aipricepatterns
sott095/aipricepatternsAI Price Patterns provides historical market memory for AI agents through semantic vector search across years of financial market data. Use it to discover similar market regimes, analyze historical price patterns, retrieve market context, and accelerate quantitative research. Designed for algorithmic trading, quantitative finance, reinforcement learning, and autonomous AI agents. Features: • Semantic vector search • Historical market memory • Pattern similarity discovery • Market regime analysis • Quantitative research • Remote HTTP MCP • Works with Claude, Cursor, ChatGPT, Codex, VS Code and other MCP-compatible clients.
Score: 100/100
Handshake verified by our own probe.
Tools · 20
Search for historical patterns similar to a custom 'sketched' price trajectory (Sketch-to-Search). Useful when you want to find matches for a hypothetical or hand-drawn pattern.
Find dated historical situations comparable to the current BTC, ETH or SOL window and see what followed. Returns chart-parity OHLC, rankingMethod and fallback reason, historical terminal-close quantil…
Start here if you do not know which market to look at. No arguments needed: scans BTC, ETH and SOL across 15m/1h/4h (configurable) and ranks where historical twins agree most on move magnitude right n…
One call for the whole candle-memory picture of a BTC, ETH or SOL window. Quote brief before answering about that market. brief is the measured close, path lines, low and high spread, closest day, whe…
Walk forward the exact chart candle configuration (symbol, interval, q, f) over independent resolved anchors. Compares analogue and volatility-scaled range bands against the unconditional baseline wit…
Return a conservative chart-parity trader decision. V2 combines the candle snapshot contract with freshness and execution-component gates; it returns NO_TRADE while candle direction is unvalidated or …
Open one neighbour from the reported rankingMethod as an episode: query OHLC, twin OHLC, rebased forward, distance, display similarity, dated permalinks. Not a directional call.
Explore historical paths rebased to the query close: terminal-close P10/P50/P90, observed path extremes, turn clusters and which quantile threshold was reached first. forwardSummary counts how often t…
Before the next order. Required: the trader's levelPrice, invalidationPrice and targetPrice (aliases level, stop, target). Do not invent a missing price from P10, P90 or any band. One search, then whi…
Investigate a caller-named price level using the chart's historical analogues. levelPrice is required. Do not omit it. The tool does not fill a neighbour P90, P10, or any other level. Defaults for the…
Replay one already stated price hypothesis as round trips. Entry is the level, and only after a candle opens on the near side and trades through it. A candle that opens through the level is an unfille…
Autonomous researcher and standing-watch loop in one call. Scans BTC, ETH and SOL, inspects the tightest neighbour bands, and checks walk-forward range lift. It replays a price order only when the cal…
Test whether a candle window behaves differently under UTC sessions (Asia, London, US overlap / CPI / NY open, New York), a UTC event hour (for example 14:00 UTC FOMC), weekday versus weekend, or a pr…
Why this twin ranked under the reported method: distance, anatomy-token observations, and outcome vs neighbour p10/p50/p90. Shape tokens alone cannot explain a context/volume score. Similarity is not …
Explain what stands out in the retrieved evidence: terminal-outcome band versus all-history, observed coverage, closest episode versus median, era clustering and neighbour distances. Suggests a concre…
Autonomous agent feedback channel: submit feature requests, needed symbols (e.g. ETH, SOL, altcoins), smaller timeframes (e.g. 1m, 1s, tick), orderbook depth, or execution parameters directly to the e…
Start here. Returns the candle market-memory guide: what the evidence supports, the chart presets, the recommended call order, how to read bands and track records, and how access works. Free; explains…
Free. Which symbol/timeframe configurations carry range information: for the /candles default preset (q=5, f=10) a walk-forward record of the analogue band and the volatility-scaled band against the a…
Return the versioned MCP compatibility manifest, including canonical tools, aliases, and JSON argument schemas for remote clients. Defaults to the listed candle tools; profile='full' returns every cal…
Free during the trial. Returns the official Docker image (ghcr.io/sergio12s/rlxbt-server), step-by-step install instructions, and a shared Pro trial license key for self-hosting the RLXBT backtesting …
Resources · 2
What the candle tools return, the chart presets, the recommended call order, and what the evidence does not support.
market://guides/start
Versioned contract manifest for the listed tools, with aliases and argument schemas.
market://mcp/manifest
Prompts · 3
Read the current candle window against market memory: dated analogues, what is unusual, one episode, and whether the band beats baseline.
Walk-forward audit of which chart q/f presets give an analogue band that beats the unconditional baseline for one symbol and interval.
Act as an autonomous market researcher on standing watch. Run the one-call watch cycle and notify only when the standing-watch bar is crossed. The watch does not invent an order.
How to use
Claude Desktop or claude.ai: add a custom connector (Customize → Connectors; on Team and Enterprise plans an owner adds it under Organization settings → Connectors) with this server URL: https://aipricepatterns.com/api/mcp
Cursor or Cline: add to your MCP config:
{
"mcpServers": {
"aipricepatterns": {
"url": "https://aipricepatterns.com/api/mcp",
"transport": "http"
}
}
}