oos-lab

Research-grade validation toolkit for systematic trading strategies.

Most backtests are fantasies. oos-lab gives you the statistical tools to tell whether yours is real, with no buy/sell signals and no performance claims of its own. Open source, MIT licensed.

What is in v0.4

Metrics

• sharpe_ratio — annualised sample Sharpe with ddof=1.
• probabilistic_sharpe_ratio — Bailey & López de Prado (2012) PSR. The probability that the true Sharpe exceeds a benchmark, accounting for the skew and kurtosis of returns.
• deflated_sharpe_ratio — Bailey & López de Prado (2014) DSR. PSR with the benchmark raised to the expected maximum across n_trials strategy variants under the false strategy theorem. The first defence against selection bias.
• expected_max_sharpe — the underlying closed form, useful on its own when you want to see how high a "lucky max" Sharpe gets without any real edge.

Cross-validation

• WalkForward — anchored or rolling walk-forward index splitter for time-series strategies.
• CombinatorialPurgedKFold — combinatorial purged cross-validation with embargo (López de Prado 2018), the leakage-aware splitter for financial series.

Overfitting

• probability_of_backtest_overfit — Probability of Backtest Overfitting via Combinatorially Symmetric Cross-Validation (Bailey, Borwein, López de Prado & Zhu 2014). Returns the PBO scalar plus the performance-degradation regression.
• haircut_sharpe — Harvey & Liu (2015) multiple-testing haircut with Holm-Bonferroni and Benjamini-Hochberg-Yekutieli adjustments.

Every public symbol is re-exported from oos_lab/__init__.py. The package depends only on numpy and scipy. 56 unit tests pass.

Install

pip install oos-lab

Quickstart

import numpy as np
from oos_lab import (
    sharpe_ratio,
    probabilistic_sharpe_ratio,
    deflated_sharpe_ratio,
    expected_max_sharpe,
    WalkForward,
    CombinatorialPurgedKFold,
    probability_of_backtest_overfit,
    haircut_sharpe,
)

rng = np.random.default_rng(0)
returns = rng.normal(0.0008, 0.012, size=1000)

print("Sharpe annualised:", sharpe_ratio(returns, periods_per_year=252))
print("PSR vs zero      :", probabilistic_sharpe_ratio(returns))
print("DSR over 1000    :", deflated_sharpe_ratio(returns,
                                                  n_trials=1000,
                                                  var_sharpe_across_trials=0.04))
print("Expected max SR  :", expected_max_sharpe(1000, 0.04))

# PBO needs a matrix of variant returns: shape (n_obs, n_variants)
variants = rng.normal(0.0002, 0.012, size=(1000, 20))
result = probability_of_backtest_overfit(variants, n_partitions=16)
print("PBO              :", result.pbo)
print("Degradation slope:", result.performance_degradation_slope)

What this is NOT

• Not a backtester. Bring your own engine.
• Not a strategy. There is no buy or sell logic anywhere.
• Not financial advice, and not a personalised recommendation. Educational and methodological use only. oos-lab is an impersonal tool that runs the same way for everyone; it makes no claim about your account, your positions, or any future result.

A note on any performance numbers

oos-lab computes statistics; it never shows a track record of its own. If you use it to present simulated or backtested results to others, attach the standard hypothetical-performance disclaimer (CFTC Rule 4.41): simulated results have inherent limitations, do not represent actual trading, and are designed with the benefit of hindsight, so no representation is made that any account will achieve similar profits or losses.

Documentation

Per-module write-ups live in docs/: sharpe, psr, deflated_sharpe, walk_forward, cpcv, pbo, haircut, plus an overview.

Roadmap

• v0.5 — bootstrap confidence intervals for the deflated Sharpe.
• v0.6 — minimum backtest length and minimum track record length helpers.

References

• Bailey, D. H., & López de Prado, M. (2012). The Sharpe Ratio Efficient Frontier. Journal of Risk, 15(2), 3-44.
• Bailey, D. H., & López de Prado, M. (2014). The Deflated Sharpe Ratio. The Journal of Portfolio Management, 40(5), 94-107.
• Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). The Probability of Backtest Overfitting. Journal of Computational Finance, 20(4), 39-70.
• Harvey, C. R., & Liu, Y. (2015). Backtesting. The Journal of Portfolio Management, 42(1), 13-28.
• López de Prado, M. (2018). Advances in Financial Machine Learning. Wiley.

License

MIT. See LICENSE.