simulate_large_matrix.py

import pandas as pd
import numpy as np

def calculate_adx(df, period=14):
    df['h_l'] = df['high'] - df['low']
    df['h_pc'] = (df['high'] - df['close'].shift(1)).abs()
    df['l_pc'] = (df['low'] - df['close'].shift(1)).abs()
    df['tr'] = df[['h_l', 'h_pc', 'l_pc']].max(axis=1)
    df['+dm'] = np.where((df['high'] - df['high'].shift(1)) > (df['low'].shift(1) - df['low']), np.maximum(df['high'] - df['high'].shift(1), 0), 0)
    df['-dm'] = np.where((df['low'].shift(1) - df['low']) > (df['high'] - df['high'].shift(1)), np.maximum(df['low'].shift(1) - df['low'], 0), 0)
    alpha = 1 / period
    df['tr_smooth'] = df['tr'].ewm(alpha=alpha, adjust=False).mean() * period
    df['+dm_smooth'] = df['+dm'].ewm(alpha=alpha, adjust=False).mean() * period
    df['-dm_smooth'] = df['-dm'].ewm(alpha=alpha, adjust=False).mean() * period
    df['+di'] = 100 * (df['+dm_smooth'] / df['tr_smooth'])
    df['-di'] = 100 * (df['-dm_smooth'] / df['tr_smooth'])
    df['dx'] = 100 * (df['+di'] - df['-di']).abs() / (df['+di'] + df['-di']).replace(0, 1)
    df['adx'] = df['dx'].ewm(alpha=alpha, adjust=False).mean()
    return df

print("Loading 4-Months of tickflow data...")
m1 = pd.read_parquet('tickflow_M1.parquet')
m5 = pd.read_parquet('tickflow_M5.parquet')
m15 = pd.read_parquet('tickflow_M15.parquet')

# Indices are already DatetimeIndex 'dt'
print("Calculating ADX for massive dataset...")
m1 = calculate_adx(m1, 14)
m5 = calculate_adx(m5, 14)
m15 = calculate_adx(m15, 14)

df = m1[['open', 'high', 'low', 'close', 'adx', '+di', '-di']].copy()
df.rename(columns={'adx': 'adx_M1', '+di': 'di_plus_M1', '-di': 'di_minus_M1'}, inplace=True)

m5_cols = m5[['adx', '+di', '-di']].rename(columns={'adx': 'adx_M5', '+di': 'di_plus_M5', '-di': 'di_minus_M5'})
df = df.join(m5_cols, how='left').ffill()

m15_cols = m15[['adx', '+di', '-di']].rename(columns={'adx': 'adx_M15', '+di': 'di_plus_M15', '-di': 'di_minus_M15'})
df = df.join(m15_cols, how='left').ffill()
df.dropna(inplace=True)

print(f"Total Rows Ready: {len(df)}")

t = 18.0
a1 = (df['adx_M1'] >= t).astype(int)
a5 = (df['adx_M5'] >= t).astype(int)
a15 = (df['adx_M15'] >= t).astype(int)

# Correct EA Mapping
df['phase'] = 1
df.loc[(a15==0) & (a5==0) & (a1==1), 'phase'] = 2
df.loc[(a15==0) & (a5==1) & (a1==0), 'phase'] = 3
df.loc[(a15==1) & (a5==0) & (a1==0), 'phase'] = 4
df.loc[(a15==0) & (a5==1) & (a1==1), 'phase'] = 5
df.loc[(a15==1) & (a5==0) & (a1==1), 'phase'] = 6
df.loc[(a15==1) & (a5==1) & (a1==0), 'phase'] = 7
df.loc[(a15==1) & (a5==1) & (a1==1), 'phase'] = 8

df['phase_prev'] = df['phase'].shift(1)
df['phase_transition'] = df['phase'] != df['phase_prev']
df['dmi_dir_m15'] = np.where(df['di_plus_M15'] > df['di_minus_M15'], 1, -1)
df['dmi_dir_m5'] = np.where(df['di_plus_M5'] > df['di_minus_M5'], 1, -1)

results = []
tp_pips = 20.0
sl_pips = 20.0

print("Scanning 64 possible transitions over 4 months...")
for p_prev in range(1, 9):
    for p_curr in range(1, 9):
        if p_prev == p_curr: continue
        
        mask = (df['phase_transition']) & (df['phase_prev'] == p_prev) & (df['phase'] == p_curr)
        indices = np.where(mask)[0]
        if len(indices) < 20: continue # minimum threshold for stats reliability
        
        wins_trend = 0
        wins_fade = 0
        for i in indices:
            if i > len(df) - 60: continue
            
            entry = df['close'].iloc[i]
            # determine the primary active direction
            if p_prev >= 5 or p_curr >= 5:
                dir_mult = df['dmi_dir_m15'].iloc[i]
            elif p_curr in [3, 4] or p_prev in [3, 4]:
                dir_mult = df['dmi_dir_m5'].iloc[i]
            else:
                dir_mult = 1 if df['close'].iloc[i] < df['open'].iloc[i] else -1 # Sideways fade
                
            future_highs = df['high'].iloc[i+1:i+61].values
            future_lows = df['low'].iloc[i+1:i+61].values
            
            # Trend Check
            wt = 0
            if dir_mult == 1:
                for h, l in zip(future_highs, future_lows):
                    if h >= entry + (tp_pips * 0.01): wt=1; break
                    if l <= entry - (sl_pips * 0.01): break
            else:
                for h, l in zip(future_highs, future_lows):
                    if l <= entry - (tp_pips * 0.01): wt=1; break
                    if h >= entry + (sl_pips * 0.01): break
            wins_trend += wt
            
            # Fade Check
            wf = 0
            dir_fade = -dir_mult
            if dir_fade == 1:
                for h, l in zip(future_highs, future_lows):
                    if h >= entry + (tp_pips * 0.01): wf=1; break
                    if l <= entry - (sl_pips * 0.01): break
            else:
                for h, l in zip(future_highs, future_lows):
                    if l <= entry - (tp_pips * 0.01): wf=1; break
                    if h >= entry + (sl_pips * 0.01): break
            wins_fade += wf
            
        count = len(indices)
        wr_trend = (wins_trend / count) * 100
        wr_fade = (wins_fade / count) * 100
        
        results.append({
            'Trans': f'P{p_prev} -> P{p_curr}',
            'Count': count,
            'Trend_WR': round(wr_trend, 1),
            'Fade_WR': round(wr_fade, 1)
        })

res_df = pd.DataFrame(results)
res_df = res_df.sort_values('Count', ascending=False)
with open('out_large_matrix.txt', 'w') as f:
    f.write(res_df.to_string(index=False))
print("Done! Saved to out_large_matrix.txt")