ML research
Machine learning for trading without fooling yourself: labels, purged validation, overfitting diagnostics and feature importance.
Backtest Your Own Python Signal on Trigr
Upload a signal computed in Python or a notebook, read it in a formula with upload(), and backtest it point in time on Trigr. Timestamps, lags and limits.
Data Leakage in Trading ML: The Checks That Matter
Data leakage in machine learning trading models makes backtests look brilliant and live trading fail. Learn the leak sources and the checks that catch them.
LightGBM vs Random Forest vs XGBoost for Trading
LightGBM, random forest or XGBoost for an XGBoost trading strategy? How the three tree models differ on noisy market data, and how to choose honestly.
Out-of-Sample Testing for Trading Strategies Done Right
Out-of-sample testing for trading strategies: when a holdout works, why walk-forward is better, and why a paper forward test is the only truly unseen data.
Sample Uniqueness and Overlapping Labels in Financial ML
Sample uniqueness in financial machine learning fixes overlapping labels that count one market move many times. How average uniqueness works, with examples.
Feature Importance for Trading Models: MDA, SFI, SHAP
Feature importance for trading models done right: why MDI misleads, how MDA, SFI and SHAP differ, and how to read them together without fooling yourself.
Machine Learning Trading Strategies Without Overfitting
How to build a machine learning trading strategy that survives out of sample: point-in-time features, triple-barrier labels, purged validation, DSR and PBO.
Probability of Backtest Overfitting (PBO) Explained
The probability of backtest overfitting (PBO) estimates how often your best in-sample pick lands in the bottom half out of sample. How to read it.
Purged Walk-Forward Validation, Explained Simply
Purged walk-forward validation removes training samples whose labels overlap the test window, then adds an embargo. Here is why it matters and how it works.
Triple-Barrier Method and Meta-Labeling in Practice
The triple-barrier method labels trades by which of a profit target, stop or time limit is hit first. Meta-labeling then learns which signals to take.
Put the idea to an honest test.
Describe a strategy in plain English or from your own AI assistant, backtest it on point-in-time data, and forward-test it on paper before any real money is involved.