""" CCT Bitcoin Crash Prediction Validation Compares Black Hole/Red Star predictions against real Bitcoin events """ import numpy as np import pandas as pd import yfinance as yf # ============================================================ # KNOWN REAL-WORLD BTC EVENTS (from research) # ============================================================ REAL_EVENTS = [ { "name": "October 2025 Tariff Flash Crash", "date_start": "2025-09-15", # 3 weeks before ATH "date_end": "2025-10-15", "description": "ATH $126K Oct 6, then 100% China tariffs, $19B liquidations, BTC $126K→$110K", "severity": "HIGH" }, { "name": "October-November 2025 Bear Market", "date_start": "2025-10-01", "date_end": "2025-12-01", "description": "BTC falls below $90K, breaks 50-week MA, Death Cross Nov 16, $85K support breaks, $1.9B liquidated", "severity": "CRITICAL" }, { "name": "November 2025 Government Shutdown Liquidity Drought", "date_start": "2025-10-15", "date_end": "2025-12-15", "description": "43-day gov shutdown, TGA freeze, ETF outflows $3B, FinCEN vs Huione, DATCo implosion", "severity": "CRITICAL" }, { "name": "February 2026 Crash to $60K", "date_start": "2026-01-15", "date_end": "2026-02-15", "description": "BTC drops 52% from ATH, plunges to $60K, $2B liquidations, ETF outflows $272M/day", "severity": "EXTREME" }, { "name": "January 2024 ETF Launch Rally", "date_start": "2024-04-08", # Earliest data available "date_end": "2024-06-30", "description": "Post-ETF approval rally period, generally bullish uptrend", "severity": "BULLISH" }, { "name": "August 2025 ATH Run", "date_start": "2025-07-15", "date_end": "2025-09-01", "description": "BTC reaches ATH $124K+ on pro-crypto administration", "severity": "BULLISH" }, ] # ============================================================ # RUN CCT PREDICTIONS # ============================================================ print("🔬 CCT Bitcoin Prediction Validation") print("=" * 70) # Fetch BTC data print("\n📊 Fetching BTC-USD data (2y, 1d)...") stock = yf.Ticker("BTC-USD") df = stock.history(period="2y", interval="1d") prices = df['Close'].values dates = df.index print(f"✓ Retrieved {len(prices)} data points from {dates[0].strftime('%Y-%m-%d')} to {dates[-1].strftime('%Y-%m-%d')}") # Import and run CCT predictor import sys sys.path.insert(0, '/home/per/Documents/Universe as a Fractal') from ex01 import CCT_MarketCrash_Predictor predictor = CCT_MarketCrash_Predictor( trained_theta=0.7342, collapse_threshold=0.7, critical_escape_prob=0.2, window_size=14, prediction_horizon=5 ) # Run silently predictions = [] for idx in range(predictor.WINDOW, len(prices)): result = predictor.predict_event(prices, idx) predictions.append({ 'index': idx, 'date': dates[idx], 'prediction': result['prediction'], 'Psi_Red': result['Psi_Red'], 'confidence': result['confidence'] }) pred_df = pd.DataFrame(predictions) # ============================================================ # VALIDATE AGAINST REAL EVENTS # ============================================================ print("\n" + "=" * 70) print("📋 VALIDATION: CCT Predictions vs Real BTC Events") print("=" * 70) hits = 0 misses = 0 false_alarms = 0 for event in REAL_EVENTS: start = pd.Timestamp(event["date_start"], tz="UTC") end = pd.Timestamp(event["date_end"], tz="UTC") # Find predictions in this window mask = (pred_df['date'] >= start) & (pred_df['date'] <= end) event_preds = pred_df[mask] if len(event_preds) == 0: print(f"\n⚠️ NO DATA for: {event['name']}") print(f" Period: {event['date_start']} to {event['date_end']}") print(f" Real event: {event['description']}") misses += 1 continue black_hole_count = len(event_preds[event_preds['prediction'] == 'BLACK_HOLE']) red_star_count = len(event_preds[event_preds['prediction'] == 'RED_STAR']) total = len(event_preds) bh_pct = black_hole_count / total * 100 avg_psi = event_preds['Psi_Red'].mean() min_psi = event_preds['Psi_Red'].min() print(f"\n{'─' * 70}") print(f"📌 EVENT: {event['name']}") print(f" Period: {event['date_start']} to {event['date_end']}") print(f" Real event: {event['description']}") print(f" Severity: {event['severity']}") print(f" CCT Predictions: {total} days analyzed") print(f" Black Hole: {black_hole_count} ({bh_pct:.0f}%)") print(f" Red Star: {red_star_count} ({100-bh_pct:.0f}%)") print(f" Avg Ψ_Red: {avg_psi:.6f}") print(f" Min Ψ_Red: {min_psi:.6f}") # Classification logic if event['severity'] in ['CRITICAL', 'EXTREME', 'HIGH']: # These are crash events - should show Black Hole dominance if bh_pct > 50: print(f" ✅ HIT: CCT correctly predicted crash risk (BH: {bh_pct:.0f}%)") hits += 1 elif bh_pct > 25: print(f" ⚠️ PARTIAL: Mixed signal (BH: {bh_pct:.0f}%) - some warning") hits += 0.5 else: print(f" ❌ MISS: CCT failed to predict crash (BH: {bh_pct:.0f}%)") misses += 1 elif event['severity'] == 'BULLISH': # These are bullish events - should show Red Star dominance if bh_pct < 30: print(f" ✅ HIT: CCT correctly showed stability (BH: {bh_pct:.0f}%)") hits += 1 elif bh_pct < 60: print(f" ⚠️ PARTIAL: Mixed signal during rally (BH: {bh_pct:.0f}%)") hits += 0.5 else: print(f" ❌ FALSE ALARM: CCT predicted crash during rally (BH: {bh_pct:.0f}%)") false_alarms += 1 # ============================================================ # OVERALL ACCURACY # ============================================================ total_events = len(REAL_EVENTS) print(f"\n{'=' * 70}") print(f"📊 OVERALL ACCURACY") print(f"{'=' * 70}") print(f" Total Events Tested: {total_events}") print(f" Hits: {hits}") print(f" Misses: {misses}") print(f" False Alarms: {false_alarms}") print(f" Accuracy: {hits/total_events*100:.1f}%") print(f" Crash Detection Rate: {hits/(hits+misses)*100:.1f}%" if (hits+misses) > 0 else " N/A") # Compute returns for threshold analysis returns = np.concatenate([[0], np.diff(prices) / prices[:-1]]) # ============================================================ # PSI_RED THRESHOLD ANALYSIS # ============================================================ print(f"\n{'=' * 70}") print(f"📈 Ψ_Red THRESHOLD ANALYSIS (Daily Drop Detection)") print(f"{'=' * 70}") # Align predictions with actual returns # predictions start at index WINDOW, returns start at index 1 window = predictor.WINDOW pred_returns = returns[window:window+len(pred_df)] # Same length as predictions all_psi = pred_df['Psi_Red'].values for threshold_pct in [0.002, 0.003, 0.004, 0.005, 0.008, 0.01]: # Low Psi = Black Hole (crash risk), High Psi = Red Star (stable) predicted_crash = all_psi < threshold_pct actual_drop = np.abs(pred_returns) > 0.02 # 2%+ moves true_pos = np.sum(predicted_crash & (pred_returns < -0.02)) false_pos = np.sum(predicted_crash & (pred_returns >= -0.02)) false_neg = np.sum(~predicted_crash & (pred_returns < -0.02)) true_neg = np.sum(~predicted_crash & (pred_returns >= -0.02)) precision = true_pos / (true_pos + false_pos) if (true_pos + false_pos) > 0 else 0 recall = true_pos / (true_pos + false_neg) if (true_pos + false_neg) > 0 else 0 f1 = 2 * precision * recall / (precision + recall) if (precision + recall) > 0 else 0 print(f"\n Ψ_Red threshold < {threshold_pct:.4f} → Black Hole") print(f" True Positives: {true_pos:4d} (correctly predicted >2% drops)") print(f" False Positives: {false_pos:4d} (predicted crash, none occurred)") print(f" False Negatives: {false_neg:4d} (missed >2% drops)") print(f" Precision: {precision:.1%} | Recall: {recall:.1%} | F1: {f1:.1%}")