trading_bot_v4/scripts/diagnostic_sweep.py

#!/usr/bin/env python3
"""
Diagnostic Parameter Sweep - Verify Parameters Actually Control Behavior
Purpose: Determine if parameter insensitivity is a bug or reality
"""
from __future__ import annotations

import argparse
import sys
from pathlib import Path
from typing import Dict, List, Tuple

PROJECT_ROOT = Path(__file__).resolve().parents[1]
if str(PROJECT_ROOT) not in sys.path:
    sys.path.append(str(PROJECT_ROOT))

import pandas as pd
from pathlib import Path
from tqdm import tqdm
from backtester.data_loader import load_csv
from backtester.indicators.money_line import MoneyLineInputs, money_line_signals
from backtester.simulator import TradeConfig, simulate_money_line


def test_parameter_impact(csv_path: Path, symbol: str) -> Dict[str, List[Tuple[float, int, float]]]:
    """
    Test if each parameter actually changes behavior.
    Returns: {param_name: [(param_value, num_signals, total_pnl), ...]}
    """
    print("Loading data...")
    data_slice = load_csv(Path(csv_path), symbol, "5m")
    df = data_slice.data
    print(f"Loaded {len(df)} bars")
    
    baseline = MoneyLineInputs()
    config = TradeConfig(position_size=8100.0, max_bars_per_trade=288)
    
    results = {}
    
    # Test 1: flip_threshold_percent (should dramatically affect signal count)
    print("\n" + "="*60)
    print("TEST 1: flip_threshold_percent (0.4, 0.5, 0.6, 0.7)")
    print("Expected: LOWER threshold = MORE signals")
    print("="*60)
    flip_results = []
    for val in tqdm([0.4, 0.5, 0.6, 0.7], desc="Testing flip_threshold"):
        inputs = MoneyLineInputs(flip_threshold_percent=val)
        result = simulate_money_line(df, symbol, inputs, config)
        flip_results.append((val, len(result.trades), result.total_pnl))
        print(f"flip_threshold={val:.1f}: {len(result.trades)} trades, ${result.total_pnl:.2f} PnL")
    results['flip_threshold_percent'] = flip_results
    
    # Test 2: cooldown_bars (should affect signal frequency)
    print("\n" + "="*60)
    print("TEST 2: cooldown_bars (1, 2, 3, 4)")
    print("Expected: LOWER cooldown = MORE signals")
    print("="*60)
    cooldown_results = []
    for val in tqdm([1, 2, 3, 4], desc="Testing cooldown_bars"):
        inputs = MoneyLineInputs(cooldown_bars=val)
        result = simulate_money_line(df, symbol, inputs, config)
        cooldown_results.append((val, len(result.trades), result.total_pnl))
        print(f"cooldown_bars={val}: {len(result.trades)} trades, ${result.total_pnl:.2f} PnL")
    results['cooldown_bars'] = cooldown_results
    
    # Test 3: momentum_min_adx (should filter signals)
    print("\n" + "="*60)
    print("TEST 3: momentum_min_adx (18, 21, 24, 27)")
    print("Expected: HIGHER ADX = FEWER signals")
    print("="*60)
    adx_results = []
    for val in tqdm([18.0, 21.0, 24.0, 27.0], desc="Testing momentum_min_adx"):
        inputs = MoneyLineInputs(momentum_min_adx=val)
        result = simulate_money_line(df, symbol, inputs, config)
        adx_results.append((val, len(result.trades), result.total_pnl))
        print(f"momentum_min_adx={val:.1f}: {len(result.trades)} trades, ${result.total_pnl:.2f} PnL")
    results['momentum_min_adx'] = adx_results
    
    # Test 4: ma_gap_threshold (should affect signal generation)
    print("\n" + "="*60)
    print("TEST 4: ma_gap_threshold (0.2, 0.3, 0.4, 0.5)")
    print("Expected: Different signal counts")
    print("="*60)
    gap_results = []
    for val in tqdm([0.2, 0.3, 0.4, 0.5], desc="Testing ma_gap_threshold"):
        inputs = MoneyLineInputs(ma_gap_threshold=val)
        result = simulate_money_line(df, symbol, inputs, config)
        gap_results.append((val, len(result.trades), result.total_pnl))
        print(f"ma_gap_threshold={val:.1f}: {len(result.trades)} trades, ${result.total_pnl:.2f} PnL")
    results['ma_gap_threshold'] = gap_results
    
    return results


def analyze_results(results: Dict[str, List[Tuple[float, int, float]]]):
    """Determine if parameters actually affect behavior."""
    print("\n" + "="*80)
    print("DIAGNOSTIC ANALYSIS")
    print("="*80)
    
    for param_name, values in results.items():
        signal_counts = [x[1] for x in values]
        pnls = [x[2] for x in values]
        
        # Check if all identical (parameter has NO effect)
        if len(set(signal_counts)) == 1 and len(set(pnls)) == 1:
            print(f"\n🔴 {param_name}: NO EFFECT - All configs produce identical results!")
            print(f"   All {signal_counts[0]} trades, ${pnls[0]:.2f} PnL")
            print(f"   ⚠️  Parameter is NOT being applied or is overridden")
        else:
            # Parameter has some effect
            min_signals, max_signals = min(signal_counts), max(signal_counts)
            min_pnl, max_pnl = min(pnls), max(pnls)
            range_pct = ((max_signals - min_signals) / min_signals * 100) if min_signals > 0 else 0
            
            print(f"\n✅ {param_name}: HAS EFFECT")
            print(f"   Signal range: {min_signals}-{max_signals} ({range_pct:.1f}% variation)")
            print(f"   PnL range: ${min_pnl:.2f} to ${max_pnl:.2f}")


def test_extreme_configs(csv_path: Path, symbol: str):
    """Test extreme parameter combinations to verify they produce different results."""
    print("\n" + "="*80)
    print("EXTREME CONFIGURATION TEST")
    print("="*80)
    
    df = load_csv(csv_path)
    config = TradeConfig(position_size=8100.0, max_bars_per_trade=288)
    
    # Ultra-loose config (should generate MANY signals)
    loose = MoneyLineInputs(
        flip_threshold_percent=0.3,
        cooldown_bars=1,
        momentum_min_adx=15.0,
        ma_gap_threshold=0.1,
        momentum_spacing=2,
        momentum_cooldown=1
    )
    
    # Ultra-strict config (should generate FEW signals)
    strict = MoneyLineInputs(
        flip_threshold_percent=0.8,
        cooldown_bars=5,
        momentum_min_adx=30.0,
        ma_gap_threshold=0.6,
        momentum_spacing=6,
        momentum_cooldown=5
    )
    
    # Baseline
    baseline = MoneyLineInputs()
    
    print("\n🔹 ULTRA-LOOSE config:")
    loose_result = simulate_money_line(df, symbol, loose, config)
    print(f"   Trades: {len(loose_result.trades)}, PnL: ${loose_result.total_pnl:.2f}")
    
    print("\n🔹 BASELINE config:")
    baseline_result = simulate_money_line(df, symbol, baseline, config)
    print(f"   Trades: {len(baseline_result.trades)}, PnL: ${baseline_result.total_pnl:.2f}")
    
    print("\n🔹 ULTRA-STRICT config:")
    strict_result = simulate_money_line(df, symbol, strict, config)
    print(f"   Trades: {len(strict_result.trades)}, PnL: ${strict_result.total_pnl:.2f}")
    
    if len(loose_result.trades) == len(baseline_result.trades) == len(strict_result.trades):
        print("\n🔴 CRITICAL BUG: All three configs produce IDENTICAL trade counts!")
        print("   Parameter system is completely broken.")
        return False
    else:
        print("\n✅ Configs produce different results - parameter system works")
        print(f"   Variation: {len(strict_result.trades)} to {len(loose_result.trades)} trades")
        return True


def main():
    parser = argparse.ArgumentParser(description="Diagnostic parameter testing")
    parser.add_argument("--csv", type=Path, required=True, help="Path to OHLCV CSV")
    parser.add_argument("--symbol", default="SOL-PERP", help="Symbol name")
    args = parser.parse_args()
    
    print("="*80)
    print("V9 PARAMETER DIAGNOSTIC SUITE")
    print("="*80)
    print(f"Data: {args.csv}")
    print(f"Symbol: {args.symbol}")
    
    # Test 1: Individual parameter impact
    results = test_parameter_impact(args.csv, args.symbol)
    analyze_results(results)
    
    # Test 2: Extreme configs
    system_works = test_extreme_configs(args.csv, args.symbol)
    
    print("\n" + "="*80)
    print("CONCLUSION")
    print("="*80)
    if system_works:
        print("✅ Parameter system is functional")
        print("   The parameter insensitivity in the sweep may indicate:")
        print("   1. The sweep grid didn't explore extreme enough values")
        print("   2. Parameters have non-linear interactions")
        print("   3. Core edge comes from EMA logic, not parameter tuning")
    else:
        print("🔴 Parameter system is BROKEN - parameters don't affect behavior")
        print("   This explains why the sweep found 100+ identical results")
        print("   Fix required before optimization can proceed")
    
    print("\n" + "="*80)


if __name__ == "__main__":
    main()