Ora
Financial markets decision model · ora-fin

A decision model for financial markets.

Send a symbol or your position, and the questions you decide on. ora-fin brings the market data, works out the answers and the odds, and returns one typed answer per question in milliseconds. Rules on those answers fire your alerts and webhooks.

Request · POST /v1/answerstate {"symbol":"AAPL"}above_200d Is AAPL above its 200-day moving average?rsi_14 =rsi(C, 14)close =C
Response · 200 · data as of 2026-11-05
above_200dyes_no0.9463
rsi_14value53.01
closevalue344.17

Just a symbol and three questions. Ora supplies the market data and returns one typed answer per question.

Request · POST /v1/answerstate {"symbol":"MSFT","entry":410,"stop":395,"shares":50}pnl =(C - entry) * sharesstop_hit =C < stoptrend choice: up / down / sidewaysoutput your JSON template
Response · 200 · data as of 2026-11-05
{
  "ticker": "MSFT",
  "session": "2026-11-05",
  "position": {
    "entry": 410,
    "stop": 395,
    "pnl": 6296.5,
    "risk_to_stop": 7046.5
  },
  "signal": {
    "trend": "up",
    "trend_confidence": 0.702,
    "rsi_band": 1,
    "exit_now": false
  },
  "note": "MSFT P&L 6296.5 with the stop at 395"
}

Your entry, stop and size in; back comes exactly the JSON your app stores, built from typed answers.

Request · POST /v1/answerstate {"symbol":"NVDA"}below_50d =C < sma(C, 50)oversold =rsi(C, 14) < 30when below_50d >= 0.6 → webhook
Response · 200 · data as of 2026-11-05
below_50dyes_no0.9463
oversoldyes_no0.7485
fired pullback_hook webhook · dry run
{
  "symbol": "NVDA",
  "session": "2026-11-05",
  "event": "pullback",
  "below_50d": 0.9463,
  "oversold": 0.7485
}

A rule on the answers decides. Here it fired, as a dry run: the response shows exactly what would be sent, and nothing was.

Request · POST /v1/answersymbols AAPL, MSFT, NVDA, AMZN, GOOGLabove_50d =C > sma(C, 50)above_200d =C > sma(C, 200)rsi_14 =rsi(C, 14)schedule weekdays, after the close → digest webhook
Response · 200 · data as of 2026-11-05
Symbolabove_50dabove_200drsi_14
AAPL0.890.9553.0
MSFT0.850.9554.5
NVDA0.050.0725.4
AMZN0.830.9455.0
GOOGL0.050.0628.0

The same questions over a watchlist on a schedule: one digest per run to your webhook.

One modelora-fin answers each request the right way: a symbol or numbers get market answers, text gets text answers
3 answer typesyes_no from 0 to 1, choice with a probability per option, score on your ordered levels
60,000+ symbolsStocks and ETFs worldwide, crypto, FX and indexes: send just the symbol
Rules → actionsSlack and signed HTTPS webhooks, with dry-run previews; email coming soon
WorkflowsThe same questions over a watchlist on a schedule, one digest per run
8 languagescURL, Python, JavaScript, PHP, Go, Java, C# and Ruby from any request

What Ora helps with

  • Check a stock before you trade: trend, overbought, volume and days to earnings, answered in one request.
  • Apply your rules the same way every time, with the rule and data date on every answer.
  • Scan a watchlist each morning: one line per name.
  • When a rule is true, send a Slack message or a webhook to your own systems.
  • Gate an order first: block, reduce or allow.
  • Watch risk limits: drawdown, stops, volatility spikes and upcoming earnings.
  • Pull price history with one formula: =series(C, "2026-10-05", "2026-11-02").
  • See forecast odds with a plain label when they carry little edge.

Models

One model to call, ora-fin, on one endpoint, POST /v1/answer. Ask about a stock or ETF, or send your own bars or a piece of text, and get one answer per question. Every release is a stable version you can pin; new updates are evaluated in a preview first.

Financial rules and forecasts

ora-fin

Your state. Your questions. Our market data.

Putting market answers into software usually means buying a data feed, cleaning it, coding the calculations and keeping all of it running. With ora-fin, it's one request.

Tell it your situation and what you want to know. We supply the market data, run the calculations and return one named answer per question, ready for your code to use.

You send your state

A symbol and your own numbers, such as an entry price, a stop or a loss limit. No market data to send.

You define the questions

Name each one and write it your way, from a simple condition to a formula built on other formulas.

Your code gets the answers

JSON under the names you chose, with the rule and the data date. Use the numbers however you like.

Where it stands today

What you needTodayStatus
Send your own stateNumbers and formulas, such as an entry price or a stop, and lists of positions, up to 50 symbols in one request.Live
No market data to sendWe supply it, for more than 60,000 symbols.Live
Complex calculationsBeta, z-scores, volatility and formulas built on other formulas.Live
Your own questions as formulasName them and write them your way.Live
Plain English about your own state“Is the close above my stop?” reads as C > stop and shows you the rule. If your state has no such field, the answer says so.Live
Your choice of answer typeNumbers, yes/no scores, plain true or false, and choice and score where each option is a rule. Options in plain English are in development.Partly live
Your own output JSONYou name every answer, and an output template returns your exact shape.Live
Answers your code can useNumbers come back exact. A condition returns a score from 0.05 to 0.95, or a plain true or false if you ask for one, and null with a reason when it can't be known.Live

What happens to each question

Name a symbol and ask your questions. One call returns one answer per question. ora-fin does three things, and every answer tells you which one it did.

1. Calculates

Indicators, levels and conditions, worked out exactly from daily bars. Same rule, same data, same answer, with the rule and the data date on the answer.

  • =C
  • =rsi(C, 14)
  • =C > sma(C, 200) a score from 0 to 1
  • =iif(C > sma(C, 200), 1, 0) a hard yes or no
  • =series(C, -21) a table of values

2. Reads common questions

Ask in plain English and it turns the question into a rule, then shows you the rule. “Is the stock above its 200-day moving average?” gives C > sma(C, 200).

It reads your own state too: with a stop in state, “Is the close above my stop?” gives C > stop. It checks that the rule accounts for the conditions and numbers you state, and declines with a reason if it doesn't. Read the rule it shows: that is what ran.

3. Forecasts odds

The probability of a future market event, such as “Will it make a new 20-day high in the next week?”, based on how similar market conditions played out historically.

New highs, new lows, volatility and large drops are where it holds up best. Gap-fill questions need the details of the gap in the request. Direction (“higher in a week”) has little edge and is labelled, and two opposite direction questions aren't guaranteed to add up to one.

What it doesn't do yet

Decision questions such as “Should I trim this position before earnings?” are declined: it doesn't choose for you. You can combine rules into your own logic, and act on the answers with alerts or webhooks. It isn't investment advice, and answers use final daily bars, not live prices.

ora-fin · one callState and questions in, typed answers out
You send one request: a state (JSON such as a symbol, your own bars or numbers, or plain text) and named questions. ora-fin answers every question in the same call; each answer says which path gave it. Optional actions run on the answers.
state{"symbol": "AAPL"}JSON: a symbol, your bars, = formulas, numbers
stateor texta message, a document, pasted data
questionsabove_200d overbought sma50named and typed: yes_no, choice, score
POST /v1/answerOraora-fin: every question in one call
answersabove_200d 0.946one per question, with its basis
actionsoptionalSlack or a signed webhook when a rule is true
POST /v1/systemone is the same endpoint under its older name. Keys look like ora_sk_…; examples read them from ORA_API_KEY.
Send

state as JSON (symbol, bars, formulas, context) or text, plus yes_no questions (choice and score for text) to POST /v1/answer

Get back

probabilities, exact rule answers, picks and scores, each with its basis and the version that answered

Latency

Model answer about 2 to 60 ms; a round trip through the API is typically under 200 ms; text state about 0.2 to 1.5 s on CPU

Price

$0.042 per million input tokens, output free
The state plus every question, plus the fixed per-request overhead.
Market data: 1,000 input tokens per symbol Ora loads for you; free if you send your own prices. · Pricing

VersionChannelReleasedWhat changed
ora-fin-0.1.6
alias ora-fin
Stable2026-11-04The stable financial model: send your state and your questions, and it supplies the market data. English questions keep every condition and number they state or are declined, and every answer says whether it is an exact rule answer, a forecast or a text answer.
ora-fin-0.1.5 Stable2026-11-04The stable financial model: send your state and your questions, and it supplies the market data. English questions keep every condition and number they state or are declined, and every answer says whether it is an exact rule answer, a forecast or a text answer.
ora-fin-0.1.4 Stable2026-11-04The stable financial model: send your state and your questions, and it supplies the market data. English questions keep every condition and number they state or are declined, and every answer says whether it is an exact rule answer, a forecast or a text answer.
ora-fin-0.1.3 Stable2026-11-03The stable financial model: send your state and your questions, and it supplies the market data. English questions can now use your own state fields, and every answer says whether it is an exact rule answer, a forecast or a text answer. Questions it cannot answer are declined.
ora-fin-0.1.2 Stable2026-11-01The stable financial model, now also the text model: send state as JSON (a symbol, your own bars, formulas, numbers) or as text, and every answer says whether it is an exact rule answer, a forecast or a text answer. Questions it cannot answer are declined.
ora-fin-0.1.1 Stable2026-10-29The stable financial model: forecasts for daily-bar and opening-gap questions, the rule language, and ask-by-symbol context for any symbol with daily prices.

Stable, preview and aliases

Stable

ora-fin-0.1.6

A released version never changes behaviour. Pin it in production and log the version each response returns.

Preview

<family>-X.Y.Z-preview

New updates land in a preview first. Previews are internal while they are evaluated; one is released to everyone, under the stable name it was given, once it passes.

Aliases

ora-fin

Always the latest stable of a family: use one to move with each release.

Ask about a stock, act on it, or judge text

Real requests from the examples library, with what came back. Action examples are dry runs: they show exactly what would be sent, and nothing was.

Financial · send a symbol

AAPL by symbol, with a dry-run alert

Ask by symbol and act on it: rule questions, and exactly one of two messages. The message shows the session the server used.

Request

{
  "model": "ora-fin-0.1.1",
  "state": {
    "symbol": "AAPL"
  },
  "questions": {
    "above_200d": "=C > sma(C, 200)",
    "overbought": "=rsi(C, 14) > 70",
    "volume_spike": "=V > 2 * avg(V, 20)"
  },
  "actions": [
    {
      "id": "stretched",
      "when": "above_200d >= 0.6 and overbought >= 0.6",
      "do": "slack",
      "connection": "trading-desk",
      "message": "{{symbol}} ({{as_of}}): above its 200-day average and overbought on RSI."
    },
    {
      "id": "not_stretched",
      "when": "else",
      "do": "slack",
      "connection": "trading-desk",
      "message": "{{symbol}} ({{as_of}}): not both above its 200-day average and overbought on RSI. No alert."
    }
  ],
  "dry_run": true
}

Answers and what it would send

above_200d 0.946 · overbought 0.054 · volume_spike 0.084

Actions 1 of 2 fired · dry run, nothing sent
stretchedrule false: skipped
not_stretchedwould send (dry run)slack → trading-desk
{
  "text": "AAPL (2026-11-02): not both above its 200-day average and overbought on RSI. No alert."
}
Slack alert

Pullback alert to the trading desk (Slack)

Post to the desk's Slack channel when a stock in an uptrend has pulled back under its 20-day average; otherwise say there is no setup. Rule questions, with a threshold on how clearly each one holds. Needs a Slack connection named trading-desk.

Request

{
  "model": "ora-fin-0.1.1",
  "state": {
    "symbol": "MSFT"
  },
  "questions": {
    "above_200d": "=C > sma(C, 200)",
    "pullback": "=C < sma(C, 20)"
  },
  "actions": [
    {
      "id": "pullback_alert",
      "when": "above_200d >= 0.6 and pullback >= 0.6",
      "do": "slack",
      "connection": "trading-desk",
      "message": "{{symbol}} ({{as_of}}): above its 200-day average and back under its 20-day. Pullback setup."
    },
    {
      "id": "no_setup",
      "when": "else",
      "do": "slack",
      "connection": "trading-desk",
      "message": "{{symbol}} ({{as_of}}): no pullback setup today."
    }
  ],
  "dry_run": true
}

Answers and what it would send

above_200d 0.946 · pullback 0.894

Actions 1 of 2 fired · dry run, nothing sent
pullback_alertwould send (dry run)slack → trading-desk
{
  "text": "MSFT (2026-11-02): above its 200-day average and back under its 20-day. Pullback setup."
}
no_setuprule false: skipped
Digest workflow

Pre-open market read

A workflow runs the same questions for every symbol on a watchlist (Weekdays 08:30 ET) and sends one digest per run. Before the open, one Slack message for every index ETF: above its 50-day (a rule), a new 20-day high within a week and choppier (labelled forecasts). No per-symbol actions: the digest is the product.

Workflow

{
  "name": "Pre-open market read",
  "symbols": "SPY QQQ IWM DIA",
  "days": [
    1,
    2,
    3,
    4,
    5
  ],
  "at_time": "08:30",
  "model": "ora-fin-0.1.6",
  "questions": {
    "above_50d": "=C > sma(C, 50)",
    "new_high": {
      "type": "yes_no",
      "instructions": "Will it make a new 20-day high in the next week?"
    },
    "vol_up": {
      "type": "yes_no",
      "instructions": "Will volatility rise over the next 5 days?"
    }
  },
  "digest": {
    "do": "slack",
    "connection": "trading-desk",
    "when": "always"
  }
}

Per-symbol answers and the digest it would send

SymbolAnswers (as of 2026-11-02)
SPY

above_50d 0.086 · new_high 0.068 · vol_up 0.562

QQQ

above_50d 0.059 · new_high 0.024 · vol_up 0.496

IWM

above_50d 0.220 · new_high 0.263 · vol_up 0.447

DIA

above_50d 0.396 · new_high 0.252 · vol_up 0.527

Actions 1 of 1 fired · dry run, nothing sent
digestwould send (dry run)slack → trading-desk
{
  "text": "Pre-open market read · data as of 2026-11-02 · 0 of 4 symbols fired:\n• SPY: above_50d 9%, new_high 7%, vol_up 56%\n• QQQ: above_50d 6%, new_high 2%, vol_up 50%\n• IWM: above_50d 22%, new_high 26%, vol_up 45%\n• DIA: above_50d 40%, new_high 25%, vol_up 53%"
}
Text · choice and score

Sentiment (choice)

Request

{
  "model": "ora-fin-0.1.6",
  "state": "Shares jumped after the company beat estimates and raised its full-year guidance.",
  "questions": {
    "sentiment": {
      "type": "choice",
      "instructions": "Overall sentiment of the message",
      "criteria": {
        "positive": "Happy, pleased, enthusiastic",
        "neutral": "Factual, no strong feeling",
        "negative": "Unhappy, annoyed, disappointed"
      }
    }
  }
}

Response

{
  "model": "ora-fin-0.1.6",
  "answers": {
    "sentiment": {
      "type": "choice",
      "choice": "positive",
      "confidence": 0.903,
      "probabilities": {
        "positive": 0.935,
        "neutral": 0.038,
        "negative": 0.027
      }
    }
  },
  "usage": {
    "input_tokens": 332,
    "output_tokens": 7
  },
  "version": "ora-fin-0.1.1"
}
Text · choice and score

Support ticket triage

Request

{
  "model": "ora-fin-0.1.6",
  "state": "Our API integration started returning 500 errors on every request about 20 minutes ago, and we can't process any customer orders until this is fixed.",
  "questions": {
    "department": {
      "type": "choice",
      "instructions": "Which team should handle this",
      "criteria": {
        "billing": "Payment or subscription issues",
        "technical": "Bugs, errors or integration problems",
        "sales": "Pricing or account questions"
      }
    },
    "frustration": {
      "type": "score",
      "instructions": "How frustrated the customer appears",
      "criteria": [
        "Calm, just stating facts",
        "Frustrated but civil",
        "Very angry, strong language"
      ]
    },
    "is_urgent": {
      "type": "yes_no",
      "instructions": "The message conveys urgency: orders blocked, errors on every request, needs fixing now"
    }
  }
}

Response

{
  "model": "ora-fin-0.1.6",
  "answers": {
    "department": {
      "type": "choice",
      "choice": "technical",
      "confidence": 1,
      "probabilities": {
        "billing": 0,
        "technical": 1,
        "sales": 0
      }
    },
    "frustration": {
      "type": "score",
      "score": 0.631,
      "confidence": 0.362,
      "legend": [
        "Calm, just stating facts",
        "Frustrated but civil",
        "Very angry, strong language"
      ],
      "probabilities": [
        0.397,
        0.575,
        0.028
      ]
    },
    "is_urgent": {
      "type": "yes_no",
      "yes_no": 0.92
    }
  },
  "usage": {
    "input_tokens": 437,
    "output_tokens": 18
  },
  "version": "ora-fin-0.1.1"
}
All examples

Pricing

$0.042 per million input tokens. Output tokens are free.

Billed per answered request from prepaid credits: the request's input tokens plus a fixed 250 per request, shown in every response as usage.input_tokens. Rejected requests (4xx) are not charged. Market data: 1,000 input tokens per symbol Ora loads for you; free if you send your own prices.

Pricing details

Ask your first question

Read the docs now; run requests in the Playground once you have an account.