Metrics

import numpy as np
from scipy.stats import skew, kurtosis, anderson, shapiro

def compute_risk_metrics(portfolio_returns, portfolio_beta, rf_rate=0.0):
    """
    Compute various risk metrics for the portfolio.
    Note, that it works for RF and CPI Adjusted returns.
    """

    mean_return = portfolio_returns.mean() * 12
    std_dev = portfolio_returns.std(ddof=1) * np.sqrt(12)
    downside_returns = portfolio_returns[portfolio_returns < rf_rate]
    downside_dev = downside_returns.std(ddof=1) * np.sqrt(12)

    cumulative = (1 + portfolio_returns).cumprod()
    running_max = np.maximum.accumulate(cumulative)
    drawdowns = (cumulative - running_max) / running_max

    max_drawdown = drawdowns.min()
    cumulative_return = cumulative.iloc[-1] - 1
    calmar_ratio = mean_return / abs(max_drawdown) if max_drawdown != 0 else np.nan

    VaR_95 = np.percentile(portfolio_returns, 5)
    CVaR_95 = portfolio_returns[portfolio_returns <= VaR_95].mean()
    omega_ratio = (portfolio_returns[portfolio_returns > rf_rate].sum() /
                   abs(portfolio_returns[portfolio_returns < rf_rate].sum()))

    anderson_stat = anderson(portfolio_returns).statistic
    shapiro_stat, shapiro_p = shapiro(portfolio_returns)

    neg_deviation = abs(portfolio_returns[portfolio_returns < rf_rate] - rf_rate)
    sasr = mean_return / neg_deviation.mean() if not neg_deviation.empty else np.nan

    mad = np.mean(np.abs(portfolio_returns - portfolio_returns.mean()))
    iqr = np.percentile(portfolio_returns, 75) - np.percentile(portfolio_returns, 25)

    ulcer_index = np.sqrt(np.mean(drawdowns[drawdowns < 0] ** 2))

    sorted_drawdowns = drawdowns.sort_values()
    threshold_index = int(0.05 * len(sorted_drawdowns))
    cdar_5 = sorted_drawdowns[:threshold_index].mean() if threshold_index > 0 else np.nan

    pain_ratio = mean_return / abs(drawdowns.mean()) if drawdowns.mean() != 0 else np.nan

    return {
        'Annual Return': mean_return,
        'Annual Volatility': std_dev,
        'Sharpe Ratio': mean_return / std_dev,
        'Sortino Ratio': mean_return / downside_dev if downside_dev != 0 else np.nan,
        'Treynor Ratio': mean_return / portfolio_beta if portfolio_beta != 0 else np.nan,
        'Calmar Ratio': calmar_ratio,
        'Max Drawdown': max_drawdown,
        'VaR 95%': VaR_95,
        'CVaR 95%': CVaR_95,
        'Downside Deviation': downside_dev,
        'Omega Ratio': omega_ratio,
        'Skewness': skew(portfolio_returns),
        'Kurtosis': kurtosis(portfolio_returns),
        'Anderson-Darling Stat': anderson_stat,
        'Shapiro-Wilk Stat': shapiro_stat,
        'Shapiro-Wilk p': shapiro_p,
        'SASR': sasr,
        'MAD': mad,
        'IQR': iqr,
        'Ulcer Index': ulcer_index,
        'CDaR (5%)': cdar_5,
        'Pain Ratio': pain_ratio,
    }