Console / Research / Analysis & Validation
Proof that an edge is real.
A dedicated validation layer quantifies robustness before deployment: out-of-sample and holdout performance, overfitting probability, configuration comparison, and multi-dimensional parameter analysis.
Between a promising backtest and committed capital sits the most important step in the process: proving the result will hold. The validation layer reports holdout performance, a probability that the backtest is overfitted, and a side-by-side comparison of candidate configurations.
Judged on data withheld from selection
Performance is measured on holdout data, and holdout results drive the deployment decision. A strategy that is strong in-sample and weak out-of-sample is reported as unestablished.
Quantify the overfitting risk
Probability-of-backtest-overfitting metrics estimate how likely the selected configuration is a product of the search rather than a real effect. The figure is reported alongside the deflated Sharpe ratio and the in-sample to out-of-sample degradation across every split.
Compare candidates on equal footing
Evaluate configurations against one another on the same data and the same terms, with parameter landscapes and diagnostics that make the trade-offs explicit. The recommendation arrives with the comparison behind it.
For institutions
A committee-ready decision, with its evidence attached.
Out-of-sample results, a quantified overfitting probability, and side-by-side candidate comparison are the artifacts an investment or risk committee needs to record the grounds on which a strategy was approved or declined.
Next step
Analysis & Validation, live on your strategies and venues.
Response within two business days