Market intelligence platformCrypto · Power · Agents

Market memory for every decision.

AIPP finds the historical situations that look like today across crypto timeframes, shows what actually happened next with dated evidence you can replay, and turns any MCP client into a real-time market researcher that watches, tests hypotheses, and alerts you only when history looks unusual.

Explore the chart Spawn a research agent

Free during research preview · no card, no account

BTC · BTCUSDT · 4h
—
loading market memory…
Closest match · date · what followed Search years of history →

The platform

The Dual Intelligence Engine.

Two markets, one method: retrieve the closest historical analogues, measure what followed, and expose all of it to people and to AI agents.

01 / CRYPTO

Crypto Market Memory

Vector search over years of candles. Find the windows that looked like now and see how each one resolved.

  • HNSW vector index in Rust
  • Multi-timeframe pattern matching
  • Historical replay with dated proofs
Open /candles
02 / RESEARCH ENTITY New

Autonomous Market Researcher (MCP)

Turn any MCP client into a real-time analyst that patrols BTC, ETH and SOL, formulates and replays competing hypotheses, tests UTC session & event contexts, and DMs you only when history looks unusual.

  • 1-call autonomous patrol (run_candle_research_watch) with watchId memory
  • Competing upside P90 / downside P10 hypothesis replay & session conditioning
  • Works in Claude, Cursor, ChatGPT, Codex + RLXBT Docker trial
Spawn on /mcp

Interactive MCP Tool Topology · Live JSON-RPC 2.0 Sandbox

Click any MCP node to execute market memory live.

Explore how Claude, Cursor, Codex and autonomous agents chain our 16 MCP tools across BTC, ETH and SOL. Click any diamond node on the topology graph below to fire a real JSON-RPC 2.0 call and inspect the structured evidence.

⚡ MCP Protocol LIVE
JSON-RPC 2.0
POST /api/mcp · Stateless HTTP
🛠️ Interactive Tools
10core tools on graph
Click any diamond ◆ to execute
🤖 Live MCP Calls (24h)
228RPC · 100 tools
Cursor · Claude · Codex · Grok
🎯 Active Graph Node
scan_candle_memory_radar
Stage 1 · Radar Scan
🪙 Target Window
BTC1hq=5 f=10
Free preview · no key needed
MCP Tool Execution Topology Graph · Click Any Node to Run Live
5 AI clients · 10 MCP tools · 3 market feeds · 0 protocol links
● AI Client (Cursor, Claude, Codex)◆ MCP Tool (click to run)⬢ Market Feed (BTC, ETH, SOL)drag · scroll zoom · click node
Stage 1 · Radar Scan

scan_candle_memory_radar

Scans BTC, ETH and SOL across timeframes and ranks windows by how tightly 20 historical twins agree on magnitude.

Coin:
TF:
🔗 Chain Next Tool in Graph:
Open BTC 1h on /candles Chart → Check a hypothesis →

Live Researcher Entity · 4-Stage Autonomous Pipeline

Turn any AI into your personal market researcher.

Connect Claude, Cursor, ChatGPT or Codex once. Your agent becomes a standing research entity: it scans the market matrix, generates competing price-level and session hypotheses, replays historical touches with fees, and messages you only when the evidence crosses the bar.

Autonomous Research Entity STAGE 1 · RADAR SCAN

AIPP Matrix Sentinel · Real-Time Market Researcher

watchId: sentinel-matrix-v1-15m15m patrol · NO_TRADE contract

Scanning 9 windows (BTC/ETH/SOL × 15m/1h/4h) · 20 historical twins per window

1. Choose Agent Specialization 4 modes
2. Watch Schedule
Real-Time Researcher Reasoning Loop tool: run_candle_research_watch
Scanning BTC, ETH, SOL across 15m / 1h / 4hTop lead: BTC 1h (bandWidthRatio 0.38)
  • RADARBTCUSDT 1h (q=5, f=10): 20 historical twins agree on terminal magnitude (38% as wide as all-history).
  • BASELINEWalk-forward get_candle_track_record verified: analogue band beats unconditional baseline (+21% Winkler lift, 120 anchors).
  • FILTER7 noisier timeframe windows filtered out automatically before bothering the user.

Awaken this researcher in Claude, Cursor, ChatGPT or Telegram

Connect https://aipricepatterns.com/api/mcp or launch 24/7 in Telegram:

24/7 Telegram Feed ↗ 
You are my AI Price Patterns Market Research Entity (AIPP Matrix Sentinel · Real-Time Market Researcher). Read get_api_guide, topic watch. Every 15m, Run run_candle_research_watch across BTCUSDT, ETHUSDT and SOLUSDT (15m, 1h, 4h) with watchId 'sentinel-matrix-v1'. When status is ANOMALY_READY and shouldMessageUser is true, also check investigate_candle_event_hypothesis for UTC session or event-window clustering and message me in this chat with the market, anchorUTC, tested hypotheses (including rejected ones), counts, and the /candles link. Otherwise stay quiet. Never name a side or a size, and end every note with NO_TRADE.
Uses run_candle_research_watch + investigate_candle_event_hypothesis Configure & connect on /mcp →
A real research result · SOL3 Oct 2026 · 12:00 UTC

If SOL reaches $120,
what happened next?

From a $119.48 query close, we inspected similar situations over the next 10 hourly candles. Target: $121.20. Invalidation: $118.80.

13 of 20

historical analogues reached $120

Among those 13 episodes, after reach:

Target first
6
Invalidation first
3
Unknown event order
2
Unfinished within the horizon
2

Observed historical counts, not a future probability or strategy win rate. Unknown order and unfinished cases stay in the result.

Run this example →View recorded evidence

How it works

From your idea to historical evidence.

01

Choose your market and levels

Pick BTC, ETH or SOL. Enter the level you want to investigate, with an optional target and invalidation.

02

See what happened after reach

Compare historical reach counts and subsequent event order. Unknown order and unfinished cases remain visible.

03

Replay the dated episodes

Open the examples behind the counts. Inspect the whole path and weigh the evidence for your idea.

Dated evidence

Every match is a real window with a date you can open and replay.

Path, not just final price

See which levels were reached first, even when the final close tells a different story.

Uncertainty stays visible

Counts include unknown event order and unfinished cases. Sample size is always shown.

Browser and AI agents

Run your research directly or let your assistant use the same evidence through MCP.

Check the quality behind the range estimate.

Compare range estimates with an all-history baseline for each market and timeframe. These scores evaluate ranges; they do not establish that a level hypothesis is a profitable strategy.

See the full record and method →

Bring a question. Inspect the evidence.

Search crypto market memory, or hand it to your AI agent.

Open Market Memory → Check a hypothesis