Natural language in, real candidates out

Options Whale is not a chatbot bolted onto market data. It is a natural-language interface to deterministic research functions.

How the pipeline works

You
  -> intent parser
  -> structured research query
  -> deterministic filters + calculators
  -> real candidates
  -> grounded explanation

The language model's only job is to translate your request into a structured query and to explain results that were produced elsewhere. It never receives a market-data API, never runs a calculation and never chooses a strike.

Example prompts

Find conservative 30-45 DTE cash-secured puts Covered calls yielding at least 1% with earnings more than 21 days away I have $20,000 of capital. Find wheel candidates that use less than half Neutral iron condors with defined risk Show only ETFs with no earnings within 21 days

Where the data comes from

Reviewed, published research with its own data timestamp and provenance.
Cash-secured-put and covered-call premiums modelled from published at-the-money implied volatility with Black-Scholes, clearly labelled as modelled.
Greeks and expected moves from the same calculation library the calculators use.

What is delayed and what is not modelled

Data is end-of-day and delayed. Modelled premiums are not executable quotes; real fills will differ from the mid. Earnings dates come from the verified calendar. Nothing here is advice, and no returns are promised.

Guardrails

If a value is unavailable, the answer says so instead of estimating it.
Model prose is rejected if it names a ticker that the tools did not return.
Undefined-risk structures carry an explicit warning.

Start researching →