feat: Add XGBoost ML Signal Quality Predictor

- New ml_signal_predictor.py with XGBoost model
- Feature extraction from historical trades (17 features)
- 5-fold cross-validation for robust training
- Integrated into enhanced_trading_check_wrapper as optional layer
- Disabled by default until model improves (AUC: 0.508)
- Key insight: Asian session is strongest predictor of success

Features used:
- Signal: confidence, threshold, regime_strength
- Session: asian/london/ny/overlap (one-hot)
- Time: hour (cyclical), day_of_week
- Quality: signal_quality score
- Direction: long/short

Usage:
- train_ml_model() to train
- enable_ml_predictor() to activate
- get_ml_status() for info

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

ml_integration.py

@@ -0,0 +1,169 @@
+#!/usr/bin/env python3
+"""
+ML SIGNAL PREDICTOR INTEGRATION
+Einfache Integration des ML Predictors in den Trading Bot
+
+VERWENDUNG IM NOTEBOOK:
+    from ml_integration import (
+        ml_predictor,
+        check_ml_signal,
+        train_ml_model,
+        get_ml_status,
+        enable_ml_predictor,
+        disable_ml_predictor,
+        ML_CONFIG
+    )
+
+    # Status anzeigen
+    print(get_ml_status())
+
+    # Model trainieren (falls genug Daten)
+    train_ml_model()
+
+    # In enhanced_trading_check_wrapper:
+    ml_result = check_ml_signal(signal_info)
+    if not ml_result['should_trade']:
+        print(f"TRADE BLOCKIERT durch ML Predictor!")
+        return None
+"""
+
+import logging
+from typing import Dict
+
+logger = logging.getLogger(__name__)
+
+# ==========================================
+# IMPORT ML PREDICTOR
+# ==========================================
+
+try:
+    from ml_signal_predictor import (
+        MLSignalPredictor,
+        ML_CONFIG,
+        get_ml_predictor,
+        check_ml_signal,
+        train_ml_model,
+        get_ml_status,
+        enable_ml_predictor,
+        disable_ml_predictor
+    )
+
+    # Initialize global predictor
+    ml_predictor = get_ml_predictor()
+
+    ML_AVAILABLE = True
+    print("ML Signal Predictor loaded successfully")
+
+except ImportError as e:
+    logger.warning(f"ML Signal Predictor not available: {e}")
+    ML_AVAILABLE = False
+
+    # Fallback functions
+    ML_CONFIG = {'enabled': False}
+    ml_predictor = None
+
+    def check_ml_signal(signal_info: Dict) -> Dict:
+        return {
+            'win_probability': 0.5,
+            'should_trade': True,
+            'lot_multiplier': 1.0,
+            'action': 'ALLOW',
+            'reason': 'ML not available'
+        }
+
+    def train_ml_model(force: bool = False) -> Dict:
+        return {'error': 'ML not available'}
+
+    def get_ml_status() -> str:
+        return "ML Signal Predictor: NOT AVAILABLE (install xgboost)"
+
+    def enable_ml_predictor():
+        print("ML Signal Predictor not available")
+
+    def disable_ml_predictor():
+        print("ML Signal Predictor not available")
+
+
+# ==========================================
+# INTEGRATION HELPER
+# ==========================================
+
+def check_ml_and_get_multiplier(signal_info: Dict, debug: bool = True) -> tuple:
+    """
+    Prüft ML Signal und gibt (should_trade, lot_multiplier, result) zurück
+
+    Für einfache Integration in enhanced_trading_check_wrapper
+
+    Returns:
+        (should_trade: bool, lot_multiplier: float, ml_result: Dict)
+    """
+    if not ML_AVAILABLE or not ML_CONFIG.get('enabled', False):
+        return True, 1.0, {'action': 'SKIP', 'reason': 'ML disabled'}
+
+    ml_result = check_ml_signal(signal_info)
+
+    if debug:
+        action_emoji = {
+            'ALLOW': '',
+            'CAUTION': '',
+            'REDUCE': '',
+            'BLOCK': ''
+        }.get(ml_result['action'], '')
+
+        print(f"\n{action_emoji} ML SIGNAL CHECK:")
+        print("-" * 50)
+        print(f"   Win Probability: {ml_result['win_probability']:.1%}")
+        print(f"   Action: {ml_result['action']}")
+        print(f"   Lot Multiplier: {ml_result['lot_multiplier']:.0%}")
+        print(f"   Reason: {ml_result['reason']}")
+        print("-" * 50)
+
+    return ml_result['should_trade'], ml_result['lot_multiplier'], ml_result
+
+
+# ==========================================
+# AUTO-TRAINING CHECK
+# ==========================================
+
+def auto_train_if_needed():
+    """
+    Prüft ob genug neue Trades für Retraining vorhanden sind
+    und trainiert automatisch falls nötig
+    """
+    if not ML_AVAILABLE or ml_predictor is None:
+        return
+
+    if ml_predictor.should_retrain():
+        print("\n Auto-Retraining triggered...")
+        train_ml_model()
+
+
+# ==========================================
+# QUICK STATUS
+# ==========================================
+
+def ml_quick_status() -> str:
+    """Kurzer Status für Startup-Output"""
+    if not ML_AVAILABLE:
+        return " ML Predictor: NOT INSTALLED (pip install xgboost)"
+
+    if not ML_CONFIG.get('enabled', False):
+        return " ML Predictor: DISABLED"
+
+    if ml_predictor and ml_predictor.is_trained:
+        stats = ml_predictor.training_stats
+        auc = stats.get('cv_auc_mean', 0)
+        trades = stats.get('total_trades', 0)
+        return f" ML Predictor: ACTIVE (AUC: {auc:.2f}, trained on {trades} trades)"
+    else:
+        return " ML Predictor: NOT TRAINED (run train_ml_model())"
+
+
+# ==========================================
+# STANDALONE TEST
+# ==========================================
+
+if __name__ == "__main__":
+    print(ml_quick_status())
+    print()
+    print(get_ml_status())