skfolio.utils.stats.squared_standardized_euclidean_dist#

skfolio.utils.stats.squared_standardized_euclidean_dist(returns, covariance)[source]#

Squared standardized Euclidean distance.

\[d^2 = \sum_i (r_i\,/\,\sigma_i)^2\]

This is the squared Mahalanobis distance using only the diagonal of the covariance matrix (ignoring correlations).

Parameters:
returnsndarray of shape (n_assets,)

Asset return vector.

covariancendarray of shape (n_assets, n_assets)

Covariance matrix.

Returns:
float

Sum of squared standardized returns (non-negative). Under correct calibration: \(\mathbb{E}[d^2] = n_{\text{assets}}\).