aether_quant.overfitting¶

Overfitting tests, grouped by the job (V6.6.5): PBO by CSCV, the probabilistic and deflated Sharpe, the minimum track record, the haircut Sharpe, placebo tests, the leakage audit and the selection gate the research rounds apply.

pbo `module-attribute` ¶

pbo = probability_of_backtest_overfitting

audit_leakage ¶

audit_leakage(frame: DataFrame, *, feature_columns: Sequence[str], target_column: str = 'target_rank_20d', ticker_column: str = 'ticker', date_column: str = 'date', split_column: str = 'split', horizon: int = 20, class_columns: Sequence[str] = (), model_score_column: str | None = None, min_names: int = 20, seed: int = 7, group_column: str | None = None) -> dict[str, Any]

All four checks; (b) and (c) on the validation split, (d) fit on train and read on validation.

deflated_sharpe_from_inputs ¶

deflated_sharpe_from_inputs(returns: ArrayLike, *, num_trials: int, trial_variance: float | None, annualization: float) -> dict[str, Any]

The deflated Sharpe with N and the trial variance given (evaluation.trial_ledger.deflation_inputs), not derived from a list of trial Sharpes: the variance must come from trials on the window being judged.

deflated_sharpe_ratio ¶

deflated_sharpe_ratio(returns: ArrayLike, trial_annual_sharpes: list[float], periods_per_year: int = DAYS_PER_YEAR) -> dict[str, Any]

DSR (Bailey & López de Prado 2014): PSR against SR0, the best Sharpe expected from the number of configurations tried. Trial Sharpes are annual; they are moved to this series' period (÷ sqrt(periods_per_year)) first.

expected_max_sharpe ¶

expected_max_sharpe(num_trials: int, trial_sharpe_variance: float) -> float | None

SR0 of the deflated Sharpe ratio: the expected maximum Sharpe of num_trials unskilled trials whose Sharpe ratios have variance trial_sharpe_variance (same units as the returned value).

haircut_sharpe ¶

haircut_sharpe(sharpe: float, n_obs: int, num_trials: int, *, method: Literal['bonferroni', 'holm', 'bhy'] = 'bhy', periods_per_year: int = 252) -> dict[str, float]

The annual Sharpe that survives num_trials: its t-statistic SR * sqrt(n_obs / periods_per_year), the two-sided p-value, the p-value adjusted for multiple testing, and the Sharpe implied by the adjusted p-value.

At rank 1, Holm equals Bonferroni (p * M) and BHY multiplies by M * c(M) with c(M) = sum(1/i). Returns {"sharpe", "haircut_sharpe", "haircut", "t_stat", "p_value", "adjusted_p_value"}; a non-positive Sharpe gets a 100% haircut.

minimum_track_record_length ¶

minimum_track_record_length(returns: ArrayLike, *, target_sharpe: float = 0.0, confidence: float = 0.95, periods_per_year: int = 252) -> float

Observations needed before the observed Sharpe beats target_sharpe (annual) at confidence, given the series' skew and kurtosis: 1 + (1 - g3*SR + (g4-1)/4*SR^2) * (z / (SR - SR*))^2 with per-period SR and raw kurtosis g4. inf when the observed Sharpe does not exceed the target; nan for fewer than 3 finite observations or zero variance.

placebo_p_value ¶

placebo_p_value(observed: float | None, placebos: list[float]) -> float

Share of placebos at or above the observed Sharpe, counting the observation itself (never exactly zero).

probabilistic_sharpe_ratio ¶

probabilistic_sharpe_ratio(returns: ArrayLike, benchmark_sharpe_per_period: float = 0.0) -> float | None

PSR (Bailey & López de Prado 2012): P(true Sharpe > benchmark) given the sample's length, skew and kurtosis. Both Sharpe ratios are per period (not annualised).

probability_of_backtest_overfitting ¶

probability_of_backtest_overfitting(returns: ArrayLike, *, blocks: int = 16, periods_per_year: int = 252) -> dict[str, Any]

{"pbo", "num_combinations", "num_trials", "lambda_quantiles", "degradation_slope", "degradation_intercept", "prob_oos_loss"}; pbo is None when there are fewer than 2 trials or fewer rows than blocks.

selection_gate ¶

selection_gate(scores: list[dict], p_values: Mapping[str, Mapping[str, float]], *, max_survivors: int = 5, max_correlation: float = 0.7, alpha: float = 0.05) -> dict[str, Any]

R-SLEEVE2. A survivor passes both Sharpe clauses and every placebo test; among survivors the higher Sharpe keeps its place against any other with daily correlation above max_correlation; at most max_survivors remain.