MCP-native · Chart pattern search · 25M+ patterns

Stock chart pattern search, built for AI agents.

Anchor any chart. Get back the 500 historical days that look like it. Bucket them by regime. That’s the edge — empirical analogs, not predictions.

25M+
indexed patterns
<300ms
cohort retrieval
50–0
vs ungrounded agents
9
canonical MCP tools
$0
free tier
A new way to query markets

Pattern search has always been mathematical.
We made it shape-based.

Old way · Indicator search

Search by formulas someone wrote.

  • Define a “rising wedge” with a math expression.
  • Scan the universe for charts matching the rule.
  • You query in indicator space. Same query for everyone.
  • Limited to patterns someone bothered to formalize.
TradingView · ThinkOrSwim · Finviz · Trendspider
Chart Library · Pattern search

Search by raw chart shape.

  • Take an actual chart — a real stock on a real date.
  • Get back the 500 historical days that look like it as raw shape.
  • Bucket those 500 by regime — vol, sector, news, earnings window.
  • Conditional distribution shows which conditions actually predicted what happened. No formulas. The shape is the query.
25M+ patterns · 19K symbols · 10 years
How a query flows

From anchor to actionable read in one MCP call.

01

Anchor

Your agent picks any (symbol, date, timeframe). NVDA · today · 1h. The chart you want to understand.

02

Match

The system finds the 500 historical days that look most like your anchor as raw price + volume shape. No indicators. No formulas.

03

Stratify

Bucket the 500 analogs by regime — vol, sector, news, earnings. The conditional distribution shows which conditions historically predicted what happened next.

How the agent talks to you

We don’t tell the agent what to say. We give it labels.

The MCP tool returns deterministic classification flags — verdict, edge, regime alignment, swing factors. Your agent reads them and writes the answer in its own voice. No scripted prose. Numbers stay in parentheses.

YOUWhat’s the read on NVDA here? (1h chart, May 11)
decision_brief returned·summary.* flags
verdict_class=coin_flipedge_class=trivialregime_alignment=neutralsample_quality=okconviction=low
swing_factors
factor=sector_laggingfactor=narrative_passivefactor=earnings_near
caveat_flags
caveat=soft_in_regime_samplecaveat=regime_was_derived
Agent reads the flags above and writes in its own voice ↓
CL

Honestly, this one reads like a coin flip. NVDA’s in-regime analogs printed about the same as everything else — there’s no real edge in either direction here.

Of the 132 closest historical analogs, 21 printed under a regime like today’s. Their 5-day median was +0.09% with a 52% hit rate. The other 111 returned +0.27% on the same hit rate — the gap is small enough that the regime isn’t doing much work here.

WHAT I’D WATCH
  • ·Sector lagging hard. XLK 60-day RS is −10.4 — that skews historical reads on tech setups lower until it turns.
  • ·News without conviction. Pulse is +0.29 with the tape active — looks like vol repricing, not a real narrative change.
  • ·Earnings in 16 sessions. Pattern reads tend to weaken this close to the print — keep that on the radar.

Bottom line: no meaningful edge, low conviction. The in-regime sample is soft (n=21) and the vol regime was derived — treat this as a lean, not a thesis.

THE NUMBERS BEHIND IT
In-regime sample
21
of 132 total
In-regime 5d median
+0.09%
win rate 52%
Other regimes
+0.27%
win rate 52% · n=111
Vol regime
High
realized 39.3
Sector RS (60d)
−10.4
XLK lagging
Days to earnings
16
catalyst near
Narrative change
0.29
low — vol-driven
Conviction
Low
thin in-regime n
POST /api/v1/decision_brief · NVDA · 2026-05-11 · 1h● 1.4sHistorical pattern data only — not financial advice.

Every read is grounded in the underlying tools your agent can also call directly: search, cohort, discover, analyze, context, narrative, explain, portfolio, and the new decision_brief orchestrator that bundles them.

Independent evaluation · n=50 · blind LLM judge

AI agents using Chart Library beat identical agents without it 50–0.

Two identical Claude Haiku agents. 50 out-of-sample trading scenarios. Blind LLM judge scored both on 6 dimensions of reasoning quality. The agent with Chart Library tools won every single scenario. Paired t- statistic greater than 10 on every dimension. Full methodology public.

For developers

Drop-in chart pattern intelligence for any AI agent.

MCP-native. Works with Claude Desktop, Cursor, Hermes Agent, or any MCP-capable client. Under a minute to install.

# Install
pip install chartlibrary-mcp
# Register with Hermes Agent
hermes mcp add chartlibrary --command "python -m chartlibrary_mcp"
# Also works with Claude Desktop, Cursor — see the docs
Free
$0
generous limits for experimentation
Builder
$29/mo
real agent workloads, faster rate limits
Scale
$99/mo
production deployments + priority
What's next

The next ten years of markets × AI runs on grounded retrieval.

Every AI agent that talks about stocks today is making up cohort statistics. The interfaces are getting smarter; the grounding hasn't caught up. Chart Library is the layer.

Now

MCP-native retrieval

9 canonical tools. 25M+ patterns. Calibrated forward distributions, regime stratification, Layer 5 memory that compounds with every query.

Soon

Live + multi-timeframe

Real-time intraday updates. One anchor, simultaneous matches across 5min / 15min / 1h / 1d.

Next

Cross-asset

Crypto first (active agent-trading market). Then futures, then international equities. Same architecture, broader universe.

Build an agent that knows what happened before.