Forecite[Use cases]Jun '26
Built for quant desks: signal your backtest can trust
Replayable history, raw sub-scores, and per-event latency telemetry — how systematic teams wire the Verdict Engine into research and production.
Priya Nair Quant Research
[Use cases]May '26
Built for discretionary traders: conviction without the noise
A live feed that's already been judged, watchlists that remember your lens, and an analyst on call — Forecite for the trader who decides in minutes, not milliseconds.
Elena Marsh Product Marketing
[Use cases]May '26
Built for AI agents: a feed your agent can act on
Structured verdicts, an MCP server, and a scoring endpoint for anything the agent finds on its own — market intelligence designed for software that reads.
Sam Whitaker Developer Advocate
Latest articles
Agents: orchestrate market intelligence, on every deskAgents are a new way to put the Verdict Engine to work alongside your team. Wired to the live feed, so every agent acts on scored, sourced intelligence.Jun '26[Product]Built for quant desks: signal your backtest can trustReplayable history, raw sub-scores, and per-event latency telemetry — how systematic teams wire the Verdict Engine into research and production.Jun '26[Use cases]Built for discretionary traders: conviction without the noiseA live feed that's already been judged, watchlists that remember your lens, and an analyst on call — Forecite for the trader who decides in minutes, not milliseconds.May '26[Use cases]Built for AI agents: a feed your agent can act onStructured verdicts, an MCP server, and a scoring endpoint for anything the agent finds on its own — market intelligence designed for software that reads.May '26[Use cases]Feeds API v1: keyset pagination and lookback windowsThe historical feeds API is now generally available — cursor-based pagination, plan-scoped lookback, and tag filters that compose.May '26[Product]Work anywhere with ForeciteOrchestrate your market intelligence with direct connections in Claude, ChatGPT, and Cursor.Apr '26[Product]How we run weekly retrains without breaking FridaysInside the Verdict Engine's training loop: labels from realized returns, rolling-window retrains, and the evaluation harness that keeps regressions out of production.Mar '26[Inside Forecite]Verifying webhook signatures, in five languagesEvery Forecite webhook delivery is signed with HMAC-SHA256. Here's how to verify X-Forecite-Signature in Python, TypeScript, Go, Rust, and Ruby.Mar '26[Guides]Introducing WarrenWarren is a new way to work with the market. Powered by the Verdict Engine, it deeply understands every headline in the feed — and can score any document you hand it.Feb '26[Product]Latency is a product decisionSub-50ms isn't an engineering vanity metric. Where the milliseconds go — and why we publish per-event telemetry instead of an average.Jan '26[Perspectives]Filtering the feed: tags, tickers, and watchlistsA practical tour of Forecite's filter model — four tag dimensions, symbol lists, and how watchlists layer URL overrides on top of saved filters.Dec '25[Guides]Forecite raises $20M Series AWe're building the intelligence layer for the markets — and scaling the Verdict Engine to judge every headline, everywhere.Nov '25[Company]Sentiment is not a directionWhy the Verdict Engine scores actionability before sentiment — and why 'positive' headlines lose money.Oct '25[Perspectives]From zero to first verdict in sixty secondsInstall the SDK, paste your key, subscribe to the live feed. A quickstart for Python, TypeScript, and the CLI.Sep '25[Guides]Designing Forecite: the blueprint aestheticDashed guides, hatched gutters, register crosses — why the interface for a trading platform borrows its language from drafting sheets.Sep '25[Inside Forecite]