What is Chart Library?

Chart Library is a chart-pattern intelligence engine for humans and AI agents. Anchor any (symbol, date, timeframe) and it returns the cohort of historical analogs — from 25M+ indexed patterns across 19,000+ symbols and 10 years — what those analogs did next, and a calibrated forward-return distribution, over web, REST API, and MCP.

Does Chart Library predict stock prices?

No. Chart Library never forecasts. It returns historical distributions — what actually happened after setups like yours (median, p10/p90 band, up rate) — plus the receipts to audit them. Direction and decisions stay with you or your agent.

How calibrated are Chart Library's bands?

The nominal 80% forward-return band held 80.8% across 300,000+ live, audited cases under symbol-disjoint evaluation. The coverage record is public and recomputed continuously at chartlibrary.io/calibration.

Market déjà vu · 2026-08-11

DDOG today looks like T did in Mar 2019

DDOG's chart over the last 30 trading days is a 55% shape match for T in Mar 2019 — its closest historical analog across a decade of market data. In the 15 trading days after that point, T went +4%.

Shape match

55%

T, next 15d

+4%

Analog

T · Mar 2019

Read this honestly

The forward path is T's actual history after that date — not a prediction for DDOG. A close shape match is a starting point for research, not a signal. Different stocks with similar charts can do very different things next.

How we find the twin

We embed DDOG's recent daily chart into a 256-dimension shape vector and search millions of historical chart windows for the nearest match by shape, then re-score the finalists on the exact window shown so the lines actually overlap. The match is amplitude-aware, so it's a real twin — not just "both went up."

Go deeper

See DDOG's full cohort of historical analogs — the distribution of what they all did next, with calibrated confidence:

DDOG cohort →Open in Chart Library