Chart Library vs Finviz — the screen versus what the screen's hits did next.
Straight answer: if you want to filter the market by P/E, market cap, float, RSI, or a heatmap glance, use Finviz. It is the best free screener for a human, and Chart Library is not trying to replace it. They sit in different categories and answer different questions.
Finviz answers “which stocks match these filters?” — for a person reading a table. Chart Library answers “given this exact setup, what did its closest historical analogs actually do over the next 1/5/10 days, and how confident can I be?” — and serves it to an AI agent over API/MCP, as a calibrated distribution with an audit receipt, never a forecast.
Two different jobs
- Finviz = discovery for humans. Rule-based filters and a visual heatmap to surface a watchlist. The output is a list of tickers a person scans.
- Chart Library = calibrated base rates for agents. Anchor a
(symbol, date, timeframe)and get the cohort of historical analogs by raw chart shape, their forward-return distribution, and a calibrated band. The output is a structured, auditable answer an agent reasons on. - Screens are rules; analogs are shapes. A screener can only find what someone formalized into a filter. Chart Library matches the actual price/volume shape — no rule to write — and tells you what that shape’s history did.
- Humans vs programs. Finviz is a website you look at. Chart Library is an endpoint your code calls (MCP, REST, SDKs) — built for the agent in the loop, not the eyeball.
They compose: screen, then get the base rates
The natural pipeline uses both. Finviz (or any screener) narrows the universe to candidates; Chart Library then tells your agent what each candidate’s setup historically led to:
- Screen → a shortlist of names that match your thesis.
- For each, call
pull_comps/cohort_analyze→ the calibrated outcome distribution + coverage receipt for that exact setup. - Your agent ranks/sizes/explains using honest base rates, not a screener row and a hunch.
And the trust layer is the part a screener structurally can’t offer: the nominal 80% band held 80.8% across 300K+ audited cases, PIT-flat to ~0.5pp. Descriptive history, never a directional call.
Frequently asked questions
- Is Chart Library a Finviz alternative?
- Not really — it's a different layer. Finviz screens the market for a human; Chart Library gives an agent the calibrated historical base rates for a specific setup over API/MCP. Many stacks use a screener for discovery and Chart Library for the outcome read.
- Does it have a screener / heatmap?
- No. Chart Library is a retrieval-and-calibration engine for analogs and outcome distributions, designed to be called by agents and code — not a visual screening UI. If you need filtering, screen first, then anchor the survivors here.
- Does it predict which stocks will go up?
- Never. It surfaces what historical analogs of a setup did next, as a calibrated distribution with a coverage receipt. Similarity-only by design; 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.