What question did this study set out to answer?

The paper aims to resolve the issue of non-stationarity in equity signal generation using a market regime approach.

March 14, 2026Open Access

OLYMPIA: Market-Regime-Conditioned Machine Learning for Equity Signal Generation

View Full Paper

Ask AI

Bookmark

Authors

This framework enhances out-of-sample precision in equity signal generation, indicating significant model improvement.

The paper aims to resolve the issue of non-stationarity in equity signal generation using a market regime approach.
Developed Olympia, a machine learning framework incorporating market regime conditions.
Used gradient-boosted decision trees combined with isotonic probability calibration.
Employed a walk-forward backtesting methodology on 11,845 US equities.
Analyzed approximately 3.46 million observations to assess model performance.
Increased out-of-sample precision from 12% to 62% with regime features.
Achieved approximately 90% top-10 out-of-sample precision under favorable market conditions.
Demonstrated a fully automated pipeline that evaluates the US equity universe within 22 minutes.

Michael Rupert (2026) studied this question.

View Full Paper

Ask AI

Bookmark

March 14, 2026Open Access

View Full Paper

Ask AI

Bookmark

Authors

This framework enhances out-of-sample precision in equity signal generation, indicating significant model improvement.

The paper aims to resolve the issue of non-stationarity in equity signal generation using a market regime approach.
Developed Olympia, a machine learning framework incorporating market regime conditions.
Used gradient-boosted decision trees combined with isotonic probability calibration.
Employed a walk-forward backtesting methodology on 11,845 US equities.
Analyzed approximately 3.46 million observations to assess model performance.
Increased out-of-sample precision from 12% to 62% with regime features.
Achieved approximately 90% top-10 out-of-sample precision under favorable market conditions.
Demonstrated a fully automated pipeline that evaluates the US equity universe within 22 minutes.

Michael Rupert (2026) studied this question.

View Full Paper

Ask AI

Bookmark