mt5-ai-xauusd-trader/TRADING_STANDARDS.md

Trading Standards for MT5 AI/ML Trading Bot

1. Risk Management Standards

1.1 Position Sizing Rules

• Maximum Position Size: Limited to 5% of account equity per trade
• Leverage Limits: Maximum 1:50 for gold (XAUUSD) trading
• Correlation Limits: No 2+ correlated positions simultaneously
• Portfolio Heat: Maximum 70% of total budget committed across all strategies (default)
• Symbol Concentration: Maximum 40% of budget per symbol (default)
• Model Family Concentration: Maximum 40% of budget per model family (default)

1.2 Risk Parameters

# Example position sizing calculation
def calculate_position_size(
    account_equity: float,
    risk_percentage: float,
    stop_loss_pips: float,
    pip_value: float
) -> float:
    """Calculate position size based on risk management rules."""
    # Risk per trade in currency
    risk_amount = account_equity * (risk_percentage / 100)
    
    # Position size in lots
    position_size = risk_amount / (stop_loss_pips * pip_value)
    
    # Apply maximum position limit (5% of equity)
    max_position = (account_equity * 0.05) / (pip_value * 100)
    
    return min(position_size, max_position)

1.3 Stop-Loss & Take-Profit Rules

• Minimum Stop Distance: 10 pips for XAUUSD
• Mandatory Stop-Loss: Every trade must have a hard stop-loss
• Risk-Reward Ratio: Minimum 1:1.5 (1 risked to gain 1.5) [ENFORCED IN SCHEMA]
• Trailing Stops: Implement after profit threshold (e.g., +50 pips)
• Profit Target Levels: Multiple profit levels with partial exits

1.4 Capital Allocation & Performance Scaling

• Adaptive Sizing: Strategy risk is scaled by a performance multiplier (0.0 to 2.0).
• Cooling-Off: Automatic multiplier floor (0.1) after 5 consecutive losses.
• Diversification Guard: Linear scaling of allocations as heat or concentration limits are approached.
• Strategy Caps: Each strategy has an absolute capital cap.

1.5 Daily Loss Limits (Circuit Breaker)

• Daily Loss Limit: 3% of account equity
• Weekly Loss Limit: 8% of account equity
• Monthly Loss Limit: 15% of account equity
• Consequence: Halt trading when limits reached, resume next period
• Alert System: Email/SMS notification when approaching limits

2. Order Management & Execution

2.1 Order Validation Framework

class OrderValidator:
    """Validate all orders before execution."""
    
    def validate_order(self, order: Order) -> Tuple[bool, str]:
        """Validate order against all trading rules."""
        checks = [
            self._check_position_size(order),
            self._check_daily_loss_limit(order),
            self._check_correlation(order),
            self._check_stop_loss(order),
            self._check_take_profit(order),
            self._check_market_hours(order),
            self._check_news_event_risk(order),
        ]
        
        for is_valid, message in checks:
            if not is_valid:
                return False, message
        
        return True, "Order validated successfully"
    
    def _check_stop_loss(self, order: Order) -> Tuple[bool, str]:
        """Ensure stop-loss is within acceptable range."""
        if order.stop_loss is None:
            return False, "Stop-loss is required"
        
        distance_pips = abs(order.entry_price - order.stop_loss) / self.pip_value
        if distance_pips < 10:
            return False, f"Stop-loss too close: {distance_pips} pips"
        
        return True, "Stop-loss validated"

2.2 Order Execution Standards

• Order Types: Market, Limit, Stop-Loss only (no pending orders beyond 5 minutes)
• Slippage Tolerance: Maximum 2 pips slippage acceptable
• Order Confirmation: All fills must be confirmed against MT5 platform
• Rejected Orders: Log and analyze all rejected orders
• Partial Fills: Handle partial fills explicitly in code

2.3 Trade Reconciliation

• Frequency: Real-time reconciliation with MT5
• Procedure: Every hour: verify open positions, compare P&L, check balance
• Discrepancy Handling: Halt trading, investigate, resolve before resuming
• Documentation: Log all reconciliation results

3. Backtesting Standards

3.1 Backtesting Requirements

• Historical Data Range: Minimum 5 years of OHLCV data
• Data Quality: Verified against multiple sources for accuracy
• Market Conditions: Test through multiple market regimes:
  • Bull markets (2015-2017)
  • Bear markets (2018, 2020)
  • Sideways/consolidation periods
  • High volatility periods (Brexit, COVID, Fed announcements)

3.2 Backtest Metrics

Required Performance Metrics:
- Total Return: Target 50-100% annually
- Sharpe Ratio: Target > 2.0
- Sortino Ratio: Target > 3.0
- Max Drawdown: Maximum 25% of peak equity
- Win Rate: Target 55-65%
- Profit Factor: Target > 2.0 (wins/losses)
- Recovery Factor: Target > 2.0 (net profit/max drawdown)

3.3 Backtest Validation

• Out-of-Sample Testing: 20% of data reserved for testing
• Walk-Forward Analysis: Rolling 1-year windows with quarterly recalibration
• Sensitivity Analysis: Test parameter variations (±10%, ±20%)
• Monte Carlo Simulation: Randomize order sequence, test robustness
• Optimization Risk: Avoid over-fitting with parameter limits

4. Live Trading Standards

4.1 Paper Trading Phase

• Duration: Minimum 2 weeks before live trading
• Success Criteria:
  • 70%+ win rate in paper trading
  • Sharpe ratio > 2.0
  • No losses exceeding 2% per trade
  • Consistent performance across market conditions
• Documentation: Log all paper trades for review

4.2 Live Trading Graduation

• Phase 1 (Weeks 1-2): Micro lots, 0.01 lot per trade, $100-500 risk per trade
• Phase 2 (Weeks 3-4): Mini lots, 0.1 lot per trade, $500-1,000 risk per trade
• Phase 3 (Weeks 5+): Standard lots, 0.5-1.0 lot per trade, $1,000+ risk per trade
• Graduation Criteria: Each phase requires 70%+ win rate before advancing

4.3 Live Trading Monitoring

class LiveTradingMonitor:
    """Real-time monitoring of live trading operations."""
    
    def monitor_trade(self, trade: Trade) -> None:
        """Continuous monitoring of active trades."""
        while trade.is_open:
            current_price = self.get_current_price()
            
            # Check stop-loss and take-profit
            self._check_exit_conditions(trade, current_price)
            
            # Monitor drawdown
            self._check_max_drawdown(trade)
            
            # Check for unusual price movements
            self._check_price_anomalies(trade, current_price)
            
            # Log performance metrics
            self._log_trade_metrics(trade)
            
            time.sleep(60)  # Check every minute

4.4 Emergency Procedures

• System Failure: Automatic stop-loss conversion to market order
• Broker Disconnect: Hold all positions, reconnect and verify
• Unusual Price: Manual review before execution
• Margin Call: Immediately liquidate worst-performing positions
• Circuit Breaker: Halt trading if daily loss limit reached

5. Data Feed & Market Hours Standards

5.1 Approved Trading Hours

• XAUUSD Hours: Sunday 17:00 - Friday 16:00 GMT
• High Volatility Hours: 12:30-14:30 GMT (US market open)
• Low Liquidity Avoid: Friday 14:00-16:00 GMT (market close)
• Holiday Trading: NO trading on major holidays (Christmas, New Year, etc.)

5.2 Data Feed Requirements

• Price Feed: Real-time from MT5 with < 100ms latency
• Update Frequency: Ticks updated every 100-500ms
• Data Validation: Check for gaps, spikes, and anomalies
• Backup Feed: Alternative data source if primary fails
• Data Logging: Timestamp all price data for audit trail

5.3 Economic Calendar Integration

HIGH IMPACT NEWS - PAUSE TRADING 30 MINUTES BEFORE & AFTER:
- Non-Farm Payroll (1st Friday of month)
- FOMC Rate Decisions (8 times per year)
- Fed Chair Speeches (sporadic)
- ISM Manufacturing/Services

MEDIUM IMPACT - CAUTION TRADING:
- CPI/PPI Releases
- Treasury Auctions
- Unemployment Claims
- Retail Sales

6. Model Performance Validation

6.1 Strategy Performance Review

• Daily: Review P&L, win rate, drawdown
• Weekly: Analyze trade patterns, identify issues
• Monthly: Comprehensive performance audit
• Quarterly: Full strategy validation and recalibration

6.2 Signal Accuracy Metrics

• True Positive Rate: Profitable signals / Total signals
• False Positive Rate: Losing signals / Total signals
• Signal Frequency: Trades per week (target: 5-15)
• Signal Timing: Entry accuracy vs optimal entry points

6.3 Performance Degradation Detection

Retraining Triggers (Automatic):
- Win rate drops below 50% (consecutive 10 trades)
- Sharpe ratio falls below 1.5
- Monthly loss exceeds 10% of equity
- Strategy underperforms random entries for 5 consecutive days

Action: Stop trading, retrain model, run validation before resume

7. Audit & Compliance

7.1 Trade Audit Trail

• Every Trade Logged: Entry time, price, size, exit time, P&L
• Decision Log: Why each trade was taken (signals, reasoning)
• Modification Log: Any manual overrides or interventions
• Performance Attribution: Credit correct predictions, analyze errors

7.2 Compliance Checklist

• All trades have stop-loss orders
• Position sizes comply with equity rules
• Daily loss limits enforced
• No trades during prohibited hours
• Data reconciliation completed daily
• Model performance within expected parameters
• Risk metrics within tolerance
• Backup systems operational

7.3 Reporting Standards

• Daily Report: Generated automatically, sent to stakeholders
• Weekly Summary: Performance metrics, trades analysis
• Monthly Report: Detailed performance review, strategy metrics
• Quarterly Review: Strategy effectiveness, recommendations

8. Emergency Protocols

8.1 Margin Call Procedures

1. Immediately notify all stakeholders
2. Liquidate 50% of worst-performing positions
3. If still critical, liquidate all positions
4. Suspend trading until account restored to 2:1 margin

8.2 System Failure Recovery

1. Detect failure within 1 minute (automated)
2. Convert all pending orders to market orders
3. Activate backup MT5 connection
4. Reconnect within 5 minutes
5. Verify all positions after reconnection

8.3 Force Stop Procedures

• Daily Loss Limit Hit: Stop all new trades immediately
• Consecutive Losses (5): Manual review required before resuming
• System Error (2+): Halt trading for 24 hours, investigate
• Model Failure: Revert to previous validated version

---

Last Updated: January 2026 Version: 1.0 Maintained By: Trading Operations Team Review Frequency: Monthly