Stay tuned: Full source code and detailed documentation are on the way! Follow https://x.com/Sayantan317 for updates.
AI-powered trading chat interface with real-time market analysis
Trade Copilot is a trading assistant that combines institutional-grade market analysis with cutting-edge AI. Built on the Model Context Protocol (MCP), it provides real-time order flow analysis, options monitoring, and technical analysis through a clean, professional chat interface.
mindmap
root((Trade Copilot<br/>))
Trading Memory
get_current_bias
check_consistency
store_trading_decision
force_reset
Technical Analysis
volume_profile
technical_analysis
technical_zones
orb_analysis
fvg_analysis
Order Flow
equity_order_flow
pattern_detection
options_order_flow
options_monitoring
Options Intelligence
open_interest_analysis
institutional_positioning
gamma_exposure
- Real-time Order Flow Analysis - Track institutional vs retail sentiment
- Options Flow Monitoring - Detect sweeps, blocks, and unusual activity
- Volume Profile Analysis - Identify key support/resistance levels (POC, VAH, VAL)
- Technical Analysis Suite - ORB, FVG, technical zones, and indicators
- Parallel Tool Execution - Sub-30 second comprehensive market analysis
- AWS Bedrock - Claude Sonnet 4, Claude 3.5 Sonnet, Claude Haiku
- Anthropic API - Direct Claude API integration
- Runtime Switching - Change providers/models without restart
- Optimized Prompts - Military-precision system prompts for speed and accuracy
- MCP Protocol - Native integration with market data servers
- Go-based Data Broker - Sub-10ms latency for real-time data
- Concurrent Processing - Parallel tool execution for maximum speed
- Professional Interface - Rich terminal UI with formatted responses
- Python 3.11+
- AWS credentials (for Bedrock)
- Anthropic API key (for Claude API)
- Market data API keys
# Clone the repository
git clone https://github.com/your-org/trade-copilot.git
cd trade-copilot
# Install dependencies
pip install -r requirements.txt
# Or for development
pip install -r requirements-dev.txt
# Copy environment file
cp .env.example .env
# Edit configuration
nano config/config.yaml# config/config.yaml
llm_providers:
bedrock:
region: "us-east-1"
models:
claude_sonnet_4: "bedrock.anthropic.claude-sonnet-4-20250514"
claude_sonnet_35: "anthropic.claude-3-5-sonnet-20241022-v2:0"
claude_haiku: "anthropic.claude-3-haiku-20240307-v1:0"
claude_api:
models:
claude_sonnet_4: "claude-3-5-sonnet-20250514"
claude_sonnet_35: "claude-3-5-sonnet-20241022"
claude_haiku: "claude-3-haiku-20240307"
# Default settings
default_provider: "bedrock"
default_model: "claude_sonnet_4"mcp_servers:
market_data:
command: "python"
args: ["-m", "mcp_market_data_server"]
env:
TWELVE_DATA_API_KEY: "${TWELVE_DATA_API_KEY}"
order_flow:
command: "python"
args: ["-m", "mcp_order_flow_server"]
env:
GRPC_BROKER_URL: "localhost:9090"
options_flow:
command: "python"
args: ["-m", "mcp_options_order_flow_server"]
env:
GRPC_BROKER_URL: "localhost:9090"# Start chat interface
trade-copilot chat
# Use specific provider/model
trade-copilot chat --provider claude_api --model claude_sonnet_4
# In chat interface
help # Show help
status # System status
clear # Clear history
switch bedrock claude_sonnet_4 # Switch provider
exit # End sessionQuery: "Analyze SPY order flow for the last 10 minutes"
Expected Tools Called:
├─ analyze_order_flow_tool(ticker="SPY", minutes=10)
├─ financial_volume_profile(ticker="SPY", timeframe="1m")
└─ financial_technical_analysis(ticker="SPY", timeframes=["1m", "5m"])
Sample Response:
**SIGNAL:** BUY CALLS
**CONVICTION:** 78/100 (3 confirming signals)
**ENTRY:** $472.25
**STOP:** $469.80 (2% risk)
**TARGET:** $477.15 (2:1 R/R)
**TIMEFRAME:** 15 minutes validity
**EVIDENCE:**
- Order flow: 68% institutional buying pressure, large block at 472.00
- Volume profile: POC at 471.95 acting as support, 89% buy volume above POC
- Technical: Clean break above 5min EMA with increasing momentumQuery: "Show me NVDA options unusual activity and sweeps"
Expected Tools Called:
├─ get_options_flow(ticker="NVDA", history_minutes=30)
├─ configure_options_monitoring_tool(ticker="NVDA", min_premium=50000)
└─ analyze_order_flow_tool(ticker="NVDA", minutes=15)
Sample Response:
**SIGNAL:** WAIT - Mixed signals
**CONVICTION:** 45/100 (conflicting data)
**ANALYSIS:**
- Options: $2.3M in call sweeps at 520 strike (2x normal), but also 1.2M put volume
- Unusual activity: 400% increase in 525C volume, premium suggesting 3% move expected
- Underlying flow: Retail buying pressure (78%), institutions net selling
**INVALIDATION:** Clear above 522.50 = bullish confirmationQuery: "Check QQQ opening range breakout and volume profile"
Expected Tools Called:
├─ financial_orb_analysis(ticker="QQQ", orb_minutes=30)
├─ financial_volume_profile(ticker="QQQ", timeframe="5m")
├─ financial_technical_zones(ticker="QQQ")
└─ financial_technical_analysis(ticker="QQQ")
Sample Response:
**SIGNAL:** BUY CALLS
**CONVICTION:** 82/100 (4 confirming signals)
**ENTRY:** $398.75 (ORB high + 0.05)
**STOP:** $396.90 (ORB low - 0.10)
**TARGET:** $402.45 (1.2x ORB range)
**EVIDENCE:**
- ORB: Clean breakout above 30min range (396.95-398.70), volume 2.1x average
- Volume profile: 73% volume above POC (397.85), strong buying interest
- Technical zones: Breaking resistance cluster at 398.60-398.80
- ADX: 28.4 showing strong trend, RSI healthy at 58.7Query: "Compare tech vs financials order flow - XLK vs XLF"
Expected Tools Called:
├─ analyze_order_flow_tool(ticker="XLK", minutes=20)
├─ analyze_order_flow_tool(ticker="XLF", minutes=20)
├─ financial_volume_profile(ticker="XLK")
├─ financial_volume_profile(ticker="XLF")
└─ sector_comparison_analysis(tickers=["XLK", "XLF"])
Sample Response:
**SECTOR ROTATION DETECTED:**
**XLK (Tech):** 72% institutional selling, volume 1.8x average
**XLF (Financials):** 81% institutional buying, volume 2.3x average
**ROTATION SIGNAL:** Tech → Financials rotation confirmed
**TRADE IDEA:** Long XLF, Short XLK (pairs trade)┌─────────────────────────────────────────────────────────────────────────────────────┐
│ CLIENT INTERACTION │
├─────────────────────────────────────────────────────────────────────────────────────┤
│ User Query: "Analyze SPY order flow and options activity" │
│ │ │
│ ▼ │
├─────────────────────────────────────────────────────────────────────────────────────┤
│ TRADE COPILOT CORE │
├─────────────────────────────────────────────────────────────────────────────────────┤
│ ┌─────────────────────────────────────────────────────────────────────────────┐ │
│ │ CHAT INTERFACE (Rich Terminal) │ │
│ │ • Input validation and parsing │ │
│ │ • Real-time response streaming │ │
│ │ • Conversation history persistence (SQLite) │ │
│ │ • Model switching commands │ │
│ └─────────────────────────────────────────────────────────────────────────────┘ │
│ │ │
│ ┌─────────────────────────────────────────────────────────────────────────────┐ │
│ │ LLM PROVIDER MANAGER │ │
│ │ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────────┐ │ │
│ │ │ AWS Bedrock │ │ Claude API │ │ Model Switching │ │ │
│ │ │ (boto3) │ │ (anthropic-sdk)│ │ (Runtime) │ │ │
│ │ │ │ │ │ │ │ │ │
│ │ │ • Claude-3.5 │ │ • Claude-3.5 │ │ • Provider select │ │ │
│ │ │ • Claude Haiku │ │ • Claude Haiku │ │ • Model switching │ │ │
│ │ │ • Auto-scaling │ │ • Direct API │ │ • Config hot-reload │ │ │
│ │ └─────────────────┘ └─────────────────┘ └─────────────────────┘ │ │
│ └─────────────────────────────────────────────────────────────────────────────┘ │
│ │ │
│ ┌─────────────────────────────────────────────────────────────────────────────┐ │
│ │ MCP TOOL MANAGER │ │
│ │ • Tool discovery and registration │ │
│ │ • Parallel tool execution (15+ concurrent) │ │
│ │ • Error handling and retry logic │ │
│ │ • Tool result aggregation and formatting │ │
│ └─────────────────────────────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────────────────────────────┘
│
PARALLEL TOOL EXECUTION
│
┌─────────────────────────┼─────────────────────────┐
▼ ▼ ▼
┌─────────────────────────┐ ┌─────────────────────────┐ ┌─────────────────────────┐
│ MCP MARKET DATA │ │ MCP ORDER FLOW │ │ MCP OPTIONS FLOW │
│ SERVER │ │ SERVER │ │ SERVER │
├─────────────────────────┤ ├─────────────────────────┤ ├─────────────────────────┤
│ TOOLS: │ │ TOOLS: │ │ TOOLS: │
│ • financial_technical_ │ │ • analyze_order_flow_ │ │ • get_options_flow │
│ analysis │ │ tool │ │ • configure_options_ │
│ • financial_volume_ │ │ • get_level_2_data │ │ monitoring_tool │
│ profile │ │ • get_trade_prints │ │ • get_unusual_activity │
│ • financial_technical_ │ │ • detect_patterns │ │ • analyze_sweep_blocks │
│ zones │ │ │ │ │
│ • financial_orb_ │ │ DATA SOURCES: │ │ DATA SOURCES: │
│ analysis │ │ • Real-time quotes │ │ • Options chains │
│ • financial_fvg_ │ │ • Level 2 book data │ │ • Unusual activity │
│ analysis │ │ • Trade patterns │ │ • Greeks calculations │
│ │ │ • Volume analysis │ │ • Flow direction │
│ DATA SOURCES: │ │ │ │ │
│ • Twelve Data API │ │ ┌─────────────────────┐ │ │ ┌─────────────────────┐ │
│ • Technical indicators │ │ │ gRPC CLIENT │ │ │ │ gRPC CLIENT │ │
│ • Volume calculations │ │ │ localhost:9090 │ │ │ │ localhost:9090 │ │
│ • Multi-timeframe │ │ └─────────────────────┘ │ │ └─────────────────────┘ │
└─────────────────────────┘ └─────────────────────────┘ └─────────────────────────┘
│ │
gRPC CONNECTION (sub-10ms latency) │
│ │
▼ ▼
┌─────────────────────────────────────────────────────────────────────────────────────┐
│ MCP TRADING DATA BROKER (Go) │
├─────────────────────────────────────────────────────────────────────────────────────┤
│ ┌─────────────────────────────────────────────────────────────────────────────┐ │
│ │ CORE SERVICES │ │
│ │ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────────────────┐ │ │
│ │ │ Order Flow │ │ Options Flow │ │ Market Data Cache │ │ │
│ │ │ Aggregator │ │ Processor │ │ │ │ │
│ │ │ │ │ │ │ • Real-time price feeds │ │ │
│ │ │ • Pattern │ │ • Sweep/Block │ │ • Historical OHLCV │ │ │
│ │ │ detection │ │ detection │ │ • Level 2 order book │ │ │
│ │ │ • Institutional │ │ • Unusual │ │ • Trade tape analysis │ │ │
│ │ │ vs retail │ │ activity │ │ • Volume profile cache │ │ │
│ │ │ • Volume │ │ • Greeks │ │ │ │ │
│ │ │ analysis │ │ calculation │ │ │ │ │
│ │ └─────────────────┘ └─────────────────┘ └─────────────────────────────┘ │ │
│ └─────────────────────────────────────────────────────────────────────────────┘ │
│ │ │
│ ┌─────────────────────────────────────────────────────────────────────────────┐ │
│ │ WEBSOCKET CLIENT POOL │ │
│ │ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────────────────┐ │ │
│ │ │ 3P WS │ │ Market WS │ │ Health Monitor │ │ │
│ │ │ Client │ │ Client │ │ │ │ │
│ │ │ │ │ │ │ • Connection monitoring │ │ │
│ │ │ • Real-time │ │ • Level 2 data │ │ • Auto-reconnection │ │ │
│ │ │ options data │ │ • Trade prints │ │ • Latency tracking │ │ │
│ │ │ • Order flow │ │ • Quote stream │ │ • Error logging │ │ │
│ │ │ • Auto-retry │ │ • Volume feeds │ │ • Performance metrics │ │ │
│ │ │ • 60s timeout │ │ │ │ │ │ │
│ │ │ monitoring │ │ │ │ │ │ │
│ │ └─────────────────┘ └─────────────────┘ └─────────────────────────────┘ │ │
│ └─────────────────────────────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────────────────────────────┘
│
EXTERNAL DATA INGESTION
│
┌─────────────────────────┼─────────────────────────┐
▼ ▼ ▼
┌─────────────────────────┐ ┌─────────────────────────┐ ┌─────────────────────────┐
│ 3P API │ │ TWELVE DATA API │ │ POLYGON/OTHER APIs │
├─────────────────────────┤ ├─────────────────────────┤ ├─────────────────────────┤
│ • Real-time options │ │ • Stock market data │ │ • Alternative feeds │
│ • Order flow data │ │ • Technical indicators │ │ • News sentiment │
│ • Institutional trades │ │ • Historical OHLCV │ │ • Economic data │
│ • Unusual activity │ │ • Volume data │ │ • Social sentiment │
│ • WebSocket streams │ │ • Fundamental data │ │ │
└─────────────────────────┘ └─────────────────────────┘ └─────────────────────────┘
┌─────────────────────────────────────────────────────────────────────────────────────┐
│ CONFIGURATION SYSTEM │
├─────────────────────────────────────────────────────────────────────────────────────┤
│ ┌─────────────────────────────────────────────────────────────────────────────┐ │
│ │ /etc/config/TradingInsightsAgentConfig.json │ │
│ │ { │ │
│ │ "ANTHROPIC_API_KEY": "sk-ant-...", │ │
│ │ "CLAUDE_API_KEY": "sk-ant-...", │ │
│ │ "3P _API_KEY": "3p_key_...", │ │
│ │ "TWELVE_DATA_API_KEY": "td_key_...", │ │
│ │ "AWS_ACCESS_KEY_ID": "AKIA...", │ │
│ │ "AWS_SECRET_ACCESS_KEY": "...", │ │
│ │ "GRPC_BROKER_URL": "localhost:9090" │ │
│ │ } │ │
│ └─────────────────────────────────────────────────────────────────────────────┘ │
│ │ │
│ ┌─────────────────────────────────────────────────────────────────────────────┐ │
│ │ config/config.yaml │ │
│ │ llm_providers: │ │
│ │ bedrock: │ │
│ │ region: "us-east-1" │ │
│ │ models: │ │
│ │ claude_sonnet_4: "bedrock.anthropic.claude-sonnet-4-20250514" │ │
│ │ claude_api: │ │
│ │ models: │ │
│ │ claude_sonnet_4: "claude-3-5-sonnet-20250514" │ │
│ │ │ │
│ │ mcp_servers: │ │
│ │ market_data: "python -m mcp_market_data_server" │ │
│ │ order_flow: "python -m mcp_order_flow_server" │ │
│ │ options_flow: "python -m mcp_options_order_flow_server" │ │
│ └─────────────────────────────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────────────────────────────┘
USER QUERY → INPUT VALIDATION → LLM ANALYSIS → TOOL EXECUTION → RESPONSE GENERATION
│ │ │ │ │
▼ ▼ ▼ ▼ ▼
┌─────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ "SPY │ │ Parse & │ │ Claude │ │ Parallel │ │ Aggregate │
│ options │ │ validate │ │ decides │ │ MCP tool │ │ results & │
│ flow" │ │ ticker, │ │ which tools │ │ execution: │ │ format for │
│ │ │ timeframe │ │ to call │ │ • Options │ │ trading │
│ │ │ params │ │ based on │ │ • Order │ │ decision │
│ │ │ │ │ context │ │ • Volume │ │ │
└─────────┘ └─────────────┘ └─────────────┘ └─────────────┘ └─────────────┘
│
CONCURRENT EXECUTION
(Sub-30s total)
User Input: "Analyze SPY order flow and options activity"
Step 1: Query Processing (50ms)
- Input validation and ticker extraction
- Context analysis for tool selection
- Historical conversation context loading
Step 2: LLM Tool Selection (200ms)
- Claude analyzes query requirements
- Selects 3-5 relevant tools based on system prompt
- Plans parallel execution strategy
Step 3: Parallel Tool Execution (1-2s)
TOOL 1: analyze_order_flow_tool(ticker="SPY", minutes=10)
├─ gRPC call to data broker
├─ Real-time order flow aggregation
└─ Pattern detection & sentiment analysis
TOOL 2: get_options_flow(ticker="SPY", history_minutes=20)
├─ Options chain analysis
├─ Unusual activity detection
└─ Sweep/block identification
TOOL 3: financial_volume_profile(ticker="SPY", timeframe="5m")
├─ Twelve Data API call
├─ Volume profile calculation
└─ POC/VAH/VAL identification
TOOL 4: financial_technical_analysis(ticker="SPY")
├─ Multi-timeframe indicator analysis
├─ Trend strength assessment
└─ Support/resistance levels
