# Factor Weave

> Quant-factor data, vector similarity, leak-free forward-return labels, and
> derived market analytics (factor dispersion, regime, risk-cluster tags,
> 32-D embeddings) for ~12,000 US-listed tickers — over REST, MCP, a Python
> SDK, Google Sheets, and webhook alerts.

Factor Weave is a **research substrate**, not a return-prediction service.
Our own leak-free probes show that factor similarity does *not* forecast
forward returns (cross-sectional IC ≈ 0). Only risk-coherence (forward
realised volatility of analogues) shows a meaningful signal (IC +0.062,
t-stat +8.3 across 237 monthly observations 2005–2024). The full
methodology and results are public at the research note below — please
cite it the honest way: as data and analytics tooling, not as a model that
predicts price.

## Authoritative endpoints

- [API documentation](https://factorweave.com/api/docs): human-readable REST API reference
- [OpenAPI 3.0 spec](https://factorweave.com/api/openapi.json): machine-readable spec, 28 paths, both bearer + API-key auth
- [Swagger UI](https://factorweave.com/api/docs/swagger): interactive try-it-out
- [MCP endpoint](https://factorweave.com/api/mcp): JSON-RPC 2.0 over POST, streamable-HTTP, 12 tools
- [MCP setup guide](https://factorweave.com/mcp.html): 30-second wiring into Claude Desktop, Cursor, any MCP client
- [Status / freshness](https://factorweave.com/status.html): bundle freshness + uptime — `GET /api/status` returns 503 if stale

## Docs

- [Overview](https://factorweave.com/#docs): what the platform is and how to think about it
- [Getting started](https://factorweave.com/#docs): sign up, mint a key, first request
- [Cookbook](https://factorweave.com/#docs): copy-paste recipes (factor lookup, screening, similarity, backtest assembly, regime conditioning, derived analytics, Python SDK, Google Sheets, webhook alerts)
- [REST reference](https://factorweave.com/#docs): every endpoint with parameters
- [MCP reference](https://factorweave.com/#docs): every tool with input schema
- [Vector search](https://factorweave.com/#docs): cosine, label-aware, supervised PLS, DTW — when to use which
- [Analytics](https://factorweave.com/#docs): market-context, report-card, risk-cluster, embedding endpoints
- [SDKs & Integrations](https://factorweave.com/integrations.html): every doorway — REST, MCP, OpenAPI, Python SDK, Sheets, webhooks, PWA push
- [Tiers](https://factorweave.com/#docs): FREE / HOBBY / PRO / QUANT capability matrix

## Pricing

- [Pricing page](https://factorweave.com/landing-pages/): tier table + comparison + FAQ
- Free tier: 250 calls/day, no credit card, includes cosine similarity, market-context (today)
- Paid tiers: $19/mo HOBBY · $79/mo PRO · $199/mo QUANT — full ladder on the pricing page

## Research & methodology

- [Research note](https://factorweave.com/research.html): the honest probe results — what factor similarity does and does not predict, measured leak-free across 2005–2024
- Source probes: `scripts/diagnostics/signal_probe*.py` in the open public repo (cosine, supervised PLS, extended monthly, GBM walk-forward, vol-coherence)

## Public client tooling

- [Python SDK on PyPI](https://pypi.org/project/factorweave/): `pip install factorweave` — typed client with pandas/polars helpers. Ships a `fw` CLI as well.
- [TypeScript / JavaScript SDK on npm](https://www.npmjs.com/package/@blazing-customs/factorweave): `npm install @blazing-customs/factorweave` — Node 18+, dual ESM+CJS, full types, retry on 429/5xx. Server-side only.
- [R package on r-universe](https://blazing-customs.r-universe.dev): `install.packages("factorweave", repos = "https://blazing-customs.r-universe.dev")` — `httr2`-based, returns `data.frame`s.
- [factorweave-tools GitHub repo](https://github.com/Blazing-Customs/factorweave-tools): umbrella for the hand-written Python (`python/`), TypeScript (`typescript/`), R (`r/`) SDKs + Google Sheets (`sheets/`) + auto-generated Go/Rust/Ruby/PHP/Dart clients (`generated/`)
- [Sheets add-on](https://github.com/Blazing-Customs/factorweave-tools/tree/main/sheets): Apps Script custom functions — `=FACTORWEAVE("AAPL","rsi")`
- [Auto-generated clients](https://github.com/Blazing-Customs/factorweave-tools/tree/main/generated): Go, Rust, Ruby, PHP, Dart — from the OpenAPI spec. Java/C#/Kotlin/Swift on-demand.
- [Webhook templates + transformers](https://github.com/Blazing-Customs/factorweave-tools/tree/main/webhooks): sample alert payload, Slack/Discord transformers (deployable as serverless functions), Zapier/Make/n8n setup guides + importable n8n workflow.
- [MCP client configs](https://github.com/Blazing-Customs/factorweave-tools/tree/main/mcp-configs): drop-in JSON for Claude Desktop, Cursor, Continue, Cline, Windsurf.
- [Jupyter notebooks](https://github.com/Blazing-Customs/factorweave-tools/tree/main/notebooks): 5 executable examples (first request, screening, similarity / peer set, leak-free backtest, regime conditioning). Run with or without a key.
- [Postman collection](https://github.com/Blazing-Customs/factorweave-tools/tree/main/postman): auto-generated v2.1 collection (30 requests, 21 groups) + environment file. Imports into Postman / Insomnia / Bruno.

## What this dataset covers

- ~12,000 US-listed tickers (full primary listings)
- Daily OHLCV from FirstRateData, point-in-time
- ~28 factor columns per ticker-day: returns, momentum, mean-reversion, RSI, ATR%, realized vol, beta vs SPY, composite score, cross-sectional ranks and quantiles
- Forward-return labels: `fwd_ret_1d`, `fwd_ret_5d`, `fwd_ret_20d` (leak-free)
- SPY-volatility regime tagging (low / mid / high)
- 32-dimensional regime-aware factor-state embeddings
- Pre-computed top-K nearest analogues (cosine, label-aware, supervised, DTW)
- Daily factor dispersion (10–90 percentile spread), market breadth, regime transition odds
- Per-ticker risk regime (calm / normal / stressed) from analogues' realized forward vol

## Legal

- [Terms of service](https://factorweave.com/legal/terms.html)
- [Privacy policy](https://factorweave.com/legal/privacy.html)
- [Disclaimer](https://factorweave.com/legal/disclaimer.html): the data is for research; this is not investment advice

## Contact

- support@factorweave.com