Experimental models — research results
Backtesting
Comparison of model configurations across a training window and a held-out test window, with confidence intervals where the API reports them.
single-number 12.2%top-3 33.2%
Experimental model — unvalidated
In testing. No result here is confirmed out-of-sample.
Period selector
Applies to server-side backtest runs
Selections are submitted to the backtest endpoint. No client-side backtest is simulated.
Querying API…
Audited out-of-sample summary
performanceSummary from /dashboard — Clopper-Pearson 95% intervals as reported
No summary rows returned.
Distinction labels are mandatory: a historical observation is measured on data the model was fitted to; an out-of-sample result is measured only on pre-draw records.