analyze_zero_fee.py

import json
import glob
import os
import pandas as pd

LOG_DIR = "logs/whale_paper_trader"

def load_recent_trades(n=10):
    files = glob.glob(os.path.join(LOG_DIR, "trades_*.json"))
    files.sort(key=os.path.getmtime, reverse=True)
    recent_files = files[:n]
    
    all_trades = []
    for f in recent_files:
        try:
            with open(f, 'r') as file:
                data = json.load(file)
                trades = data.get('trades', [])
                all_trades.extend(trades)
        except:
            pass
    return all_trades

def analyze_zero_fee(trades):
    if not trades:
        print("❌ No trades to analyze.")
        return

    df = pd.DataFrame(trades)
    
    # In the logs, 'pnl_pct' is typically the Gross PnL % (Leverage Adjusted) BEFORE fees 
    # (Because net_pnl_usdt is separate). 
    # Let's Verify: Trade 84: pnl_pct 0.54, fee 3.92, net -3.37.
    # 0.5% profit on $100 is $0.50. So pnl_pct IS Gross.
    
    print(f"\n🚀 dYdX ZERO FEE SIMULATION (Based on recent 10 sessions)")
    print("=" * 60)
    
    # 1. Total Potential
    total_gross_pnl = df['pnl_pct'].sum()
    gross_win_rate = (df['pnl_pct'] > 0).mean() * 100
    
    print(f"   Total Trades Considered: {len(df)}")
    print(f"   Gross Win Rate:          {gross_win_rate:.1f}%")
    print(f"   Total Theoretical Profit: +{total_gross_pnl:.1f}% (Accumulated PnL%)")
    print(f"   Avg Profit per Trade:    +{df['pnl_pct'].mean():.3f}%")
    
    # 2. Strategy Analysis (Which entry logic is best?)
    def get_dominant_strategy(probs):
        if not isinstance(probs, dict) or not probs: return "UNKNOWN"
        return max(probs, key=probs.get)
        
    df['dom_strategy'] = df['strategy_probs'].apply(get_dominant_strategy)
    
    print(f"\n🏆 BEST STRATEGIES (Entries)")
    strat_stats = df.groupby('dom_strategy').agg({
        'pnl_pct': ['count', 'sum', 'mean', lambda x: (x > 0).mean() * 100]
    })
    strat_stats.columns = ['Count', 'Total PnL%', 'Avg PnL%', 'Win Rate%']
    # Filter reasonable sample size
    strat_stats = strat_stats[strat_stats['Count'] > 2].sort_values('Total PnL%', ascending=False)
    print(strat_stats)
    
    # 3. Scalping Viability (Duration Analysis)
    print(f"\n⚡ SCALPING VIABILITY (Duration)")
    df['duration_bin'] = pd.cut(df['hold_seconds'], bins=[0, 10, 60, 300, 99999], 
                                labels=['Ultra-Scalp (<10s)', 'Scalp (10s-1m)', 'Swing (1m-5m)', 'Hold (>5m)'])
    
    dur_stats = df.groupby('duration_bin', observed=False).agg({
        'pnl_pct': ['count', 'sum', 'mean', lambda x: (x > 0).mean() * 100]
    })
    dur_stats.columns = ['Count', 'Total PnL%', 'Avg PnL%', 'Win Rate%']
    print(dur_stats)

    # 4. Exit Logic (Management)
    print(f"\n🚪 EXIT STYLE PERFORMANCE")
    exit_stats = df.groupby('status').agg({
        'pnl_pct': ['count', 'sum', 'mean', lambda x: (x > 0).mean() * 100]
    })
    exit_stats.columns = ['Count', 'Total PnL%', 'Avg PnL%', 'Win Rate%']
    print(exit_stats.sort_values('Total PnL%', ascending=False))
    
    print("\n💡 CONCLUSION FOR dYdX:")
    best_dur = dur_stats['Avg PnL%'].idxmax()
    best_strat = strat_stats.index[0] if not strat_stats.empty else "N/A"
    
    if dur_stats.loc['Ultra-Scalp (<10s)', 'Avg PnL%'] > 0:
        print(f"   ✅ YES! <10s Scalping is profitable!")
        print(f"      Avg Trade: +{dur_stats.loc['Ultra-Scalp (<10s)', 'Avg PnL%']:.3f}%")
        print(f"      Win Rate:  {dur_stats.loc['Ultra-Scalp (<10s)', 'Win Rate%']:.1f}%")
        print("      -> With 0 fees, you can run the bot 'as is' and make money.")
    else:
        print(f"   ⚠️ Even with 0 fees, ultra-scalping might be risky (Avg PnL negative).")
        
    print(f"   💪 Best Approach: {best_strat} holding for {best_dur}")

if __name__ == "__main__":
    analyze_zero_fee(load_recent_trades(10))