sf_quant.optimizer.mve_optimizer

sf_quant.optimizer.mve_optimizer(ids: list[str], alphas: ndarray, covariance_matrix: ndarray, constraints: list[Constraint], gamma: float = 2, betas: ndarray | None = None) → DataFrame

Mean-variance optimizer with constraints.

This function solves the constrained mean-variance optimization problem and returns optimal portfolio weights aligned with the provided ids.

Parameters

idslist of str

Identifiers for the assets, used to label the output.

alphasnp.ndarray

Expected returns for each asset, shape (n_assets,).

covariance_matrixnp.ndarray

Covariance matrix of asset returns, shape (n_assets, n_assets).

constraintslist of Constraint

List of constraint objects implementing the Constraint protocol.

gammafloat

Risk-aversion parameter. Higher values penalize variance more strongly. Defaults to 2.

betasnp.ndarray, optional

Predicted betas, required for certain constraints (e.g., UnitBeta).

Returns

pl.DataFrame

A Polars DataFrame with barrid and weight columns.

Examples

>>> import sf_quant.optimizer as sfo
>>> import numpy as np
>>> ids = ['AAPL', 'IBM']
>>> alphas = np.array([1.1, 1.2])
>>> covariance_matrix = np.array([
...     [.5, .1],
...     [.1, .2]
... ])
>>> constraints = [sfo.FullInvestment()]
>>> weights = sfo.mve_optimizer(
...     ids=ids,
...     alphas=alphas,
...     covariance_matrix=covariance_matrix,
...     constraints=constraints
... )
>>> weights
shape: (2, 2)
┌────────┬────────┐
│ barrid ┆ weight │
│ ---    ┆ ---    │
│ str    ┆ f64    │
╞════════╪════════╡
│ AAPL   ┆ 0.1    │
│ IBM    ┆ 0.9    │
└────────┴────────┘