File size: 11,060 Bytes
c5ef85d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 | import pandas as pd
import numpy as np
from scipy import stats
import sys
def calculate_rolling_metrics(input_file, output_file):
try:
print(f"Reading {input_file}...")
df = pd.read_csv(input_file)
# 1. Data Preprocessing
# Ensure Time_deal is datetime
df['Time_deal'] = pd.to_datetime(df['Time_deal'])
# Sort by time to ensure rolling calculations are correct
df = df.sort_values('Time_deal').reset_index(drop=True)
# Calculate 'Trade Return' for each row
# Balance in the CSV is usually the balance *after* the deal.
# Previous Balance = Current Balance - Profit
df['Prev_Balance'] = df['Balance'] - df['Profit']
# Avoid division by zero
df['Trade_Return'] = np.where(
df['Prev_Balance'] > 0,
df['Profit'] / df['Prev_Balance'],
0.0
)
# 2. Define Metric Columns (using requested format)
metric_map = {
'Sharpe': 'rolling_Sharpe_Ratio',
'Sortino': 'rolling_Sortino_Ratio',
'Calmar': 'rolling_Calmar_Ratio',
'Stability': 'rolling_Stability',
'Recovery Factor': 'rolling_Recovery_Factor',
'omega ratio': 'rolling_Omega_Ratio',
'skew': 'rolling_Skewness',
'kurtosis': 'rolling_Kurtosis',
'tail ratio': 'rolling_Tail_Ratio',
'alpha': 'rolling_Alpha',
'beta': 'rolling_Beta',
'Common Sense Ratio': 'rolling_Common_Sense_Ratio',
'volatility': 'rolling_Volatility',
'Kelly Criterion': 'rolling_Kelly_Criterion',
'System Quality Number (SQN)': 'rolling_SQN_System_Quality_Number',
'K-Ratio': 'rolling_K_Ratio',
'R-Squared': 'rolling_R_Squared',
'CPC Index': 'rolling_CPC_Index',
'VaR': 'rolling_VaR_Value_at_Risk',
'CVaR': 'rolling_CVaR_Conditional_Value_at_Risk',
'return standard deviation': 'rolling_Return_Standard_Deviation',
'AHPR': 'rolling_AHPR_Average_Holding_Period_Return',
'GHPR': 'rolling_GHPR_Geometric_Holding_Period_Return',
'Drawdown duration': 'rolling_Drawdown_Duration',
'Maximum drawdown duration': 'rolling_Max_Drawdown_Duration',
'Time-weighted return (TWR)': 'rolling_TWR_Time_Weighted_Return',
'Money-weighted return (MWR / IRR)': 'rolling_MWR_Money_Weighted_Return',
'Ulcer Index': 'rolling_Ulcer_Index',
'MAE': 'rolling_MAE_Max_Adverse_Excursion',
'MFE': 'rolling_MFE_Max_Favorable_Excursion',
'MAR Ratio': 'rolling_MAR_Ratio',
'Information Ratio': 'rolling_Information_Ratio',
'Treynor Ratio': 'rolling_Treynor_Ratio',
'Tracking Error': 'rolling_Tracking_Error',
'Active Share': 'rolling_Active_Share'
}
# Initialize new columns with NaN
for col in metric_map.values():
df[col] = np.nan
# 3. Expanding Window Calculation
print("Calculating metrics (this may take a moment)...")
# Pre-convert columns to numpy arrays for speed in loop
returns = df['Trade_Return'].values
profits = df['Profit'].values
balances = df['Balance'].values
times = df['Time_deal'].values
n = len(df)
# State variables for expensive tracking
max_dd_duration_sec = 0.0
for i in range(n):
# Slices for expanding window (0 to i)
hist_rets = returns[:i+1]
hist_profits = profits[:i+1]
hist_bal = balances[:i+1]
current_time = times[i]
# --- Basic Return Stats ---
if len(hist_rets) > 1:
mean_ret = np.mean(hist_rets)
std_ret = np.std(hist_rets, ddof=1)
else:
mean_ret = hist_rets[0]
std_ret = 0.0
# Volatility
df.at[i, metric_map['volatility']] = std_ret
df.at[i, metric_map['return standard deviation']] = std_ret
# Sharpe (Assuming Risk Free = 0, Simple Trade Sharpe)
if std_ret > 1e-9:
df.at[i, metric_map['Sharpe']] = mean_ret / std_ret
else:
df.at[i, metric_map['Sharpe']] = 0.0
# Sortino
neg_rets = hist_rets[hist_rets < 0]
if len(neg_rets) > 1:
down_std = np.std(neg_rets, ddof=1)
if down_std > 1e-9:
df.at[i, metric_map['Sortino']] = mean_ret / down_std
# Skew / Kurtosis
if len(hist_rets) > 2:
df.at[i, metric_map['skew']] = stats.skew(hist_rets)
df.at[i, metric_map['kurtosis']] = stats.kurtosis(hist_rets)
# --- Equity Curve Stats (Drawdown) ---
cum_max = np.maximum.accumulate(hist_bal)
drawdowns = (cum_max - hist_bal) / cum_max
max_dd = np.max(drawdowns)
# Ulcer Index
if len(drawdowns) > 0:
df.at[i, metric_map['Ulcer Index']] = np.sqrt(np.mean(drawdowns**2))
# Calmar / MAR Ratio (Approximated with simple CAGR)
# Calculate years elapsed
elapsed_seconds = (current_time - times[0]).astype('timedelta64[s]').astype(float)
years = elapsed_seconds / (365 * 24 * 3600)
total_return = (hist_bal[-1] / (hist_bal[0] - hist_profits[0])) - 1 if (hist_bal[0] - hist_profits[0]) > 0 else 0
cagr = 0
if years > 0:
# Handle negative base for power
if total_return > -1:
cagr = (1 + total_return) ** (1 / years) - 1
if max_dd > 0:
df.at[i, metric_map['Calmar']] = cagr / max_dd
df.at[i, metric_map['MAR Ratio']] = cagr / max_dd
# Recovery Factor (Net Profit / Max Drawdown Amount)
net_profit = np.sum(hist_profits)
dd_amounts = cum_max - hist_bal
max_dd_amt = np.max(dd_amounts)
if max_dd_amt > 0:
df.at[i, metric_map['Recovery Factor']] = net_profit / max_dd_amt
# Drawdown Duration
# Current DD duration: Time since last High Water Mark
hwm_idx = np.argmax(hist_bal)
current_dd_duration = (current_time - times[hwm_idx]).astype('timedelta64[s]').astype(float)
df.at[i, metric_map['Drawdown duration']] = current_dd_duration
# Update Max Drawdown Duration
max_dd_duration_sec = max(max_dd_duration_sec, current_dd_duration)
df.at[i, metric_map['Maximum drawdown duration']] = max_dd_duration_sec
# Stability (R-Squared of Equity Log Linearity)
if len(hist_bal) > 2:
try:
# Log of equity (handle negatives/zeros)
y = np.log(np.abs(hist_bal) + 1e-9)
x = np.arange(len(y))
slope, intercept, r_val, p_val, std_err = stats.linregress(x, y)
df.at[i, metric_map['Stability']] = r_val ** 2
# K-Ratio (Slope / StdErr of equity curve)
slope_k, _, _, _, std_err_k = stats.linregress(x, hist_bal)
if std_err_k > 0:
df.at[i, metric_map['K-Ratio']] = slope_k / std_err_k
except:
pass
# --- Trade Stats ---
wins = hist_profits[hist_profits > 0]
losses = np.abs(hist_profits[hist_profits < 0])
# Omega Ratio
if np.sum(losses) > 0:
df.at[i, metric_map['omega ratio']] = np.sum(wins) / np.sum(losses)
# Kelly & CPC
if len(wins) > 0 and len(losses) > 0:
win_rate = len(wins) / len(hist_profits)
avg_win = np.mean(wins)
avg_loss = np.mean(losses)
if avg_loss > 0:
b_ratio = avg_win / avg_loss
# Kelly = p - q/b
df.at[i, metric_map['Kelly Criterion']] = win_rate - (1 - win_rate) / b_ratio
# CPC Index = ProfitFactor * WinRate * PayoffRatio
pf = np.sum(wins) / np.sum(losses)
df.at[i, metric_map['CPC Index']] = pf * win_rate * b_ratio
# SQN
if len(hist_profits) > 1:
std_profit = np.std(hist_profits, ddof=1)
if std_profit > 0:
sqn = np.sqrt(len(hist_profits)) * np.mean(hist_profits) / std_profit
df.at[i, metric_map['System Quality Number (SQN)']] = sqn
# Tail Ratio & VaR
if len(hist_rets) > 10:
t_95 = np.percentile(hist_rets, 95)
t_05 = np.abs(np.percentile(hist_rets, 5))
if t_05 > 0:
df.at[i, metric_map['tail ratio']] = t_95 / t_05
# Common Sense Ratio = Profit Factor * Tail Ratio
pf = np.sum(wins)/np.sum(losses) if np.sum(losses) > 0 else 0
df.at[i, metric_map['Common Sense Ratio']] = pf * (t_95 / t_05)
# VaR (5%)
var_val = np.percentile(hist_rets, 5)
df.at[i, metric_map['VaR']] = var_val
# CVaR (Mean of returns <= VaR)
cvar_vals = hist_rets[hist_rets <= var_val]
if len(cvar_vals) > 0:
df.at[i, metric_map['CVaR']] = np.mean(cvar_vals)
# AHPR / GHPR / TWR
df.at[i, metric_map['AHPR']] = np.mean(1 + hist_rets)
if np.all((1 + hist_rets) > 0):
df.at[i, metric_map['GHPR']] = stats.gmean(1 + hist_rets)
df.at[i, metric_map['Time-weighted return (TWR)']] = np.prod(1 + hist_rets) - 1
# 4. Cleanup & Save
# Drop temporary calculation columns
df = df.drop(columns=['Prev_Balance', 'Trade_Return'])
print(f"Saving to {output_file}...")
df.to_csv(output_file, index=False)
print("Done.")
except FileNotFoundError:
print(f"Error: The file '{input_file}' was not found.")
except Exception as e:
print(f"An error occurred: {e}")
if __name__ == "__main__":
calculate_rolling_metrics(
'merged_extracted_orders_and_deals.csv',
'7_layer_output.csv'
) |