400,
        "entry_price": "445.4000",
        "exit_price": "448.2000",
        "return_rate": "0.63",
        "interval": "15m",
        "exit_reason": "Booked 0.63% profits @448.20 at 12:30 PM"
      }
    ],
    "total": 1
  }
}

AI Stock Picker

A model portfolio, not a signal feed. AI Stock Picker holds a basket of roughly ten US stocks by weight, rebalances weekly and stops names out intraday. Unlike /swingmax and /daytrading, where every row is an independent trade, these rows add up to one book and performance is measured on the book. Three endpoints mirror the product page: current holdings, closed positions, performance. Scope: analysis:stock-picker.

POST/v1/openapi/analysis/stock-picker/current

StockPickerCurrent

Current holdings of the AI Stock Picker book, with entry, stop, latest price, unrealized return, weight and daily bars for each name. Returns every holding in one call by default; the paging, sorting and keyword parameters are there for callers that want them.

Returns and weights are percent numbers everywhere in the two portfolio products (2.35 = +2.35%, 10 = 10%), and prices are always double — even where the upstream service carries them as strings or fractions.

There is no lang parameter: narrative and rationale exist in English only.

Empty positions is a valid state right after a full stop-out or on the data pipeline's rebuild window, not an outage. as_of still tells you which snapshot you are looking at.

Request parameters

Field	Type	Required	Description
page	int32	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Omit to return every holding in one call — the book holds about ten names. Set it only if you want pages.
sort_type	int32	no	0 = endpoint default (upstream order: weight, high to low), 1 = newest join_date first, 2 = oldest join_date first, 3 = highest return first, 4 = lowest return first.
keyword	string	no	Case-insensitive substring match on symbol and code. Omit to disable filtering. Max 64 characters.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisStockPickerCurrentData	Response payload.
└positions	AnalysisStockPickerPosition[]	One row per holding.
└symbol	string	Bare ticker, e.g. NVDA.
└code	string	Same as symbol for this product today — the upstream feed carries no exchange suffix here (e.g. MSFT, not MSFT.O).
└locale	string	Always US.
└asset_type	int32	0 = stock.
└narrative	string	Theme the name was picked under, e.g. "AI infrastructure". The portfolio is diversified across themes, so this is the grouping key. English only.
└rationale	string	Why this name was picked, as written by the model at entry. English only.
└join_date	string	Entry date, YYYY-MM-DD (US Eastern trading day). Compare with as_of to tell a fresh pick from a name that has been held for weeks.
└entry_price	double	Cost basis.
└stop_loss	double	Stop price. When it triggers the row moves to status=stopped and drops out of this list at the next snapshot; the exit then shows up on /stock-picker/history with exit_reason=stop_loss.
└current_price	double	Latest price. 0 means the quote could not be fetched — in that case return is also 0, which is NOT "flat".
└return	double	(current_price / entry_price - 1) * 100, in percent, 2 decimals. Unrealized.
└weight	double	Portfolio weight in percent (10 = 10% of the book).
└status	string	active, or stopped for a name whose stop triggered intraday and is waiting for the next snapshot to leave the book.
└kline	AnalysisKlinePointV2[]	Daily bars from join_date to today, oldest first. OHLC only — volume is always 0 on this endpoint.
└total	int32	Rows after keyword filtering, before paging.
└as_of	string	Snapshot date of the book, YYYY-MM-DD US Eastern. The snapshot is refreshed every trading day around 08:00 ET, but rebalancing only happens weekly — a new as_of does not mean the holdings changed; look at join_date for that.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/stock-picker/current
{}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "positions": [
      {"symbol":"ULTA","code":"ULTA","locale":"US","asset_type":0,
       "narrative":"Beauty retail resilience","rationale":"...",
       "join_date":"2026-10-06","entry_price":412.5,"stop_loss":371.25,
       "current_price":430.1,"return":4.27,"weight":10,"status":"active",
       "kline":[{"close":415.2,"timestamp":1759723200,"open":411.9,"high":417.3,"low":410.8,"volume":0}]}
    ],
    "total": 1,
    "as_of": "2026-10-15"
  }
}

POST/v1/openapi/analysis/stock-picker/history

StockPickerHistory

Closed positions of the AI Stock Picker book, newest exit first, paged upstream. Each row is one complete holding: entry and exit dates and prices, realized return, weight while held, and why it left the book.

There is deliberately no sort_type and no keyword on this endpoint. Rows are paged upstream, so this layer never holds the whole table; sorting the current page by return would look right and be wrong. Walk the pages and sort client-side, or use /stock-picker/performance for the aggregate numbers.

Rows start in early 2026 when the current strategy went live; earlier picks from the previous strategy generation are not included, so the history and /stock-picker/performance describe the same set of trades.

A position whose stop triggers today appears here the same day with exit_reason=stop_loss; rebalance exits land on the rebalance day. Cached for about a minute when the requested window includes today, longer for windows entirely in the past.

Request parameters

Field	Type	Required	Description
page	int32	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Default 20, maximum 20; larger values are clamped, not rejected.
start_date	string	no	Earliest exit_date to include, YYYY-MM-DD, inclusive. Omit for no lower bound.
end_date	string	no	Latest exit_date to include, YYYY-MM-DD, inclusive. Omit for no upper bound. Must not be before start_date — a reversed range is InvalidArgument, not an empty list.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisStockPickerHistoryData	Response payload.
└list	AnalysisStockPickerClosedPosition[]	Closed positions, newest exit_date first.
└symbol	string	Bare ticker, e.g. NVDA.
└code	string	Same as symbol for this product today — the upstream feed carries no exchange suffix here (e.g. MSFT, not MSFT.O).
└locale	string	Always US.
└asset_type	int32	0 = stock.
└narrative	string	Theme the name was picked under, e.g. "AI infrastructure". The portfolio is diversified across themes, so this is the grouping key. English only.
└rationale	string	Why this name was picked, as written by the model at entry. English only.
└join_date	string	Entry date, YYYY-MM-DD.
└exit_date	string	Exit date, YYYY-MM-DD. This is what start_date / end_date filter on.
└entry_price	double	Cost basis.
└exit_price	double	Fill price at exit. 0 on the rare early rows where the pipeline did not record it.
└return	double	Realized return of the position, (exit_price / entry_price - 1) * 100, in percent.
└weight	double	Portfolio weight while held, in percent.
└exit_reason	string	rebalance = swapped out at the weekly rebalance; stop_loss = the stop triggered intraday.
└total	int32	Rows matching the date filter, before paging.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/stock-picker/history
{"page": 1, "size": 20, "start_date": "2026-09-01", "end_date": "2026-09-30"}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "list": [
      {"symbol":"NVDA","code":"NVDA","locale":"US","asset_type":0,
       "narrative":"AI infrastructure","rationale":"...",
       "join_date":"2026-09-01","exit_date":"2026-09-29",
       "entry_price":100,"exit_price":112.35,"return":12.35,"weight":8,
       "exit_reason":"rebalance"}
    ],
    "total": 1
  }
}

POST/v1/openapi/analysis/stock-picker/performance

StockPickerPerformance

Track record of the AI Stock Picker book since inception: headline statistics plus the cumulative-return curve against the S&P 500 on the same time axis. Recomputed upstream once a day after the US close; there is no period parameter.

Every ratio except sharpe_ratio is a percent number here, converted from the fractions the upstream table stores — 12.3 means +12.3%.

If the nightly recompute has not produced a row yet (new deployment, pipeline outage) the endpoint returns ret=ERR_RECORD_NOT_EXIST instead of a payload full of zeros, so a missing track record cannot be mistaken for a flat one.

Statistics here and rows on /stock-picker/history come from the same set of trades (the current strategy generation, early 2026 onward).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisStockPickerPerformanceData	Response payload.
└strategy_name	string	Product name, "AI Stock Picker".
└description	string	One-paragraph strategy description. English only.
└total_return	double	Cumulative return since inception, percent.
└annualized_return	double	Annualized return, percent.
└win_rate	double	Share of closed positions that were profitable, percent.
└max_drawdown	double	Largest peak-to-trough decline of the book, percent, negative.
└sharpe_ratio	double	Sharpe ratio. Not a percent.
└day1_change	double	Change of the book over the latest trading day, percent.
└biggest_lose_return	double	Worst single closed position, percent, negative.
└strategy_curve	AnalysisPerformancePoint[]	Cumulative return of the book, one point per trading day, oldest first.
└date	string	Trading day, YYYY-MM-DD in US Eastern time.
└ts	int64	Unix seconds at 00:00 US Eastern on that trading day. Both curves carry date and ts so one charting routine works for every portfolio product.
└cumulative_return	double	Cumulative return since the first point of the curve, in percent (2.35 = +2.35%).
└sp500_curve	AnalysisPerformancePoint[]	S&P 500 over the same days, same time axis, so the two can be drawn on one chart.
└date	string	Trading day, YYYY-MM-DD in US Eastern time.
└ts	int64	Unix seconds at 00:00 US Eastern on that trading day. Both curves carry date and ts so one charting routine works for every portfolio product.
└cumulative_return	double	Cumulative return since the first point of the curve, in percent (2.35 = +2.35%).

Example request

POST https://api.alphio.ai/v1/openapi/analysis/stock-picker/performance
{}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "strategy_name": "AI Stock Picker",
    "description": "...",
    "total_return": 23.45, "annualized_return": 30.1, "win_rate": 61.23,
    "max_drawdown": -10.2, "sharpe_ratio": 1.23, "day1_change": 0.42,
    "biggest_lose_return": -14.8,
    "strategy_curve": [{"date":"2026-01-02","ts":1767330000,"cumulative_return":0}, {"date":"2026-01-05","ts":1767589200,"cumulative_return":0.8}],
    "sp500_curve":    [{"date":"2026-01-02","ts":1767330000,"cumulative_return":0}, {"date":"2026-01-05","ts":1767589200,"cumulative_return":0.3}]
  }
}

SwingMax Portfolio

SwingMax Portfolio is the book built from SwingMax signals: up to ten US names held by weight, with the same R2/R3 ladder exits as the single-name signals, plus a cash sleeve. It is a separate product from /swingmax/* — those endpoints give you every signal, this one gives you the managed portfolio — and a separate scope: analysis:swingmax-portfolio. Three endpoints: current book, closed trades, performance.

POST/v1/openapi/analysis/swingmax-portfolio/current

SwingmaxPortfolioCurrent

The book as it stands: current positions with weight and unrealized return, the industry breakdown, and the fills of the most recent trading session. One call, no paging — the book holds at most ten names.

Which session latest_trades covers: on weekends and US holidays, the previous trading day; on a trading day before 08:00 ET, also the previous trading day; from 08:00 ET onward, today. Pull before 08:00 ET and you get yesterday's fills.

price is the latest daily close, so return moves once a day. For intraday marks, quote the symbols yourself.

The path is /swingmax-portfolio/, not /swingmax/portfolio: scopes match by path prefix, and this product is granted separately from analysis:swingmax. A key with only analysis:swingmax gets 10403 here.

Request parameters

Field	Type	Required	Description
lang	string	no	Localizes description. "zh" / "zh-cn" = Simplified Chinese, "zh-tw" = Traditional Chinese; anything else, or empty, is English. Takes priority over the lang request header.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisSwingmaxPortfolioCurrentData	Response payload.
└description	string	One-paragraph product description, localized by lang.
└positions	AnalysisSwingmaxPortfolioPosition[]	Current holdings, weight high to low, at most ten.
└symbol	string	Bare ticker, e.g. NVDA.
└name	string	Company name.
└locale	string	Always US.
└asset_type	int32	0 = stock.
└price	double	Latest daily close — not a live quote.
└cost	double	Entry price.
└return	double	(price / cost - 1) * 100, percent, 2 decimals. Unrealized.
└weight	double	Portfolio weight in percent (10 = 10% of the book).
└entry_ts	int64	Entry time, unix seconds.
└industries	AnalysisSwingmaxPortfolioIndustry[]	Industry breakdown of the book, Cash first then by weight.
└industry	string	TRBC industry group name. The first row is always "Cash" — the part of the book not invested.
└weight	double	Percent of the book. All rows add up to 100.
└latest_trades	AnalysisSwingmaxPortfolioTrade[]	Fills of the most recent trading session, newest first. Empty on a session with no trades.
└symbol	string	Bare ticker, e.g. NVDA.
└name	string	Company name.
└locale	string	Always US.
└asset_type	int32	0 = stock.
└signal_ts	int64	Fill time, unix seconds.
└filled_price	double	Fill price.
└weight_before	double	Weight of this name in the book before the fill, percent.
└weight_after	double	Weight after the fill, percent. 0 on a full exit.
└event	string	BUY, R2, R3 or SELL — the same ladder names as the /swingmax signals: R2 and R3 are partial profit-taking at the first and second targets, SELL is the final exit (stop or max hold).
└side	int32	1 = buy, 2 = sell. R2, R3 and SELL are all sells.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/swingmax-portfolio/current
{"lang": ""}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "description": "SwingMax Portfolio is a curated AI portfolio built from the best SwingMax signals. ...",
    "positions": [
      {"symbol":"AAPL","name":"Apple Inc","locale":"US","asset_type":0,
       "price":230.5,"cost":220,"return":4.77,"weight":9.5,"entry_ts":1759723200}
    ],
    "industries": [
      {"industry":"Cash","weight":12.5},
      {"industry":"Software & IT Services","weight":87.5}
    ],
    "latest_trades": [
      {"symbol":"NVDA","name":"NVIDIA Corp","locale":"US","asset_type":0,
       "signal_ts":1760000000,"filled_price":180.25,
       "weight_before":9.8,"weight_after":4.9,"event":"R2","side":2}
    ]
  }
}

POST/v1/openapi/analysis/swingmax-portfolio/history

SwingmaxPortfolioHistory

Closed trades of the SwingMax Portfolio book, one row per completed round trip: entry and exit time and price, weight while held, realized return. Paged and sorted upstream, so all four sort orders are available.

No keyword on this endpoint: the upstream query has no symbol filter. To find one name, walk the pages.

A name can appear several times — each round trip is its own row, and the book re-enters names it has traded before.

Request parameters

Field	Type	Required	Description
page	int32	no	1-based. 0 or negative is treated as unset and means page 1.
size	int32	no	Rows per page. Default 20, maximum 20; larger values are clamped, not rejected.
sort_type	int32	no	0 or 1 = newest entry first, 2 = oldest entry first, 3 = highest realized_return first, 4 = lowest first. The default is newest first, not the upstream table's natural order.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisSwingmaxPortfolioHistoryData	Response payload.
└list	AnalysisSwingmaxPortfolioClosedTrade[]	Closed trades in the requested order.
└symbol	string	Bare ticker, e.g. NVDA.
└name	string	Company name.
└locale	string	Always US.
└asset_type	int32	0 = stock.
└entry_ts	int64	Entry time, unix seconds.
└entry_price	double	Entry price.
└exit_ts	int64	Exit time, unix seconds.
└exit_price	double	Exit price.
└weight	double	Portfolio weight while held, in percent.
└realized_return	double	Realized return of the trade, percent, 2 decimals.
└total	int32	All closed trades, before paging.

Example request

POST https://api.alphio.ai/v1/openapi/analysis/swingmax-portfolio/history
{"page": 1, "size": 20, "sort_type": 3}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "list": [
      {"symbol":"TSLA","name":"Tesla Inc","locale":"US","asset_type":0,
       "entry_ts":1757376000,"entry_price":100,"exit_ts":1759104000,"exit_price":110,
       "weight":9.9,"realized_return":10}
    ],
    "total": 1
  }
}

POST/v1/openapi/analysis/swingmax-portfolio/performance

SwingmaxPortfolioPerformance

Performance of the SwingMax Portfolio book over a chosen window: the book's return and the S&P 500's over the same window, both cumulative-return curves on one time axis, and the since-inception statistics.

Long windows are downsampled so the curves stay drawable: up to ~180 trading days daily, up to ~520 weekly, beyond that monthly, and period 5 is always weekly. The last point is always kept, so the curve's end matches portfolio_return.

stats is the same since-inception block regardless of period; only portfolio_return, sp500_return and the two curves follow the window.

Request parameters

Field	Type	Required	Description
period	int32	no	Window for the curves and the two window returns: 0 or 5 = everything since 2024-01-01, 1 = last 1 month, 2 = last 3 months, 3 = last 6 months, 4 = last 1 year. Any other value is InvalidArgument.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	AnalysisSwingmaxPortfolioPerformanceData	Response payload.
└period	int32	The window actually applied — 0 is echoed back as 5.
└portfolio_return	double	Return of the book over the window, percent: last capital in the window against the first.
└sp500_return	double	S&P 500 (SPY close) over the same window, percent.
└stats	AnalysisSwingmaxPortfolioStats	Since-inception statistics. Independent of period.
└total_return	double	Cumulative return since the book's inception, percent.
└annualized_return	double	Annualized return, percent.
└day1_change	double	Change of the book over the latest trading day, percent.
└sharpe_ratio	double	Sharpe ratio. Not a percent.
└portfolio_curve	AnalysisPerformancePoint[]	Cumulative return of the book inside the window, rebased to 0 at the first point. Aligned to SPY's trading days; days the book has no record for carry the previous value forward.
└date	string	Trading day, YYYY-MM-DD in US Eastern time.
└ts	int64	Unix seconds at 00:00 US Eastern on that trading day. Both curves carry date and ts so one charting routine works for every portfolio product.
└cumulative_return	double	Cumulative return since the first point of the curve, in percent (2.35 = +2.35%).
└sp500_curve	AnalysisPerformancePoint[]	S&P 500 cumulative return over the same days, same time axis.
└date	string	Trading day, YYYY-MM-DD in US Eastern time.
└ts	int64	Unix seconds at 00:00 US Eastern on that trading day. Both curves carry date and ts so one charting routine works for every portfolio product.
└cumulative_return	double	Cumulative return since the first point of the curve, in percent (2.35 = +2.35%).

Example request

POST https://api.alphio.ai/v1/openapi/analysis/swingmax-portfolio/performance
{"period": 2}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "period": 2,
    "portfolio_return": 6.4,
    "sp500_return": 3.1,
    "stats": {"total_return": 40.1, "annualized_return": 20.2, "day1_change": -0.3, "sharpe_ratio": 1.4},
    "portfolio_curve": [{"date":"2026-07-15","ts":1784088000,"cumulative_return":0}, {"date":"2026-07-16","ts":1784174400,"cumulative_return":0.5}],
    "sp500_curve":     [{"date":"2026-07-15","ts":1784088000,"cumulative_return":0}, {"date":"2026-07-16","ts":1784174400,"cumulative_return":0.2}]
  }
}

Whales tracker

Endpoints tracking institutional (13F) and Congressional holdings. entity_type: 1=institution, 2=congress member. seo_name is the stable partner-facing identifier. All return/chart values are decimal fractions (0.012 = +1.2%). Rankings are cached ~10s server-side.

POST/v1/openapi/whales/top-returns

TopReturns

Institutions ranked by the chosen period's return, descending. Institutions only — for congress members use Ranking with entity_type=2.

Request parameters

Field	Type	Required	Description
period	string	yes	"1d" / "1w" / "1m" / "3m" / "1y".
limit	int32	yes	1 – 50.
include	string	no	Comma-separated: "chart" (equity curve) / "holdings" (top_holdings ≤2 tickers + holding_count) / "performance" (sharpe_ratio + mdd) / "returns" (return_1d / return_1w / return_1m / return_1y). Empty = base fields only.
keyword	string	no	Fuzzy entity-name filter (matches names containing the keyword). Empty = no filtering.
sort	string	no	"<field>:<asc\|desc>", one field at a time. Fields: total_return (the requested period's return), sharpe_ratio, return_1d, return_1w, return_1m, return_1y, mdd. Empty = period return, descending. Sorting works regardless of include — include only controls which fields are returned.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesTopReturnsData	Response payload.
└entities	WhalesEntity[]	Ranked by return_value descending by default; sort overrides the order.
└name	string	Entity display name.
└action	string	Currently always "Holding" on ranking endpoints; empty on FollowingRanking.
└return_value	double	Return for the requested period, decimal fraction (0.618 = +61.8%). TopReturnsByAsset always uses the 3m return.
└top_holdings	WhalesHolding[]	Largest positions, tickers only. Only populated when include contains holdings (capped at 2-3 tickers).
└ticker	string	Ticker symbol.
└holding_count	int32	Number of positions in the latest filing. 0 unless include contains holdings.
└chart	WhalesChartPoint[]	Equity-curve points. Only populated when include contains chart.
└date	string	YYYY-MM-DD.
└value	double	Cumulative return since tracking began, decimal fraction (0.618 = +61.8%).
└avatar	string	CDN avatar URL.
└entity_type	int32	1=institution, 2=congress.
└seo_name	string	Stable slug — the partner-facing identifier, feed it to DetailsBySeoName.
└sharpe_ratio	double	Sharpe ratio of the tracked portfolio. Only populated when include contains performance.
└mdd	double	Maximum drawdown, computed from the equity curve. Positive decimal fraction (0.2 = 20% drawdown). Only populated when include contains performance.
└return_1d	double	1-day return, decimal fraction (0.012 = +1.2%). Only populated when include contains returns.
└return_1w	double	1-week return, decimal fraction. Only populated when include contains returns.
└return_1m	double	1-month return, decimal fraction. Only populated when include contains returns.
└return_1y	double	1-year return, decimal fraction. Only populated when include contains returns.

POST/v1/openapi/whales/ranking

Ranking

Ranking filtered by entity_type. Query one period (period → data.entities) or up to 4 at once (periods → data.period_groups).

top_holdings is always empty on this endpoint for partner API calls.

Sending both period and periods returns an InvalidArgument error.

Request parameters

Field	Type	Required	Description
period	string	no	Single-period query: "1d" / "1w" / "1m" / "3m" / "1y". Exactly one of period / periods must be set.
periods	string[]	no	Multi-period query: up to 4 unique values from the same set. Response returns one period_groups entry per requested period, in request order.
limit	int32	yes	1 – 50 (applies to each period's list).
include	string	no	Comma-separated: "chart" (mini equity curve per entity) / "performance" (sharpe_ratio + mdd) / "returns" (return_1d / return_1w / return_1m / return_1y).
entity_type	WhalesEntityType	yes	INSTITUTION (1) / CONGRESS (2).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesRankingData	Response payload.
└entities	WhalesEntity[]	Populated for single-period requests.
└name	string	Entity display name.
└action	string	Currently always "Holding" on ranking endpoints; empty on FollowingRanking.
└return_value	double	Return for the requested period, decimal fraction (0.618 = +61.8%). TopReturnsByAsset always uses the 3m return.
└top_holdings	WhalesHolding[]	Largest positions, tickers only. Only populated when include contains holdings (capped at 2-3 tickers).
└ticker	string	Ticker symbol.
└holding_count	int32	Number of positions in the latest filing. 0 unless include contains holdings.
└chart	WhalesChartPoint[]	Equity-curve points. Only populated when include contains chart.
└date	string	YYYY-MM-DD.
└value	double	Cumulative return since tracking began, decimal fraction (0.618 = +61.8%).
└avatar	string	CDN avatar URL.
└entity_type	int32	1=institution, 2=congress.
└seo_name	string	Stable slug — the partner-facing identifier, feed it to DetailsBySeoName.
└sharpe_ratio	double	Sharpe ratio of the tracked portfolio. Only populated when include contains performance.
└mdd	double	Maximum drawdown, computed from the equity curve. Positive decimal fraction (0.2 = 20% drawdown). Only populated when include contains performance.
└return_1d	double	1-day return, decimal fraction (0.012 = +1.2%). Only populated when include contains returns.
└return_1w	double	1-week return, decimal fraction. Only populated when include contains returns.
└return_1m	double	1-month return, decimal fraction. Only populated when include contains returns.
└return_1y	double	1-year return, decimal fraction. Only populated when include contains returns.
└period_groups	WhalesRankingPeriodGroup[]	Populated for multi-period requests, in the same order as req.periods.
└period	string	The requested period this group belongs to.
└entities	WhalesEntity[]	Ranking list for this period.
└name	string	Entity display name.
└action	string	Currently always "Holding" on ranking endpoints; empty on FollowingRanking.
└return_value	double	Return for the requested period, decimal fraction (0.618 = +61.8%). TopReturnsByAsset always uses the 3m return.
└top_holdings	WhalesHolding[]	Largest positions, tickers only. Only populated when include contains holdings (capped at 2-3 tickers).
└ticker	string	Ticker symbol.
└holding_count	int32	Number of positions in the latest filing. 0 unless include contains holdings.
└chart	WhalesChartPoint[]	Equity-curve points. Only populated when include contains chart.
└date	string	YYYY-MM-DD.
└value	double	Cumulative return since tracking began, decimal fraction (0.618 = +61.8%).
└avatar	string	CDN avatar URL.
└entity_type	int32	1=institution, 2=congress.
└seo_name	string	Stable slug — the partner-facing identifier, feed it to DetailsBySeoName.
└sharpe_ratio	double	Sharpe ratio of the tracked portfolio. Only populated when include contains performance.
└mdd	double	Maximum drawdown, computed from the equity curve. Positive decimal fraction (0.2 = 20% drawdown). Only populated when include contains performance.
└return_1d	double	1-day return, decimal fraction (0.012 = +1.2%). Only populated when include contains returns.
└return_1w	double	1-week return, decimal fraction. Only populated when include contains returns.
└return_1m	double	1-month return, decimal fraction. Only populated when include contains returns.
└return_1y	double	1-year return, decimal fraction. Only populated when include contains returns.

POST/v1/openapi/whales/top-returns-by-asset

TopReturnsByAsset

Reverse-lookup: institutions currently holding a specific ticker, ranked by their 3-month return (descending). return_value is always the 3m return.

Request parameters

Field	Type	Required	Description
asset	WhalesAsset	yes	The ticker to look up.
└asset_type	int32	no	0=stock, 1=etf.
└ticker	string	yes	Ticker symbol, e.g. "AAPL".
└locale	string	no	"US" (default) / "HK".
include	string	no	Comma-separated: "chart" / "holdings" / "performance" / "returns". With holdings, the queried ticker is guaranteed to appear in top_holdings.
limit	int32	no	Defaults to 6 when omitted.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesTopReturnsData	Response payload.
└entities	WhalesEntity[]	Ranked by 3m return, descending.
└name	string	Entity display name.
└action	string	Currently always "Holding" on ranking endpoints; empty on FollowingRanking.
└return_value	double	Return for the requested period, decimal fraction (0.618 = +61.8%). TopReturnsByAsset always uses the 3m return.
└top_holdings	WhalesHolding[]	Largest positions, tickers only. Only populated when include contains holdings (capped at 2-3 tickers).
└ticker	string	Ticker symbol.
└holding_count	int32	Number of positions in the latest filing. 0 unless include contains holdings.
└chart	WhalesChartPoint[]	Equity-curve points. Only populated when include contains chart.
└date	string	YYYY-MM-DD.
└value	double	Cumulative return since tracking began, decimal fraction (0.618 = +61.8%).
└avatar	string	CDN avatar URL.
└entity_type	int32	1=institution, 2=congress.
└seo_name	string	Stable slug — the partner-facing identifier, feed it to DetailsBySeoName.
└sharpe_ratio	double	Sharpe ratio of the tracked portfolio. Only populated when include contains performance.
└mdd	double	Maximum drawdown, computed from the equity curve. Positive decimal fraction (0.2 = 20% drawdown). Only populated when include contains performance.
└return_1d	double	1-day return, decimal fraction (0.012 = +1.2%). Only populated when include contains returns.
└return_1w	double	1-week return, decimal fraction. Only populated when include contains returns.
└return_1m	double	1-month return, decimal fraction. Only populated when include contains returns.
└return_1y	double	1-year return, decimal fraction. Only populated when include contains returns.

POST/v1/openapi/whales/following/rankingX-Partner-User required

FollowingRanking

Ranks the entities the calling user follows (their watchlist) by the chosen period's return, descending. Requires X-Partner-User, like the watchlist endpoints.

A user's first-ever call auto-seeds their watchlist with the featured entities from Home.

Request parameters

Field	Type	Required	Description
period	string	yes	"1d" / "1w" / "1m" / "3m" / "1y".
limit	int32	yes	1 – 50. Truncates the sorted list.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesFollowingRankingData	Response payload.
└entities	WhalesEntity[]	Only name, return_value, avatar, entity_type and seo_name are populated here — no action / top_holdings / holding_count / chart / performance metrics.
└name	string	Entity display name.
└action	string	Currently always "Holding" on ranking endpoints; empty on FollowingRanking.
└return_value	double	Return for the requested period, decimal fraction (0.618 = +61.8%). TopReturnsByAsset always uses the 3m return.
└top_holdings	WhalesHolding[]	Largest positions, tickers only. Only populated when include contains holdings (capped at 2-3 tickers).
└ticker	string	Ticker symbol.
└holding_count	int32	Number of positions in the latest filing. 0 unless include contains holdings.
└chart	WhalesChartPoint[]	Equity-curve points. Only populated when include contains chart.
└date	string	YYYY-MM-DD.
└value	double	Cumulative return since tracking began, decimal fraction (0.618 = +61.8%).
└avatar	string	CDN avatar URL.
└entity_type	int32	1=institution, 2=congress.
└seo_name	string	Stable slug — the partner-facing identifier, feed it to DetailsBySeoName.
└sharpe_ratio	double	Sharpe ratio of the tracked portfolio. Only populated when include contains performance.
└mdd	double	Maximum drawdown, computed from the equity curve. Positive decimal fraction (0.2 = 20% drawdown). Only populated when include contains performance.
└return_1d	double	1-day return, decimal fraction (0.012 = +1.2%). Only populated when include contains returns.
└return_1w	double	1-week return, decimal fraction. Only populated when include contains returns.
└return_1m	double	1-month return, decimal fraction. Only populated when include contains returns.
└return_1y	double	1-year return, decimal fraction. Only populated when include contains returns.

POST/v1/openapi/whales/institutions/popular-trades

InstitutionsPopularTrades

Most net-bought / net-sold tickers, aggregated across all tracked institutions' latest quarterly (13F) filings and ranked by aggregate share change.

Request parameters

Field	Type	Required	Description
locale	string	yes	"us" = US stocks + ETFs / "hk" = HK stocks.
tab	string	yes	"increase" = largest aggregate share increases (net buying) / "decrease" = largest aggregate decreases (net selling).
limit	int32	yes	1 – 50.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesInstitutionsPopularTradesData	Response payload.
└trades	WhalesPopularTrade[]	One row per ticker, ordered by \|share_change\| (largest net buys first on increase, largest net sells first on decrease).
└ticker	string	Ticker symbol, e.g. "NVDA".
└code	string	Exchange-qualified code, e.g. "NVDA.O".
└name	string	Security display name.
└asset_type	int32	0=stock, 1=etf.
└locale	string	"US" / "HK".
└share_change	double	Net change in shares held vs the previous quarter, summed across all tracked institutions. Positive = net buying, negative = net selling. This is the ranking key.
└holding_mv	double	Total market value (USD) of the position summed across all tracked institutions, per the latest filings.

POST/v1/openapi/whales/details

Details

Full entity detail by id (from any ranking response).

Request parameters

Field	Type	Required	Description
id	int32	yes	≥ 1.
include	string	no	Comma-separated: "chart" (equity curves), "trade_update" (recent trades + last_updated), "current_holdings" (sector distribution). Base fields and positions are always returned.
entity_type	WhalesEntityType	yes	INSTITUTION (1) / CONGRESS (2).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesEntityDetails	Response payload.
└name	string	Entity display name.
└is_watched	bool	Whether the requesting user watches this entity. Always false on partner API calls — no user context is forwarded on this endpoint.
└top20_holding_mv	int64	Aggregate market value (USD) of the entity's 20 largest positions, per the latest filing.
└last_updated	string	Date (YYYY-MM-DD) of the most recent reported portfolio change. Only populated when include contains trade_update.
└holding_count	int32	Number of positions currently held.
└description	string	Short text description of the entity.
└performance	WhalesPerformance	Tracked-portfolio performance stats. Returns are decimal fractions (0.012 = +1.2%).
└return_1d	double	Latest 1-day return, decimal fraction.
└sharpe_ratio	double	Sharpe ratio of the tracked portfolio.
└total_return	double	Cumulative return since tracking began, decimal fraction.
└annualized_return	double	Annualized return, decimal fraction.
└charts	WhalesCharts	Equity curves. Only populated when include contains chart.
└strategy	WhalesChartPoint[]	Entity portfolio equity curve.
└date	string	YYYY-MM-DD.
└value	double	Cumulative return since tracking began, decimal fraction (0.618 = +61.8%).
└sp500	WhalesChartPoint[]	S&P 500 benchmark over the same window.
└date	string	YYYY-MM-DD.
└value	double	Cumulative return since tracking began, decimal fraction (0.618 = +61.8%).
└trade_updates	WhalesTradeUpdate[]	Portfolio changes from the most recent rebalance/filing. Only populated when include contains trade_update.
└date	string	Rebalance date, YYYY-MM-DD.
└action	string	"buy" / "sell" / "hold" — derived from the weight change (rounded to 2 decimals).
└ticker	string	Ticker symbol, e.g. "NVDA".
└code	string	Exchange-qualified code, e.g. "NVDA.O".
└name	string	Security display name.
└asset_type	int32	0=stock, 1=etf.
└locale	string	"US" / "HK".
└old_percent	double	Portfolio weight (%) before the change.
└new_percent	double	Portfolio weight (%) after the change.
└fill_price	double	Price at the rebalance date.
└current_holdings	WhalesCurrentHoldings	Current portfolio. positions is always returned; sector_distribution requires include=current_holdings.
└sector_distribution	WhalesSectorWeight[]	Sector weights of the current portfolio. Only populated when include contains current_holdings.
└sector	string	Sector name; "other" when unclassified.
└weight	double	Weight of the sector, in %.
└positions	WhalesPosition[]	All current positions.
└ticker	string	Ticker symbol, e.g. "NVDA".
└code	string	Exchange-qualified code, e.g. "NVDA.O".
└name	string	Security display name.
└asset_type	int32	0=stock, 1=etf.
└locale	string	"US" / "HK".
└cost_price	double	Average cost per share.
└current_percent	double	Current portfolio weight (%).
└total_return	double	Cumulative return of the position since entry.
└sector_distribution_desc	string	Server-generated text summary of the sector mix.
└entity_type	int32	1=institution, 2=congress.
└institution_profile	WhalesInstitutionProfile	Only set for institutions (entity_type=1).
└person_name	string	Fund manager / key person name.
└avatar	string	CDN avatar URL.
└profile	string	Manager / institution bio.
└investing_philosophy	string	Investing-philosophy blurb.
└congress_member_info	WhalesCongressMemberInfo	Only set for congress members (entity_type=2).
└avatar	string	Avatar URL.
└party	string	Political party, e.g. "Republican" / "Democrat".
└organization	string	Chamber, e.g. "Senate" / "House".
└faq	WhalesFAQInfo[]	Server-generated FAQ entries derived from the stats above (holding count, returns, AUM).
└question	string	—
└answer	string	—

POST/v1/openapi/whales/details-by-seo-name

DetailsBySeoName

Same as Details but keyed by stable seo_name (from any ranking / search / home response). Preferred when you cached an entity earlier.

Request parameters

Field	Type	Required	Description
seo_name	string	yes	Stable entity slug, min_len 1.
include	string	no	Comma-separated: "chart" (equity curves), "trade_update" (recent trades + last_updated), "current_holdings" (sector distribution). Base fields and positions are always returned.
entity_type	WhalesEntityType	yes	INSTITUTION (1) / CONGRESS (2).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesEntityDetails	Response payload.
└name	string	Entity display name.
└is_watched	bool	Whether the requesting user watches this entity. Always false on partner API calls — no user context is forwarded on this endpoint.
└top20_holding_mv	int64	Aggregate market value (USD) of the entity's 20 largest positions, per the latest filing.
└last_updated	string	Date (YYYY-MM-DD) of the most recent reported portfolio change. Only populated when include contains trade_update.
└holding_count	int32	Number of positions currently held.
└description	string	Short text description of the entity.
└performance	WhalesPerformance	Tracked-portfolio performance stats. Returns are decimal fractions (0.012 = +1.2%).
└return_1d	double	Latest 1-day return, decimal fraction.
└sharpe_ratio	double	Sharpe ratio of the tracked portfolio.
└total_return	double	Cumulative return since tracking began, decimal fraction.
└annualized_return	double	Annualized return, decimal fraction.
└charts	WhalesCharts	Equity curves. Only populated when include contains chart.
└strategy	WhalesChartPoint[]	Entity portfolio equity curve.
└date	string	YYYY-MM-DD.
└value	double	Cumulative return since tracking began, decimal fraction (0.618 = +61.8%).
└sp500	WhalesChartPoint[]	S&P 500 benchmark over the same window.
└date	string	YYYY-MM-DD.
└value	double	Cumulative return since tracking began, decimal fraction (0.618 = +61.8%).
└trade_updates	WhalesTradeUpdate[]	Portfolio changes from the most recent rebalance/filing. Only populated when include contains trade_update.
└date	string	Rebalance date, YYYY-MM-DD.
└action	string	"buy" / "sell" / "hold" — derived from the weight change (rounded to 2 decimals).
└ticker	string	Ticker symbol, e.g. "NVDA".
└code	string	Exchange-qualified code, e.g. "NVDA.O".
└name	string	Security display name.
└asset_type	int32	0=stock, 1=etf.
└locale	string	"US" / "HK".
└old_percent	double	Portfolio weight (%) before the change.
└new_percent	double	Portfolio weight (%) after the change.
└fill_price	double	Price at the rebalance date.
└current_holdings	WhalesCurrentHoldings	Current portfolio. positions is always returned; sector_distribution requires include=current_holdings.
└sector_distribution	WhalesSectorWeight[]	Sector weights of the current portfolio. Only populated when include contains current_holdings.
└sector	string	Sector name; "other" when unclassified.
└weight	double	Weight of the sector, in %.
└positions	WhalesPosition[]	All current positions.
└ticker	string	Ticker symbol, e.g. "NVDA".
└code	string	Exchange-qualified code, e.g. "NVDA.O".
└name	string	Security display name.
└asset_type	int32	0=stock, 1=etf.
└locale	string	"US" / "HK".
└cost_price	double	Average cost per share.
└current_percent	double	Current portfolio weight (%).
└total_return	double	Cumulative return of the position since entry.
└sector_distribution_desc	string	Server-generated text summary of the sector mix.
└entity_type	int32	1=institution, 2=congress.
└institution_profile	WhalesInstitutionProfile	Only set for institutions (entity_type=1).
└person_name	string	Fund manager / key person name.
└avatar	string	CDN avatar URL.
└profile	string	Manager / institution bio.
└investing_philosophy	string	Investing-philosophy blurb.
└congress_member_info	WhalesCongressMemberInfo	Only set for congress members (entity_type=2).
└avatar	string	Avatar URL.
└party	string	Political party, e.g. "Republican" / "Democrat".
└organization	string	Chamber, e.g. "Senate" / "House".
└faq	WhalesFAQInfo[]	Server-generated FAQ entries derived from the stats above (holding count, returns, AUM).
└question	string	—
└answer	string	—

POST/v1/openapi/whales/home

Home

Static editorial list of featured entities (ARK, Nancy Pelosi, Berkshire Hathaway, Soros, Renaissance, Pershing Square). Empty body.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesHomeData	Response payload.
└items	WhalesHomeInfo[]	Curated, fixed order — not a ranking.
└name	string	Entity display name.
└cover	string	CDN cover-image URL for the home card.
└label	string	Editorial tag; currently always "Featured".
└entity_type	int32	1=institution, 2=congress.
└seo_name	string	Feed to DetailsBySeoName.

POST/v1/openapi/whales/search

Case-insensitive substring match over all institution and congress-member names.

Request parameters

Field	Type	Required	Description
keyword	string	yes	min_len 1. Must contain a non-whitespace character.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesSearchData	Response payload.
└items	WhalesSearchInfo[]	Capped at 50 results (institutions prioritized up to 25, remainder congress members).
└name	string	Entity display name.
└entity_type	int32	1=institution, 2=congress.
└seo_name	string	Feed to DetailsBySeoName.

POST/v1/openapi/whales/faq

FAQ

Static product-level FAQ about the Whales tracker, localized via Accept-Language (English default, Chinese supported). Not entity-specific — per-entity FAQ comes in Details.faq. Empty body.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesFAQData	Response payload.
└items	WhalesFAQInfo[]	Fixed question/answer list.
└question	string	—
└answer	string	—

POST/v1/openapi/whales/watchlist/listX-Partner-User required

WatchlistList

List entities the user follows, with their returns across all periods. Requires X-Partner-User.

A user's first-ever call auto-seeds their watchlist with the featured entities from Home. If they later unfollow everything, the list stays empty.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	WhalesWatchlistListData	Response payload.
└items	WhalesWatchlistItem[]	One row per followed entity.
└name	string	Entity display name.
└entity_type	int32	1=institution, 2=congress.
└return_1d	double	1-day return, decimal fraction (0.012 = +1.2%).
└return_7d	double	7-day return, decimal fraction.
└return_1m	double	1-month return, decimal fraction.
└return_3m	double	3-month return, decimal fraction.
└return_1y	double	1-year return, decimal fraction.
└avatar	string	CDN avatar URL.
└seo_name	string	Stable slug for DetailsBySeoName.

POST/v1/openapi/whales/watchlist/createX-Partner-User required

WatchlistCreate

Add an entity to the calling user's watchlist. Requires X-Partner-User.

Request parameters

Field	Type	Required	Description
id	int32	yes	Internal numeric entity id, ≥ 1.
entity_type	WhalesEntityType	yes	INSTITUTION (1) / CONGRESS (2).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success.
msg	string	Status / error message.

POST/v1/openapi/whales/watchlist/deleteX-Partner-User required

WatchlistDelete

Remove an entity from the calling user's watchlist. Requires X-Partner-User.

Request parameters

Field	Type	Required	Description
id	int32	yes	Internal numeric entity id, ≥ 1.
entity_type	WhalesEntityType	yes	INSTITUTION (1) / CONGRESS (2).

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success.
msg	string	Status / error message.

Market data

Raw market data endpoints powering Chat card details. Keyed by symbol; partner UIs typically fetch these on-demand when expanding a card.

POST/v1/openapi/data/valuation

DataValuation

Historical valuation time-series.

Request parameters

Field	Type	Required	Description
ticker	string	yes	e.g. "AAPL".
valuation_type	int32	yes	EValuationType. 0=PE / 1=PB / 2=PS / 3=Price-to-FCF / 4=EV-EBITDA / 5=Price-to-OCF / 6=FCF-Yield / 7=EV-EBIT.
is_forward	bool	no	true=forward, false=trailing (default).
from	string	no	YYYY-MM-DD. Default last 1Y.
to	string	no	YYYY-MM-DD. Default today.
locale	string	no	"US" / "HK".

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	DataValuationData	Response payload.
└valuation	DataValuationItem[]	Time series.
└t	string	Date (YYYY-MM-DD).
└v	string	Valuation value.
└variance_info	DataVarianceInfo	Five reference bands.
└mean	string	—
└variance	string	—
└overvalued	string	mean + 1σ.
└undervalued	string	mean − 1σ.
└strongly_overvalued	string	mean + 2σ.
└strongly_undervalued	string	mean − 2σ.

Example request

POST https://api.alphio.ai/v1/openapi/data/valuation
{ "ticker":"AAPL", "valuation_type":0, "from":"2025-06-01", "to":"2026-06-01", "locale":"US" }

Example response

{
  "ret": 0, "msg": "ok",
  "data": {
    "valuation":[{"t":"2025-06-01","v":"28.5"},{"t":"2025-06-02","v":"28.7"}, ...],
    "variance_info":{
      "mean":"30.2","variance":"3.1",
      "overvalued":"33.3","undervalued":"27.1",
      "strongly_overvalued":"36.4","strongly_undervalued":"24.0"
    }
  }
}

POST/v1/openapi/data/rating

DataRating

Analyst ratings.

Request parameters

Field	Type	Required	Description
ticker	string	yes	e.g. "AAPL".
from	string	no	YYYY-MM-DD.
to	string	no	YYYY-MM-DD.
page	int32	no	1-indexed.
size	int32	no	1 – 100, default 20.
locale	string	no	"US" / "HK".

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	DataRatingData	Response payload.
└results	DataStockAnalysis	—
└ticker	string	—
└code	string	—
└analyses	DataAnalysisItem[]	—
└rating_analyst	string	—
└rating_date	string	—
└rating_action	string	"upgrade" / "downgrade" / "maintain" / "initiate".
└rating_analyst_stars	double	—
└rating_analyst_firm	string	—
└rating_price_target	string	—
└rating	string	e.g. "buy" / "hold".
└rating_upside	double	—
└analysis_summary	DataAnalysisSummary	Aggregate snapshot.
└code	string	—
└symbol	string	—
└average_price_target	string	—
└buy	int32	Count of buy ratings.
└content	string[]	Highlight bullets.
└highest_price_target	string	—
└hold	int32	Count of hold ratings.
└lowest_price_target	string	—
└pagination	DataPagination	—
└current_page	int32	—
└total_pages	int32	—
└total_items	int32	—

POST/v1/openapi/data/snapshot

DataSnapshot

Financial-metric snapshot (per-quarter or annual).

Request parameters

Field	Type	Required	Description
code	string	yes	—
type	int32	yes	EFinancialIndicatorsType. 1=EPS / 2=Revenue / 3=ROE / 4=Gross Margin / 5=Net Margin / 6=Current Ratio / 7=Debt to Equity / 8=Operating Margin.
report_type	int32	yes	FinanceReportType. 1=annual / 2=quarterly.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	DataSnapshotData	Response payload.
└columns	DataSnapshotColumn[]	Each column = one metric series.
└name	string	—
└values	DataSnapshotValue[]	One cell per period.
└title	string	e.g. "Q1 2026".
└value	double	—
└cur_type	string	Currency / unit.
└valid	bool	false → missing data; render as "—".

POST/v1/openapi/data/kline

DataKline

OHLCV bars.

Request parameters

Field	Type	Required	Description
tickers	WhalesAsset[]	yes	Max 20 per request.
└asset_type	int32	no	SymbolType: 0=stock, 1=etf, 2=crypto, 3=commodity, 4=forex, 5=perp.
└ticker	string	no	e.g. "TSLA".
└locale	string	no	"US" / "HK".
from	string	no	YYYY-MM-DD (US-Eastern) or unix seconds.
to	string	no	YYYY-MM-DD (US-Eastern) or unix seconds.
multiplier	int32	no	e.g. 1, 5, 15.
timespan	int32	no	TimespanOpt. 2=minute, 3=hour, 4=day, 5=week, 6=month, 7=quarter, 8=year. Values outside 2–8 are rejected.
limit	int32	no	Max 1000. Default 500.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	DataKlineData	Response payload.
└series	DataKlineSeries[]	One series per requested ticker.
└ticker	string	—
└asset_type	int32	—
└points	DataKlinePoint[]	OHLCV bars. Numbers are strings to avoid precision loss.
└open	string	—
└high	string	—
└low	string	—
└close	string	—
└volume	string	Traded value (turnover) for the bar — per proto this is 成交额, not share count.
└timestamp	int64	Unix seconds.

Example request

POST https://api.alphio.ai/v1/openapi/data/kline
{
  "tickers":[ { "asset_type":0, "ticker":"TSLA", "locale":"US" } ],
  "from":"2026-06-09", "to":"2026-07-09",
  "multiplier":1, "timespan":4, "limit":30
}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "series": [
      {
        "ticker": "TSLA",
        "asset_type": 0,
        "points": [
          {"open":"391.09","high":"396.14","low":"388.42","close":"394.26",
           "volume":"58231290","timestamp":"1783486800"}
        ]
      }
    ]
  }
}

POST/v1/openapi/data/price-prediction

DataPricePrediction

Historical OHLC bars backing the price-prediction card — no forecast fields in the payload itself. Request is a discriminated union by ticker_type; response has matching stock_data / crypto_data.

Request parameters

Field	Type	Required	Description
ticker_type	string	yes	"stock" / "etf" / "" → stock branch; "crypto" → crypto branch.
stock	DataPricePredictionStockReq	no	Required when ticker_type is stock / etf / empty.
└code	string	no	Ticker, e.g. "AAPL".
└multiplier	int32	no	Time multiplier.
└timespan	int32	no	TimespanOpt. 2=minute, 3=hour, 4=day, 5=week, 6=month, 7=quarter, 8=year.
└timestamp	int64	no	Anchor end-time (unix seconds). Empty = now.
└count	int32	no	Bars to return. Max 5000, recommended 800.
└locale	string	no	"US" / "HK".
crypto	DataPricePredictionCryptoReq	no	Required when ticker_type = "crypto".
└type	int32	no	QueryType. 1=crypto spot, 2=etf, 3=crypto derivative (HL perps).
└symbol	string	no	—
└multiplier	int32	no	—
└timespan	int32	no	TimespanOpt. 2=minute, 3=hour, 4=day, 5=week, 6=month, 7=quarter, 8=year.
└limit	int32	no	—
└start_time	int64	no	Unix seconds.
└end_time	int64	no	Unix seconds.
└symbol_list	string[]	no	Batch query — per-symbol series come back in crypto_data.infos.
└locale	string	no	Source platform for multi-source assets, e.g. "HYPERLIQUID". Empty = default source.

Response fields

Field	Type	Description
ret	int32	Business return code. 0 = success, non-zero = error code (see Errors).
msg	string	Human-readable status / error message. "ok" on success.
data	DataPricePredictionData	Response payload.
└ticker_type	string	Echoed back.
└stock_data	DataPricePredictionStockData	Present on the stock branch.
└points	DataPricePredictionStockPoint[]	Historical bars, oldest first.
└code	string	—
└close	double	—
└high	double	—
└low	double	—
└transactions	int64	Share volume (per proto: 交易量).
└open	double	—
└timestamp	int64	Unix seconds.
└volume	double	Traded value / turnover (per proto: 成交额), not share count.
└vwap	double	Volume-weighted average price.
└otc	bool	—
└pre_close	double	Previous-period values (pre_* mirror the base fields).
└pre_high	double	—
└pre_low	double	—
└pre_transactions	int64	—
└pre_open	double	—
└pre_timestamp	int64	—
└pre_volume	double	—
└pre_vwap	double	—
└symbol	string	—
└last_time	int64	Unix seconds.
└crypto_data	DataPricePredictionCryptoData	Present on the crypto branch.
└points	DataPricePredictionCryptoPoint[]	Bars for the primary symbol.
└high	double	—
└open	double	—
└low	double	—
└close	double	—
└timestamp	int64	Unix seconds.
└volume	int64	Legacy integer volume — kept for old clients only.
└volume_decimal	double	Preferred: float volume (needed for fractional assets like HL perps).
└infos	map<string, DataPricePredictionCryptoSeries>	Per-symbol series when symbol_list was provided.

Example request

POST https://api.alphio.ai/v1/openapi/data/price-prediction
{
  "ticker_type":"stock",
  "stock":{ "code":"AAPL", "multiplier":1, "timespan":4, "count":800, "locale":"US" }
}

Example response

{
  "ret": 0,
  "msg": "ok",
  "data": {
    "ticker_type": "stock",
    "stock_data": {
      "points": [
        {"code":"AAPL","close":195.5,"open":194.2,"high":196.1,"low":193.8,
         "timestamp":1717286400,"volume":48230000.0,"vwap":195.1,"otc":false}
      ],
      "last_time": 1717286400
    }
  }
}

POST/v1/openapi/data/screener/stock

DataScreenerStock

Stock screener. Request / response follow the upstream datafabric schema (chatfinbot.datafabric.v1.SelectStocksReq / SelectStocksRsp). Contact us for the full schema, or reuse a screenId returned by a Chat card.

POST/v1/openapi/data/screener/etf

DataScreenerETF

ETF screener. Uses chatfinbot.datafabric.v1.SelectEtfsReq / SelectEtfsRsp upstream.

POST/v1/openapi/data/screener/crypto

DataScreenerCrypto

Crypto screener. Uses chatfinbot.datafabric.v1.SelectCryptosReq / SelectCryptosRsp upstream.