Agent Skills: Execute a trade with full tracking and notifications

Python Coding Standards

Modular Architecture

The backend is modularized into packages:

• src.trading.orders: Order management (limit, market, batch, positions).
• src.data.market_data: Data fetching (polymarket, binance, indicators, price validation).
• src.trading: Centralized execution logic (execution.py, logic.py, strategy.py).
• src.trading.position_manager: Position lifecycle management (entry, exit, scale-in, stop-loss, reconciliation).

Standard Trade Execution

from src.trading import execute_trade
# Execute a trade with full tracking and notifications
trade_id = execute_trade(trade_params, is_reversal=False)

Imports

• Use absolute imports from src.*
• Group imports: standard library → third-party → local
• Always import from src.config.settings for configuration constants

Formatting

• Indentation: 4 spaces
• Line length: ~90 characters
• Strings: Double quotes preferred
• Docstrings: Triple double-quotes with brief description

Error Handling & Logging

• Use log_error(text, include_traceback=True) from src.utils.logger for all exceptions.
• Always log errors with context: log(f"[{symbol}] Order failed: {e}").
• Graceful degradation: return neutral/safe values on error.
• Use emojis in logs (👀 Monitoring, 🚀 Execution, ✅ Success, ❌ Error).

Thread Safety & Synchronization (v0.4.2+)

• Use threading locks for shared resources: with position_lock: ...
• Always confirm order fills before cancellation to prevent race conditions
• Implement reconciliation systems for order tracking consistency
• Use recent_fills tracking to prevent duplicate processing
• Add synchronization locks for all critical trading operations

Confidence Algorithm Standards (v0.4.2+)

• Cap confidence at 85% maximum to prevent overconfidence
• Validate price movements before high-confidence trades (>75%)
• Implement multi-confirmation systems for extreme confidence levels
• Use directional voting with weighted signal aggregation
• Test confidence ranges (0-100%) and bias validation thoroughly

Bayesian Confidence Calculation (v0.5.0+)

• Dual Calculation System: Both additive and Bayesian methods are calculated simultaneously for A/B testing
• Configuration Flag: BAYESIAN_CONFIDENCE in src/config/settings.py toggles between methods (default: NO)
• Bayesian Formula:

  # Start with market prior from Polymarket orderbook
  prior_odds = p_up / (1 - p_up)
  log_odds = ln(prior_odds)
  
  # Accumulate evidence from all signals
  for each signal:
      evidence = (score - 0.5) * 2  # -1 to +1
      log_LR = evidence * 3.0 * quality  # Calibration with quality
      log_odds += log_LR × weight
  
  # Convert to probability
  confidence = 1 / (1 + exp(-log_odds))
  

• Quality Factors (0.7-1.5x): Applied to log-likelihood strength for each signal
• Market Prior: Anchors calculation to Polymarket orderbook probability
• A/B Testing: Both methods stored in raw_scores dictionary and database for comparison
• Database Schema: Migration 007 adds additive_confidence, additive_bias, bayesian_confidence, bayesian_bias, market_prior_p_up columns

Audit Trail Requirements (v0.4.2+)

• Log all scale-in order lifecycle events with consistent formatting
• Use standardized emojis: 🎯 (placement), ✅ (filled), 🧹 (cancellation), ⚠️ (race condition)
• Include order IDs, sizes, prices, and timestamps in all audit logs
• Maintain separate audit trails for notifications, reconciliation, and position management
• Log notification processing with order ID extraction details

Database Safety Standards

• Always use db_connection() context manager
• Pass active cursors to write functions within transactions
• Implement migration system for schema changes
• Use schema_version table to track database versions
• Prevent deadlocks with proper cursor management

Configuration Management

• Centralize all settings in src.config.settings
• Use environment variables for sensitive data
• Implement validation for critical thresholds (MIN_EDGE, confidence caps)
• Document all configuration options in .env.example
• Add price validation settings for high-confidence trades

Testing Standards

• Create comprehensive test suites for algorithm validation
• Test confidence ranges (0-100%), bias validation, and safety features
• Implement integration tests for multi-confirmation systems
• Validate price movement detection and manipulation prevention
• Test race condition prevention and synchronization

Performance Optimization

• Use efficient API calls with proper caching
• Implement batch processing for multiple orders
• Monitor execution times and optimize critical paths
• Use WebSocket connections for real-time data where possible
• Optimize notification processing with efficient order ID extraction

Documentation Standards

• Update README.md with latest features and improvements
• Maintain detailed strategy documentation in docs/STRATEGY.md
• Document all risk profiles in docs/RISK_PROFILES.md
• Keep migration guide updated in docs/MIGRATIONS.md
• Document new position flow in docs/POSITION_FLOW.md

Code Quality Standards

• Use type hints for all function parameters and return values
• Implement proper error handling with specific exception types
• Write comprehensive docstrings for all public functions
• Follow PEP 8 style guidelines with project-specific adaptations
• Implement comprehensive audit trail logging

Security Standards

• Never commit private keys or sensitive credentials
• Use secure methods for API key storage and access
• Implement proper input validation for all external data
• Use parameterized queries for database operations
• Secure notification processing with robust parsing

Deployment Standards

• Use Docker for consistent deployment environments
• Implement proper logging aggregation for production
• Monitor system health and performance metrics
• Use proper process management and restart policies
• Deploy with comprehensive monitoring and alerting

Monitoring & Alerting

• Implement comprehensive audit trail logging
• Monitor for race conditions and synchronization issues
• Track order fill rates and reconciliation success
• Alert on unusual patterns or system errors
• Monitor confidence algorithm performance and validation
• Track notification processing success rates

Version Control Standards

• Use semantic versioning (MAJOR.MINOR.PATCH)
• Create detailed release notes for each version
• Tag releases with comprehensive change descriptions
• Maintain backward compatibility where possible
• Document all improvements in release tags

Balance Validation & Position Sync (v0.4.4+)

• Enhanced Balance Validation: Symbol-specific tolerance for API reliability issues (XRP, etc.)
  • Default: 0.8 reliability weight, 0.2 position trust factor, 5-minute grace period
  • XRP: 0.3 reliability weight (lower trust), 0.3 position trust, 15-minute grace period
• Real-Time Exit Order Validation: 1-second cycle size validation to catch mismatches immediately
• Rounding Threshold: Use 0.05 share threshold for exit order repair to prevent infinite loops
• Bi-Directional Healing: Sync database size with actual wallet balance when discrepancies detected
  • Sync if balance > DB size + 0.0001 (immediate, no cooldown)
  • Sync if age > 60s AND scale-in age > 60s AND |balance - size| > 0.0001
• Grace Period Logic:
  • 60-second grace period after buy/scale-in before balance validation
  • 10-minute grace period for zero balance before ghost trade settlement (increased from 5m)
• API Retry Strategy: Exponential backoff with symbol-specific retry counts
  • Default: 2 retries with 1.0s delay
  • XRP: 3 retries with 2.0s delay
  • Crypto markets: +1 extra retry with 1.5x delay multiplier
• Geographic Restrictions: Detect and skip retries for geo-blocked APIs
• Database as Source of Truth: For exit orders, always trust database position size over balance API

Position Monitoring & Reporting (v0.4.3+)

• Clean Position Reports: Aligned format with directional emojis (📈 UP winning, 📉 DOWN winning)
• Status Indicators: Visual indicators for exit plan status (⏰ active, ⏳ pending, ⏭️ skipped)
• Trade ID Display: Show trade ID for all positions in monitoring output
• Spam Reduction: Removed debug logging spam for cleaner production logs
• Min Size Handling: Silent skip for positions below MIN_SIZE (5.0 shares) with status display

Notification Processing (v0.4.3+)

• Multi-Field Order ID Extraction: Robust parsing with fallback to multiple field names
• Concise Logging: Streamlined notification processing logs with symbol and price info
• Batch Processing: Efficient handling of multiple notifications at once

Version History

v0.5.0 (Jan 2026)

• Bayesian Confidence Calculation: Implemented statistically principled probability calculation using log-odds and likelihood ratios
• Dual Calculation System: Both additive and Bayesian methods calculated simultaneously for A/B testing
• Market Prior Integration: Bayesian method uses Polymarket orderbook probability as prior for anchored calculations
• Quality Factor System: Evidence strength modulated by signal quality (0.7-1.5x multiplier on log-likelihood)
• Database Migration 007: Added columns for storing both methods' results and market prior
• Configuration Toggle: BAYESIAN_CONFIDENCE flag enables Bayesian method for production use

v0.4.4 (Jan 2026)

• Exit Order Repair Fix: Prevent infinite repair loops from exchange rounding differences (0.05 threshold)
• Real-Time Validation: Exit order size validation every 1-second cycle to catch mismatches immediately
• Log Cleanup: Remove noisy balance API warnings for near-zero values
• Position Display: Show "Exit skipped" status for positions below MIN_SIZE threshold

v0.4.3 (Jan 2026)

• Enhanced Position Reports: Clean, aligned format with directional emojis and status indicators
• Professional Display: Removed debug spam and redundant logging for cleaner trading logs
• Visual Clarity: Perfect alignment with trade IDs, position sizes, and PnL percentages
• Unified Format: Consolidated position monitoring with consistent visual indicators

v0.4.2 (Jan 2026)

• Confidence Algorithm Fixes: Eliminated false 100% signals with proper capping and validation
• Scale-in Order Race Condition Prevention: Implemented order fill confirmation before cancellation
• Synchronization Locks: Added comprehensive thread safety across all trading operations
• Reconciliation System: Built order tracking reconciliation to prevent ghost trades
• Smart Exit Pricing: Implemented 0.999 exit pricing for improved winning trade fills
• Enhanced Audit Trail: Comprehensive logging for scale-in orders and position management
• Notification System: Fixed unknown order ID extraction with robust multi-field parsing
• Exit Plan Self-Healing: Improved balance validation consistency and grace period logic