quantlab.portfolio.predefined.mean_variance.MeanVarianceInputs

class quantlab.portfolio.predefined.mean_variance.MeanVarianceInputs(symbols, expected_return, estimate, current_weights, locked_symbols=<factory>, locked_weights=<factory>, risk_locked_weights=<factory>, closed_without_risk=<factory>, exposures=<factory>, locked_exposures=<factory>, exposure_bounds=<factory>, closed_without_exposure=<factory>, reference_weights=<factory>, reference_without_risk=<factory>)

Bases: object

The problem one bar poses: the candidate symbols, the locked ones and their inputs.

Attributes:

symbolsnp.ndarray

The candidates the optimiser sets: tradable, not locked, covered by the covariance estimator, and either with a finite expected-return prediction or held; with candidate_top_k, only the pool.

expected_returnnp.ndarray

mu per candidate, on the expected-return label’s span; 0.0 for a held candidate without a prediction.

estimateReturnCovariance

The covariance estimator’s estimate, on the same span, over the candidates, then the locked symbols it covers, then, with a reference book, the reference’s other symbols it covers.

current_weightsnp.ndarray

The weights currently held on the candidates.

locked_symbolsnp.ndarray

The locked positions: held and not tradable at the bar.

locked_weightsnp.ndarray

Their current weights, which the solution keeps.

risk_locked_weightsnp.ndarray

The locked weights the estimate covers, in its order after the candidates.

closed_without_risknp.ndarray

Held, tradable symbols the covariance estimator does not cover, closed.

exposuresnp.ndarray

[k, candidates]: each bounded exposure per candidate; [0, candidates] without bounds.

locked_exposuresnp.ndarray

[k]: each bounded exposure of the locked positions, sum w * x.

exposure_boundsnp.ndarray

[k, 2]: each bounded exposure’s (lower, upper), in the rows’ order.

closed_without_exposurenp.ndarray

Held, tradable symbols lacking a bounded exposure, closed.

reference_weightsnp.ndarray

The reference book b on estimate’s symbols, the risk term pricing w - b; empty without one (reference_weights).

reference_without_risknp.ndarray

Symbols of the reference book the estimate does not cover, left out of it.

Examples

>>> inputs = optimizer.problem_inputs(context)
>>> inputs.symbols.tolist(), inputs.covariance.shape
(['AAA', 'BBB', 'CCC', 'DDD'], (4, 4))

property covariance

Sigma as a dense matrix, over the candidates then the covered locked symbols.

Examples

>>> inputs.covariance.shape
(4, 4)