MCP Trading Agent
Enables an AI agent to perform live demo trading analysis using market data, news, backtesting, breakout scanning, and persistent trading knowledge tools.
README
MCP Trading Agent v3.0 — ICT / SMC + News Sentiment
======================================================
A production-ready MCP server exposing 13 market-data, news,
backtesting, and persistence tools to an LLM agent (Nexus v2).
The agent learns from backtests and applies those rules to
live analysis — compounding its edge over time.
Quick Start
-----------
1. Install dependencies:
pip install -r requirements.txt
2. Run the server (stdio transport for Claude Desktop / Claude Code):
python server.py
3. Or run with HTTP transport (for MCP Inspector / web):
set MCP_TRANSPORT=streamable-http
python server.py
4. Test with the MCP Inspector:
npx -y @modelcontextprotocol/inspector
# Then connect to http://localhost:8000/mcp
Claude Desktop Integration
--------------------------
Add this to your Claude Desktop config (~/.claude/config.json):
{
"mcpServers": {
"trading-agent": {
"command": "python",
"args": ["C:\Github\ai-company\mcp-trading-agent\server.py"]
}
}
}
Project Structure
-----------------
mcp-trading-agent/
├── server.py # 13 MCP tool registrations + entry point
├── config.py # ServerConfig dataclass (v2.0.0)
├── system_prompt.py # Nexus v2 persona — 7 command workflows
├── CLAUDE.md # Auto-loaded by Claude Code (same as above)
├── requirements.txt # mcp[cli], yfinance, ddgs, pandas, numpy
├── README.md # This file
├── v2_upgrade_walkthrough.md # Architecture & evolution docs
├── tools/
│ ├── market_data.py # 7 functions: OHLC, liquidity, backtest,
│ │ # intraday backtest, MTF fetch, breakout scan
│ ├── news.py # fetch_market_news (DuckDuckGo / ddgs)
│ ├── risk_reward.py # get_risk_to_reward_setup
│ └── persistence.py # HTML reports, lessons.md CRUD,
│ # sync_trading_knowledge (SHA-256 dedup)
└── data/ # Persistent state (auto-created on first run)
├── lessons.md # Knowledge base — rules learned from backtests
├── lessons.hashes # SHA-256 fingerprints for dedup sidecar
└── reports/ # HTML analysis reports (30-day auto-purge)
All 13 MCP Tools
----------------
v1 — Original (5 tools)
┌──────────────────────────────┬──────────────────────────────────────────┐
│ Tool │ Purpose │
├──────────────────────────────┼──────────────────────────────────────────┤
│ get_daily_ohlc │ Daily OHLCV candles (60-day default) │
│ get_intraday_ohlc │ Sub-daily candles (1m/5m/15m/30m/60m) │
│ identify_liquidity_pools │ Swing high/low detection (BSL / SSL) │
│ fetch_market_news │ DuckDuckGo news search (fundamental bias)│
│ get_risk_to_reward_setup │ RR ratio + quality verdict │
└──────────────────────────────┴──────────────────────────────────────────┘
v2 — Stateful / Backtest (6 tools)
┌──────────────────────────────┬──────────────────────────────────────────┐
│ Tool │ Purpose │
├──────────────────────────────┼──────────────────────────────────────────┤
│ get_historical_backtest_data │ Extended OHLCV w/ swing flags (10-500d) │
│ run_intraday_backtest │ Auto SMC scan: sweep+FVG, RR>=3, w-fwd │
│ manage_html_report │ Save HTML + auto-open browser + 30d purge│
│ read_lessons_learned │ Read lessons.md knowledge base │
│ update_lessons_learned │ Append free-form insights to lessons.md │
│ sync_trading_knowledge │ SHA-256 dedup + persist structured rules │
└──────────────────────────────┴──────────────────────────────────────────┘
v3 — Breakout Scanner (2 tools)
┌──────────────────────────────┬──────────────────────────────────────────┐
│ Tool │ Purpose │
├──────────────────────────────┼──────────────────────────────────────────┤
│ get_multi_timeframe_data │ Monthly + Weekly + Daily OHLC in one call│
│ scan_for_breakout │ 1-10 score across M/W/D timeframes │
└──────────────────────────────┴──────────────────────────────────────────┘
Agent Commands (plain-text, not slash commands)
-----------------------------------------------
Command Data Source
──────────────────────────────────────────────────────────────────
backtest [ticker] [days] get_historical_backtest_data
backtest intraday [ticker] [days] [intv] run_intraday_backtest
analyze [ticker] get_daily_ohlc + liquidity
entry [ticker] (runs analyze silently first)
intraday [ticker] get_intraday_ohlc — 1:3 RR gate
view [ticker] get_daily_ohlc — macro swing
breakout [ticker or list] get_multi_timeframe_data +
scan_for_breakout
Example session:
backtest intraday NIFTY 60 15m
→ automated SMC scan, equity curve HTML, rules saved to lessons.md
intraday NIFTY
→ live 15m analysis with 1:3 RR gate, HTML report auto-opened
breakout NIFTY, BTC, GOLD
→ D/W/M alignment matrix, conviction scores, trigger prices
analyze AAPL
→ fundamental + technical confluence, HTML report
entry AAPL
→ tight trade card: entry / SL / target / RR / verdict
Breakout Scoring (scan_for_breakout)
-------------------------------------
Scoring breakdown (10 pts max):
Monthly (3 pts): price > EMA-6, near BSL <= 5%, 2/3 months bullish
Weekly (3 pts): price > EMA-20, volatility contraction, near BSL <= 3%
Daily (4 pts): price > EMA-20, displacement candle, FVG present,
volume spike >= 1.3× 20-day avg
Conviction tiers:
8-10 HIGH (high_conviction_confirmed when M + W both BULLISH)
5-7 MODERATE (Watch and Wait)
1-4 LOW (no confluence)
Trigger price = nearest daily BSL above current price.
Supported Ticker Aliases
------------------------
NIFTY → ^NSEI, BANKNIFTY → ^NSEBANK, SENSEX → ^BSESN,
SPX → ^GSPC, SPY → SPY, QQQ → QQQ, DXY → DX-Y.NYB,
GOLD → GC=F, CRUDE → CL=F, BTC → BTC-USD, ETH → ETH-USD
Intraday Backtest — What run_intraday_backtest Returns
-------------------------------------------------------
Per-trade fields:
setup_type, direction, sweep_time, session_label, hour,
swept_level, fvg_zone, entry, stop_loss, target,
risk_pts, reward_pts, rr, atr_at_setup, is_consecutive_sweep,
outcome (WIN/LOSS/OPEN), exit_price
Aggregate stats:
win_rate_pct, avg_rr, profit_factor, max_drawdown_r,
expectancy_r, equity_curve (R-multiple list), session_breakdown
session_breakdown keys (Opening/Morning/Midday/Afternoon/Closing):
total, wins, losses, open_trades, win_rate_pct
Knowledge Persistence
---------------------
lessons.md is automatically maintained across sessions.
sync_trading_knowledge uses SHA-256 fingerprints stored in
lessons.hashes to prevent near-duplicate rules accumulating.
HTML reports older than 30 days are auto-purged by manage_html_report.
License: MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.