Forecasting any question

Marina answers forecasting questions with a probability or a number, backed by cited and checked evidence and several models, in about 20–60 seconds for $0.03–0.10.

bun run forecast "Will the Fed cut rates at the FOMC meeting ending October 28, 2026?"
bun run forecast "What will the US regular gasoline average be on Oct 15 2026?" --unit '$/gal'
bun run forecast "…" --kind number --by 2026-12-31 --json

In the world, any entity can ask: forecast <question> (alias predict). Programs can call POST /v1/forecast with {"question": "...", "kind"?: "probability" | "number", "resolveBy"?, "unit"?} behind the model API’s auth; the reply is the full answer, below.

What happens

1. Research — a search-grounded model (OpenRouter’s web search) gathers dated, sourced facts: the current state, base rates, scheduled events, and what forecasters and markets expect.
2. Citation check — each fact’s figures are looked up in the page it cites; lines are tagged [verified], [unverified] or [unreachable]. Publishers whose terms bar bots are never fetched.
3. Analysts — one model per vendor (by default DeepSeek V4 Pro, Claude Sonnet 5, GPT-6 Luna) answers from the tagged dossier.
4. Judge — Jev scores how well each analyst’s reasoning is supported by the verified facts; weakly supported answers count for little.
5. Aggregate — probabilities are combined in log-odds, numbers as a weighted mean whose spread includes the analysts’ disagreement.

The answer carries everything needed to audit it: each analyst’s answer and reasoning, its grounding score and weight, the sources, how many facts verified, cost and time — and a caveat when the evidence was thin.

Configuration

Variable	Default	Meaning
OPENROUTER_API_KEY	required	retrieval and the Jev judge run through OpenRouter
MARINA_FORECAST_ANALYSTS	three vendors via OpenRouter	comma-separated provider/model ids
MARINA_FORECAST_RETRIEVER	openrouter-web:openai/gpt-6-luna	the research model
MARINA_FORECAST_JUDGE	jev	jev, decisions (the configured MARINA_DECISIONS backend — OpenJev, TypeSafe, a chat classifier; falls back to jev when none is set) or none (equal weights)

How good is it?

The same pipeline, pointed at the Social Simulation Arena’s questions, is measured in docs/guides/arena.md: structured data (the Civiqs nowcast) beat every leaderboard entry on backtests; the value of the web-research agents is being measured forward, on questions that had not resolved when they answered. Treat a single forecast as a well-sourced opinion, not a guarantee.