Chart Library vs Financial Modeling Prep — the fundamentals versus what the setup did next.
Not competitors. Financial Modeling Prep (FMP) is a developer-friendly financial-data API — income statements, balance sheets, ratios, valuations, prices, and calendars. If you’re building DCFs, screeners, or fundamentals dashboards, it’s a strong, affordable source of the raw financial record.
Chart Library answers a different question, and a behavioral one rather than a fundamental one: “given this exact price/volume setup, what did its closest historical analogs do over the next 1/5/10 days, with what confidence?” It returns a calibrated outcome distribution and an audit receipt, built for an AI agent to call — not statements to model.
Fundamentals data vs calibrated behavioral analogs
- FMP = the financial record. What a company is — statements, ratios, valuations, prices — as queryable data.
- Chart Library = the behavioral precedent. What setups like this one did next, as a calibrated forward-return distribution with the drivers that separated past winners from losers.
- Orthogonal, not overlapping. FMP won’t tell you the historical odds after a shape; Chart Library won’t serve you a balance sheet. A serious agent uses both — fundamentals for context, calibrated analogs for the outcome read.
- The calibration is the moat. Nominal 80% band held 80.8% across 300K+ audited cases, PIT-flat — a trust receipt a raw data API has no equivalent for.
Frequently asked questions
- Is Chart Library a Financial Modeling Prep alternative?
- No — different layer and different question. FMP serves fundamentals and prices; Chart Library returns calibrated historical-analog outcome distributions for a setup. Use FMP (or any data API) for inputs and Chart Library for the behavioral base rates.
- Does Chart Library have fundamentals, statements, or ratios?
- No. It's a retrieval-and-calibration engine over historical price/volume analogs, designed to be called by agents. For fundamentals, pair it with FMP; Chart Library supplies the calibrated outcome read.
- Does it forecast returns?
- Never. It surfaces what historical analogs did next as a calibrated distribution with a coverage receipt — similarity-only, no directional forecast.
Run a cohort_analyze call.
Free Sandbox tier — 1,000 calls/day, no authentication. MCP install for Claude or Cursor takes 30 seconds.