openquant.hrp

Compiled from Rust (crates/pyopenquant). Signatures and types are from the shipped .pyi stubs; descriptions are the functions’ own docstrings, as help() shows them. All Python modules.

Functions

allocate_hrp

allocate_hrp(
    asset_names: Sequence[str],
    asset_prices: Sequence[Sequence[float]] | None = None,
    asset_returns: Sequence[Sequence[float]] | None = None,
    covariance_matrix: Sequence[Sequence[float]] | None = None,
    resample_by: str | None = None,
    use_shrinkage: bool = False,
    distance: str | None = None,
) -> tuple[list[float], list[int]]

Hierarchical Risk Parity weights (AFML chapter 16, Snippets 16.1-16.4).

Allocates without inverting the covariance matrix: (1) tree clustering on correlation distances d = sqrt((1 - rho) / 2) by single linkage (section 16.4.1); (2) quasi-diagonalisation so similar assets are adjacent (16.4.2); (3) recursive bisection, splitting weight between halves in inverse proportion to their inverse-variance cluster variances (16.4.3). The result is long-only and fully invested. The covariance is taken from covariance_matrix if given, else estimated from asset_returns, else from simple returns of asset_prices; HRP weights do not depend on its scale.

Parameters

asset_names : list[str]

One name per asset; its length fixes the number of assets N.

asset_prices : list[list[float]] | None, default None

Prices, one inner list per observation (oldest first) and N columns. Used only when neither asset_returns nor covariance_matrix is given.

asset_returns : list[list[float]] | None, default None

Per-period returns, one inner list per observation and N columns. Used when no covariance_matrix is given.

covariance_matrix : list[list[float]] | None, default None

N x N covariance, used as given when present.

resample_by : str | None, default None

For prices only: "W"/"week"/"weekly" keeps every 5th row, "M"/"month"/"monthly" every 21st (case-insensitive); anything else, or None, keeps every row.

use_shrinkage : bool, default False

Multiply the off-diagonal terms of an estimated covariance by 0.9 (a fixed shrink, not Ledoit-Wolf); no effect on a supplied covariance_matrix.

distance : str | None, default None

Distance the tree is built on (case-insensitive): "distance_of_distances" (the default; the Euclidean distance between columns of d, as AFML Snippet 16.4 computes) or "correlation" (cluster on d itself, as mlfinlab does).

Returns

tuple[list[float], list[int]]

(weights, ordered_indices): non-negative weights summing to 1, one per asset in asset_names order, and the asset indices in quasi-diagonal (leaf) order.

Raises

ValueError

If distance is unknown, a matrix is empty or ragged, or the core rejects the input (e.g. no prices, returns or covariance given, empty asset_names, fewer than two price or return rows, a zero price, a non-positive variance, or a matrix whose shape disagrees with the number of asset names).