voly 0.0.144__tar.gz → 0.0.146__tar.gz

{voly-0.0.144/src/voly.egg-info → voly-0.0.146}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: voly
-Version: 0.0.144
+Version: 0.0.146
 Summary: Options & volatility research package
 Author-email: Manu de Cara <manu.de.cara@gmail.com>
 License: MIT

{voly-0.0.144 → voly-0.0.146}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "voly"
-version = "0.0.144"
+version = "0.0.146"
 description = "Options & volatility research package"
 readme = "README.md"
 authors = [
@@ -60,7 +60,7 @@ line_length = 100
 multi_line_output = 3
 
 [tool.mypy]
-python_version = "0.0.144"
+python_version = "0.0.146"
 warn_return_any = true
 warn_unused_configs = true
 disallow_untyped_defs = true

{voly-0.0.144 → voly-0.0.146}/src/voly/core/hd.py

@@ -16,11 +16,13 @@ from voly.formulas import iv, get_domain
 from voly.models import SVIModel
 from voly.core.fit import fit_model
 from arch import arch_model
-from arch.univariate import GARCH, EGARCH
 
 
 @catch_exception
-def get_historical_data(currency, lookback_days, granularity, exchange_name):
+def get_historical_data(currency: str,
+                        lookback_days: str,
+                        granularity: str,
+                        exchange_name: str) -> pd.DataFrame:
     """
     Fetch historical OHLCV data for a cryptocurrency.
 
@@ -37,9 +39,8 @@ def get_historical_data(currency, lookback_days, granularity, exchange_name):
 
     Returns:
     -------
-    df_hist : pandas.DataFrame containing the historical price data with OHLCV columns.
+    pd.DataFrame: Historical price data with OHLCV columns.
     """
-
     try:
         # Get the exchange class from ccxt
         exchange_class = getattr(ccxt, exchange_name.lower())
@@ -61,11 +62,15 @@ def get_historical_data(currency, lookback_days, granularity, exchange_name):
     ohlcv_list = []
     ohlcv = exchange.fetch_ohlcv(symbol, granularity, since=from_ts, limit=1000)
     ohlcv_list.append(ohlcv)
-    while True:
+
+    # Fetch all available data within the lookback period
+    while len(ohlcv) == 1000:
         from_ts = ohlcv[-1][0]
         new_ohlcv = exchange.fetch_ohlcv(symbol, granularity, since=from_ts, limit=1000)
-        ohlcv.extend(new_ohlcv)
-        if len(new_ohlcv) != 1000:
+        if len(new_ohlcv) <= 1:
+            break
+        ohlcv.extend(new_ohlcv[1:])  # Skip first element to avoid duplication
+        if len(new_ohlcv) < 1000:
             break
 
     # Convert to DataFrame
@@ -74,13 +79,13 @@ def get_historical_data(currency, lookback_days, granularity, exchange_name):
     df_hist.set_index('date', inplace=True)
     df_hist = df_hist.sort_index(ascending=True)
 
-    print(f"Data fetched successfully: {len(df_hist)} rows from {df_hist.index[0]} to {df_hist.index[-1]}")
+    logger.info(f"Data fetched successfully: {len(df_hist)} rows from {df_hist.index[0]} to {df_hist.index[-1]}")
 
     return df_hist
 
 
 @catch_exception
-def parse_window_length(window_length, df_hist):
+def parse_window_length(window_length: str, df_hist: pd.DataFrame) -> int:
     """
     Parse window length from string format (e.g., '7d', '30d') to number of data points.
 
@@ -96,7 +101,7 @@ def parse_window_length(window_length, df_hist):
     int
         Number of data points corresponding to the window length.
     """
-    if not window_length.endswith('d'):
+    if not isinstance(window_length, str) or not window_length.endswith('d'):
         raise VolyError("window_length should be in format '7d', '30d', etc.")
 
     # Extract number of days
@@ -115,21 +120,34 @@ def parse_window_length(window_length, df_hist):
 
 
 @catch_exception
-def fit_volatility_model(log_returns, df_hist, model_type='garch', distribution='normal', window_length='30d',
-                         n_fits=400):
+def fit_volatility_model(log_returns: np.ndarray,
+                         df_hist: pd.DataFrame,
+                         model_type: str = 'garch',
+                         distribution: str = 'normal',
+                         window_length: str = '30d',
+                         n_fits: int = 400) -> Dict[str, Any]:
     """
     Fit a volatility model (GARCH or EGARCH) to log returns.
 
-    Args:
-        log_returns: Array of log returns
-        df_hist: DataFrame with historical price data
-        model_type: Type of volatility model ('garch' or 'egarch')
-        distribution: Distribution type ('normal', 'studentst', or 'skewstudent')
-        window_length: Length of each window as a string (e.g., '30d')
-        n_fits: Number of sliding windows
+    Parameters:
+    -----------
+    log_returns : np.ndarray
+        Array of log returns
+    df_hist : pd.DataFrame
+        DataFrame with historical price data
+    model_type : str
+        Type of volatility model ('garch' or 'egarch')
+    distribution : str
+        Distribution type ('normal', 'studentst', or 'skewstudent')
+    window_length : str
+        Length of each window as a string (e.g., '30d')
+    n_fits : int
+        Number of sliding windows
 
     Returns:
-        Dict with model parameters and processes
+    --------
+    Dict[str, Any]
+        Dictionary with model parameters and processes
     """
     # Parse window length
     window_points = parse_window_length(window_length, df_hist)
@@ -137,36 +155,38 @@ def fit_volatility_model(log_returns, df_hist, model_type='garch', distribution=
     if len(log_returns) < window_points + n_fits:
         raise VolyError(f"Not enough data points. Need at least {window_points + n_fits}, got {len(log_returns)}")
 
-    # Adjust window sizes if necessary
-    n_fits = min(n_fits, len(log_returns) // 3)
-    window_points = min(window_points, len(log_returns) // 3)
+    # Adjust window sizes if necessary to avoid over-fitting
+    n_fits = min(n_fits, max(100, len(log_returns) // 3))
+    window_points = min(window_points, max(20, len(log_returns) // 3))
 
     start = window_points + n_fits
     end = n_fits
 
     # Different number of parameters based on model type and distribution
-    if model_type.lower() == 'garch':
-        if distribution.lower() == 'normal':
-            n_params = 4  # mu, omega, alpha, beta
-        elif distribution.lower() == 'studentst':
-            n_params = 5  # mu, omega, alpha, beta, nu
-        else:  # skewstudent
-            n_params = 6  # mu, omega, alpha, beta, nu, lambda (skew)
-    else:  # egarch
-        if distribution.lower() == 'normal':
-            n_params = 5  # mu, omega, alpha, gamma, beta
-        elif distribution.lower() == 'studentst':
-            n_params = 6  # mu, omega, alpha, gamma, beta, nu
-        else:  # skewstudent
-            n_params = 7  # mu, omega, alpha, gamma, beta, nu, lambda (skew)
+    param_names = get_param_names(model_type, distribution)
+    n_params = len(param_names)
 
     parameters = np.zeros((n_fits, n_params))
     z_process = []
 
-    logger.info(f"Fitting {model_type.upper()} model with {distribution} distribution using {n_fits} windows...")
+    logger.info(
+        f"Fitting {model_type.upper()} model with {distribution} distribution using {n_fits} windows of {window_length}...")
 
     for i in range(n_fits):
+        if i % (n_fits // 10) == 0:
+            logger.info(f"Fitting progress: {i}/{n_fits}")
+
+        # Skip if we don't have enough data
+        if end - i - 1 < 0 or start - i - 1 > len(log_returns):
+            continue
+
         window = log_returns[end - i - 1:start - i - 1]
+
+        # Skip windows that are too small or have invalid data
+        if len(window) < 10 or np.isnan(window).any() or np.isinf(window).any():
+            continue
+
+        # Mean-center the data to improve numerical stability
         data = window - np.mean(window)
 
         try:
@@ -176,101 +196,102 @@ def fit_volatility_model(log_returns, df_hist, model_type='garch', distribution=
             else:  # egarch
                 model = arch_model(data, vol='EGARCH', p=1, o=1, q=1, dist=distribution.lower())
 
-            fit_result = model.fit(disp='off')
+            fit_result = model.fit(disp='off', options={'maxiter': 1000})
 
             # Extract parameters based on model type and distribution
             params_dict = fit_result.params.to_dict()
 
-            if model_type.lower() == 'garch':
-                mu = params_dict.get("mu", 0)
-                omega = params_dict.get("omega", 0)
-                alpha = params_dict.get("alpha[1]", 0)
-                beta = params_dict.get("beta[1]", 0)
-
-                if distribution.lower() == 'normal':
-                    parameters[i, :] = [mu, omega, alpha, beta]
-                elif distribution.lower() == 'studentst':
-                    nu = params_dict.get("nu", 0)
-                    parameters[i, :] = [mu, omega, alpha, beta, nu]
-                else:  # skewstudent
-                    nu = params_dict.get("nu", 0)
-                    lam = params_dict.get("lambda", 0)
-                    parameters[i, :] = [mu, omega, alpha, beta, nu, lam]
-            else:  # egarch
-                mu = params_dict.get("mu", 0)
-                omega = params_dict.get("omega", 0)
-                alpha = params_dict.get("alpha[1]", 0)
-                gamma = params_dict.get("gamma[1]", 0)
-                beta = params_dict.get("beta[1]", 0)
-
-                if distribution.lower() == 'normal':
-                    parameters[i, :] = [mu, omega, alpha, gamma, beta]
-                elif distribution.lower() == 'studentst':
-                    nu = params_dict.get("nu", 0)
-                    parameters[i, :] = [mu, omega, alpha, gamma, beta, nu]
-                else:  # skewstudent
-                    nu = params_dict.get("nu", 0)
-                    lam = params_dict.get("lambda", 0)
-                    parameters[i, :] = [mu, omega, alpha, gamma, beta, nu, lam]
-
-            # Get last innovation
+            # Extract parameter values in correct order
+            param_values = [params_dict.get(param, 0) for param in param_names]
+            parameters[i, :] = param_values
+
+            # Get last innovation (standardized residual)
             residuals = fit_result.resid
             conditional_vol = fit_result.conditional_volatility
-            z_t = residuals[-1] / conditional_vol[-1]
-            z_process.append(z_t)
+
+            if len(residuals) > 0 and len(conditional_vol) > 0:
+                z_t = residuals[-1] / conditional_vol[-1]
+                if not np.isnan(z_t) and not np.isinf(z_t):
+                    z_process.append(z_t)
 
         except Exception as e:
             logger.warning(f"Model fit failed for window {i}: {str(e)}")
 
     # Clean up any failed fits
     if len(z_process) < n_fits / 2:
-        raise VolyError("Too many model fits failed. Check your data.")
+        raise VolyError(f"Too many model fits failed ({len(z_process)}/{n_fits}). Check your data.")
+
+    # Filter out rows with zeros (failed fits)
+    valid_rows = ~np.all(parameters == 0, axis=1)
+    parameters = parameters[valid_rows]
 
+    # Calculate average parameters and standard deviations
     avg_params = np.mean(parameters, axis=0)
     std_params = np.std(parameters, axis=0)
 
     return {
+        'model_type': model_type,
+        'distribution': distribution,
         'parameters': parameters,
         'avg_params': avg_params,
         'std_params': std_params,
         'z_process': np.array(z_process),
-        'model_type': model_type,
-        'distribution': distribution,
-        'param_names': get_param_names(model_type, distribution)
+        'param_names': param_names
     }
 
 
-def get_param_names(model_type, distribution):
-    """Get parameter names based on model type and distribution."""
+def get_param_names(model_type: str, distribution: str) -> List[str]:
+    """
+    Get parameter names based on model type and distribution.
+
+    Parameters:
+    -----------
+    model_type : str
+        Type of volatility model ('garch' or 'egarch')
+    distribution : str
+        Distribution type ('normal', 'studentst', or 'skewstudent')
+
+    Returns:
+    --------
+    List[str]
+        List of parameter names
+    """
     if model_type.lower() == 'garch':
         if distribution.lower() == 'normal':
-            return ['mu', 'omega', 'alpha', 'beta']
+            return ['mu', 'omega', 'alpha[1]', 'beta[1]']
         elif distribution.lower() == 'studentst':
-            return ['mu', 'omega', 'alpha', 'beta', 'nu']
+            return ['mu', 'omega', 'alpha[1]', 'beta[1]', 'nu']
         else:  # skewstudent
-            return ['mu', 'omega', 'alpha', 'beta', 'nu', 'lambda']
+            return ['mu', 'omega', 'alpha[1]', 'beta[1]', 'nu', 'lambda']
     else:  # egarch
         if distribution.lower() == 'normal':
-            return ['mu', 'omega', 'alpha', 'gamma', 'beta']
+            return ['mu', 'omega', 'alpha[1]', 'gamma[1]', 'beta[1]']
         elif distribution.lower() == 'studentst':
-            return ['mu', 'omega', 'alpha', 'gamma', 'beta', 'nu']
+            return ['mu', 'omega', 'alpha[1]', 'gamma[1]', 'beta[1]', 'nu']
         else:  # skewstudent
-            return ['mu', 'omega', 'alpha', 'gamma', 'beta', 'nu', 'lambda']
+            return ['mu', 'omega', 'alpha[1]', 'gamma[1]', 'beta[1]', 'nu', 'lambda']
 
 
 @catch_exception
-def simulate_volatility_paths(vol_model, horizon, simulations=5000, variate_parameters=True):
+def simulate_volatility_paths(vol_model: Dict[str, Any],
+                              horizon: int,
+                              simulations: int = 5000) -> Tuple[np.ndarray, float]:
     """
     Simulate future paths using a fitted volatility model.
 
-    Args:
-        vol_model: Dict with volatility model parameters
-        horizon: Number of steps to simulate
-        simulations: Number of paths to simulate
-        variate_parameters: Whether to vary parameters between simulations
+    Parameters:
+    -----------
+    vol_model : Dict[str, Any]
+        Dict with volatility model parameters
+    horizon : int
+        Number of steps to simulate
+    simulations : int
+        Number of paths to simulate
 
     Returns:
-        Array of simulated log returns
+    --------
+    Tuple[np.ndarray, float]
+        Simulated returns and drift
     """
     parameters = vol_model['parameters']
     z_process = vol_model['z_process']
@@ -282,18 +303,21 @@ def simulate_volatility_paths(vol_model, horizon, simulations=5000, variate_para
     pars = vol_model['avg_params'].copy()
     bounds = vol_model['std_params'].copy()
 
-    # Log parameters
-    param_str = ", ".join([f"{name}={par:.6f}" for name, par in zip(param_names, pars)])
+    # Create dictionary for easier parameter access
+    param_dict = {name: value for name, value in zip(param_names, pars)}
+
+    # Log parameters in a structured way
+    param_str = ", ".join([f"{name}={param_dict.get(name, 0):.6f}" for name in param_names])
     logger.info(f"{model_type.upper()} parameters: {param_str}")
 
-    # Create KDE for innovations based on distribution
+    # Create sampling function based on distribution
     if distribution.lower() == 'normal':
         # Use standard normal for normal distribution
         def sample_innovation(size=1):
             return np.random.normal(0, 1, size=size)
     else:
         # Use KDE for non-normal distributions to capture empirical distribution
-        kde = stats.gaussian_kde(z_process, bw_method='silverman')  # original code doesn't include bw_method
+        kde = stats.gaussian_kde(z_process, bw_method='silverman')  # original code didnt have bw_method
         z_range = np.linspace(min(z_process), max(z_process), 1000)
         z_prob = kde(z_range)
         z_prob = z_prob / np.sum(z_prob)
@@ -303,65 +327,65 @@ def simulate_volatility_paths(vol_model, horizon, simulations=5000, variate_para
 
     # Simulate paths
     simulated_returns = np.zeros(simulations)
+    mu = param_dict.get('mu', 0)
 
     for i in range(simulations):
         if (i + 1) % (simulations // 10) == 0:
             logger.info(f"Simulation progress: {i + 1}/{simulations}")
 
-        # Optionally vary parameters
-        if variate_parameters and (i + 1) % (simulations // 20) == 0:
-            new_pars = []
-            for j, (par, bound) in enumerate(zip(pars, bounds)):
-                var = bound ** 2 / len(parameters)
-                new_par = np.random.normal(par, var)
-                # Ensure omega is positive, betas are between 0 and 1, etc.
-                if j >= 1 and new_par <= 0:
-                    new_par = 0.01
-                new_pars.append(new_par)
-            sim_pars = new_pars
+        # Optionally vary parameters between simulations
+        if (i + 1) % (simulations // 20) == 0:
+            # Create parameter variations based on their estimated distribution
+            sim_params = {}
+            for j, (name, par, bound) in enumerate(zip(param_names, pars, bounds)):
+                var = bound ** 2 / max(len(parameters), 1)
+                # Generate new parameter from normal distribution around the mean
+                new_par = np.random.normal(par, np.sqrt(var))
+
+                # Apply constraints to ensure valid parameters
+                if name == 'omega':
+                    new_par = max(new_par, 1e-6)  # Must be positive
+                elif name in ['alpha[1]', 'beta[1]']:
+                    new_par = max(min(new_par, 0.999), 0.001)  # Between 0 and 1
+                elif name == 'nu':
+                    new_par = max(new_par, 2.1)  # Degrees of freedom > 2
+
+                sim_params[name] = new_par
         else:
-            sim_pars = pars.copy()
+            sim_params = param_dict.copy()
 
-        # Initialize variables based on model type
+        # Initialize volatility based on model type
         if model_type.lower() == 'garch':
-            if distribution.lower() == 'normal':
-                mu, omega, alpha, beta = sim_pars
-                sigma2 = omega / (1 - alpha - beta)
-            elif distribution.lower() == 'studentst':
-                mu, omega, alpha, beta, nu = sim_pars
-                sigma2 = omega / (1 - alpha - beta)
-            else:  # skewstudent
-                mu, omega, alpha, beta, nu, lam = sim_pars
-                sigma2 = omega / (1 - alpha - beta)
+            omega = sim_params.get('omega', 0)
+            alpha = sim_params.get('alpha[1]', 0)
+            beta = sim_params.get('beta[1]', 0)
+
+            # Initialize GARCH volatility (unconditional variance)
+            sigma2 = omega / (1 - alpha - beta) if alpha + beta < 1 else omega / 0.99
         else:  # egarch
-            if distribution.lower() == 'normal':
-                mu, omega, alpha, gamma, beta = sim_pars
-                log_sigma2 = omega / (1 - beta)
-                sigma2 = np.exp(log_sigma2)
-            elif distribution.lower() == 'studentst':
-                mu, omega, alpha, gamma, beta, nu = sim_pars
-                log_sigma2 = omega / (1 - beta)
-                sigma2 = np.exp(log_sigma2)
-            else:  # skewstudent
-                mu, omega, alpha, gamma, beta, nu, lam = sim_pars
-                log_sigma2 = omega / (1 - beta)
-                sigma2 = np.exp(log_sigma2)
+            omega = sim_params.get('omega', 0)
+            beta = sim_params.get('beta[1]', 0)
+
+            # Initialize EGARCH volatility
+            log_sigma2 = omega / (1 - beta) if beta < 1 else omega / 0.99
+            sigma2 = np.exp(log_sigma2)
 
         returns_sum = 0
 
-        # Simulate path
+        # Simulate path step by step
         for _ in range(horizon):
-            # Sample innovation
+            # Sample a random innovation
             z = sample_innovation()
 
-            # Update volatility and returns based on model type
+            # Update returns and volatility based on model type
             if model_type.lower() == 'garch':
                 # Calculate return
                 e = z * np.sqrt(sigma2)
                 returns_sum += e + mu
 
                 # Update GARCH volatility
-                sigma2 = omega + alpha * e ** 2 + beta * sigma2
+                sigma2 = sim_params.get('omega', 0) + sim_params.get('alpha[1]', 0) * e ** 2 + sim_params.get('beta[1]',
+                                                                                                              0) * sigma2
             else:  # egarch
                 # Calculate return
                 e = z * np.sqrt(sigma2)
@@ -369,6 +393,12 @@ def simulate_volatility_paths(vol_model, horizon, simulations=5000, variate_para
 
                 # Update EGARCH volatility
                 abs_z = abs(z)
+                gamma = sim_params.get('gamma[1]', 0)
+                alpha = sim_params.get('alpha[1]', 0)
+                beta = sim_params.get('beta[1]', 0)
+                omega = sim_params.get('omega', 0)
+
+                # EGARCH update equation
                 log_sigma2 = omega + beta * log_sigma2 + alpha * (abs_z - np.sqrt(2 / np.pi)) + gamma * z
                 sigma2 = np.exp(log_sigma2)
 
@@ -381,89 +411,95 @@ def get_hd_surface(model_results: pd.DataFrame,
                    df_hist: pd.DataFrame,
                    domain_params: Tuple[float, float, int] = (-1.5, 1.5, 1000),
                    return_domain: str = 'log_moneyness',
-                   method: str = 'arch_returns',
-                   model_type: str = 'garch',
+                   method: str = 'garch',
                    distribution: str = 'normal',
-                   **kwargs) -> Dict[str, Any]:
+                   window_length: str = '30d',
+                   n_fits: int = 400,
+                   simulations: int = 5000,
+                   bandwidth: str = 'silverman') -> Dict[str, Any]:
     """
     Generate historical density surface from historical price data.
 
     Parameters:
-        model_results: DataFrame with model parameters and maturities
-        df_hist: DataFrame with historical price data
-        domain_params: Tuple of (min, max, num_points) for x-domain
-        return_domain: Domain for x-axis values ('log_moneyness', 'moneyness', 'returns', 'strikes')
-        method: Method to use for HD estimation ('hist_returns' or 'arch_returns')
-        model_type: Type of volatility model to use ('garch' or 'egarch')
-        distribution: Distribution to use ('normal', 'studentst', or 'skewstudent')
-        **kwargs: Additional parameters for specific methods:
-            For volatility models ('garch'/'egarch' method):
-                n_fits: Number of sliding windows (default: 400)
-                simulations: Number of Monte Carlo simulations (default: 5000)
-                window_length: Length of sliding windows as string (default: '30d')
-                variate_parameters: Whether to vary parameters (default: True)
-                bandwidth: KDE bandwidth (default: 'silverman')
-            For 'hist_returns' method:
-                bandwidth: KDE bandwidth (default: 'silverman')
+    -----------
+    model_results : pd.DataFrame
+        DataFrame with model parameters and maturities
+    df_hist : pd.DataFrame
+        DataFrame with historical price data
+    domain_params : Tuple[float, float, int]
+        Tuple of (min, max, num_points) for x-domain
+    return_domain : str
+        Domain for x-axis values ('log_moneyness', 'moneyness', 'returns', 'strikes')
+    method : str
+        Method to use for HD estimation:
+        - 'garch': GARCH(1,1) model
+        - 'egarch': EGARCH(1,1,1) model with asymmetry
+        - 'basic': Simple histogram/KDE of historical returns
+    distribution : str
+        Distribution to use for volatility models ('normal', 'studentst', or 'skewstudent')
+    window_length : str
+        Length of sliding windows as string (e.g., '30d')
+    n_fits : int
+        Number of sliding windows for volatility model fitting
+    simulations : int
+        Number of Monte Carlo simulations for volatility models
+    bandwidth : str
+        KDE bandwidth method (default: 'silverman')
 
     Returns:
+    --------
+    Dict[str, Any]
         Dictionary containing pdf_surface, cdf_surface, x_surface, and moments
     """
-    # Check if required columns are present
+    # Validate inputs
     required_columns = ['s', 't', 'r']
     missing_columns = [col for col in required_columns if col not in model_results.columns]
     if missing_columns:
         raise VolyError(f"Required columns missing in model_results: {missing_columns}")
 
+    if len(df_hist) < 2:
+        raise VolyError("Not enough data points in df_hist")
+
     # Determine granularity from df_hist
-    if len(df_hist) > 1:
-        # Calculate minutes between consecutive timestamps
-        minutes_diff = (df_hist.index[1] - df_hist.index[0]).total_seconds() / 60
-        minutes_per_period = int(minutes_diff)
-    else:
-        raise VolyError("Cannot determine granularity from df_hist.")
+    minutes_diff = (df_hist.index[1] - df_hist.index[0]).total_seconds() / 60
+    minutes_per_period = max(1, int(minutes_diff))
 
-    # Validate model_type and distribution
-    valid_model_types = ['garch', 'egarch']
+    # Validate method and model parameters
+    valid_methods = ['garch', 'egarch', 'basic']
     valid_distributions = ['normal', 'studentst', 'skewstudent']
 
-    if model_type.lower() not in valid_model_types:
-        raise VolyError(f"Invalid model_type: {model_type}. Must be one of {valid_model_types}")
+    method = method.lower()
+    distribution = distribution.lower()
 
-    if distribution.lower() not in valid_distributions:
-        raise VolyError(f"Invalid distribution: {distribution}. Must be one of {valid_distributions}")
+    if method not in valid_methods:
+        raise VolyError(f"Invalid method: {method}. Must be one of {valid_methods}")
 
-    # Get method-specific parameters
-    if method == 'arch_returns':
-        n_fits = kwargs.get('n_fits', 400)
-        simulations = kwargs.get('simulations', 5000)
-        window_length = kwargs.get('window_length', '30d')
-        variate_parameters = kwargs.get('variate_parameters', True)
-        bandwidth = kwargs.get('bandwidth', 'silverman')
-        logger.info(
-            f"Using {model_type.upper()} method with {distribution} distribution, {n_fits} fits, {simulations} simulations")
-    elif method == 'hist_returns':
-        bandwidth = kwargs.get('bandwidth', 'silverman')
-        logger.info(f"Using returns-based KDE method with bandwidth {bandwidth}")
-    else:
-        raise VolyError(f"Unknown method: {method}. Use 'hist_returns', 'arch_returns'.")
+    if method in ['garch', 'egarch'] and distribution not in valid_distributions:
+        raise VolyError(f"Invalid distribution: {distribution}. Must be one of {valid_distributions}")
 
     # Calculate log returns from price history
     log_returns = np.log(df_hist['close'] / df_hist['close'].shift(1)) * 100
     log_returns = log_returns.dropna().values
 
-    # Fit volatility model once if using garch/egarch method
+    # Fit volatility model if using GARCH or EGARCH
     vol_model = None
-    if method == 'arch_returns':
+    if method in ['garch', 'egarch']:
+        model_type = method  # Use method as model_type
+        logger.info(
+            f"Using {model_type.upper()} with {distribution} distribution, {n_fits} fits, {simulations} simulations")
+
         vol_model = fit_volatility_model(
-            log_returns,
-            df_hist,
+            log_returns=log_returns,
+            df_hist=df_hist,
             model_type=model_type,
             distribution=distribution,
             window_length=window_length,
             n_fits=n_fits
         )
+    elif method == 'basic':
+        logger.info(f"Using basic returns-based KDE method with bandwidth {bandwidth}")
 
+    # Initialize result containers
     pdf_surface = {}
     cdf_surface = {}
     x_surface = {}
@@ -488,8 +524,8 @@ def get_hd_surface(model_results: pd.DataFrame,
 
         logger.info(f"Processing HD for maturity {i} (t={t:.4f} years, {tau_days_float:.2f} days)")
 
-        if method == 'hist_returns':
-            # Standard returns-based method
+        if method == 'basic':
+            # Simple returns-based method
             # Filter historical data for this maturity's lookback period
             start_date = pd.Timestamp.now() - pd.Timedelta(days=int(t * 365.25))
             maturity_hist = df_hist[df_hist.index >= start_date].copy()
@@ -519,7 +555,16 @@ def get_hd_surface(model_results: pd.DataFrame,
             f = stats.gaussian_kde(adj_returns, bw_method=bandwidth)
             pdf_values = f(LM)
 
-        elif method == 'arch_returns':
+            # Transform according to return domain
+            pdf_lm = pdf_values
+            pdf_m = pdf_lm / M
+            pdf_k = pdf_lm / K
+            pdf_r = pdf_lm / (1 + R)
+
+            # No model parameters to include
+            model_params = None
+
+        elif method in ['garch', 'egarch']:
             # Volatility model-based method
             if vol_model is None:
                 logger.warning(f"Volatility model fitting failed, skipping maturity {i}")
@@ -530,8 +575,7 @@ def get_hd_surface(model_results: pd.DataFrame,
             simulated_returns, simulated_mu = simulate_volatility_paths(
                 vol_model,
                 horizon,
-                simulations,
-                variate_parameters
+                simulations
             )
 
             # Scale the simulated returns to match target time horizon
@@ -548,25 +592,31 @@ def get_hd_surface(model_results: pd.DataFrame,
             # Convert to terminal prices
             simulated_prices = s * np.exp(risk_neutral_returns / 100)
 
-            # Convert to moneyness domain
+            # Convert to moneyness domain (x-domain)
             simulated_moneyness = s / simulated_prices
 
             # Perform KDE to get PDF
             kde = stats.gaussian_kde(simulated_moneyness, bw_method=bandwidth)
             pdf_values = kde(M)
 
+            # Transform according to return domain
+            pdf_m = pdf_values
+            pdf_lm = pdf_m * M
+            pdf_k = pdf_lm / K
+            pdf_r = pdf_lm / (1 + R)
+
             # Include volatility model params in moments
             avg_params = vol_model['avg_params']
             param_names = vol_model['param_names']
-            model_params = {name: value for name, value in zip(param_names, avg_params)}
-            model_params['model_type'] = model_type
+            model_params = {name.replace('[1]', ''): value for name, value in zip(param_names, avg_params)}
+            model_params['model_type'] = method
             model_params['distribution'] = distribution
 
-            # Add persistence for GARCH-type models
-            if model_type.lower() == 'garch':
+            # Add persistence for GARCH models
+            if method == 'garch':
                 model_params['persistence'] = model_params.get('alpha', 0) + model_params.get('beta', 0)
         else:
-            continue  # Skip this maturity if method is invalid
+            continue  # Skip if invalid method
 
         # Ensure density integrates to 1
         dx = LM[1] - LM[0]
@@ -577,41 +627,27 @@ def get_hd_surface(model_results: pd.DataFrame,
 
         pdf_values = pdf_values / total_area
 
-        # Common processing for both methods
-
-        # Transform densities to various domains
-        if method == 'hist_returns':
-            pdf_lm = pdf_values
-            pdf_m = pdf_lm / M
-            pdf_k = pdf_lm / K
-            pdf_r = pdf_lm / (1 + R)
-        else:  # volatility models
-            pdf_m = pdf_values
-            pdf_lm = pdf_m * M
-            pdf_k = pdf_lm / K
-            pdf_r = pdf_lm / (1 + R)
-
         # Calculate CDF
         cdf = np.cumsum(pdf_lm * dx)
-        cdf = np.minimum(cdf / cdf[-1], 1.0)
+        cdf = np.minimum(cdf / cdf[-1], 1.0)  # Ensure CDF is between 0 and 1
 
         # Select appropriate domain and calculate moments
         if return_domain == 'log_moneyness':
             x = LM
             pdf = pdf_lm
-            moments = get_all_moments(x, pdf, model_params if method in ['garch', 'egarch'] else None)
+            moments = get_all_moments(x, pdf, model_params)
         elif return_domain == 'moneyness':
             x = M
             pdf = pdf_m
-            moments = get_all_moments(x, pdf, model_params if method in ['garch', 'egarch'] else None)
+            moments = get_all_moments(x, pdf, model_params)
         elif return_domain == 'returns':
             x = R
             pdf = pdf_r
-            moments = get_all_moments(x, pdf, model_params if method in ['garch', 'egarch'] else None)
+            moments = get_all_moments(x, pdf, model_params)
         elif return_domain == 'strikes':
             x = K
             pdf = pdf_k
-            moments = get_all_moments(x, pdf, model_params if method in ['garch', 'egarch'] else None)
+            moments = get_all_moments(x, pdf, model_params)
         else:
             raise VolyError(f"Unsupported return_domain: {return_domain}")
 
@@ -621,11 +657,14 @@ def get_hd_surface(model_results: pd.DataFrame,
         x_surface[i] = x
         all_moments[i] = moments
 
+    # Check if we have any valid results
+    if not pdf_surface:
+        raise VolyError("No valid densities could be calculated. Check your input data.")
+
     # Create DataFrame with moments
     moments = pd.DataFrame(all_moments).T
 
-    logger.info(
-        f"Historical density calculation complete using {method} method with {model_type} model and {distribution} distribution")
+    logger.info(f"Historical density calculation complete using {method} method")
 
     return {
         'pdf_surface': pdf_surface,

{voly-0.0.144 → voly-0.0.146/src/voly.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: voly
-Version: 0.0.144
+Version: 0.0.146
 Summary: Options & volatility research package
 Author-email: Manu de Cara <manu.de.cara@gmail.com>
 License: MIT