Backtesting

Choose the validation workflow that matches the evidence you want to test.

How to read a backtestBacktesting checks whether a research rule held up historically without using information from the future.

1
Define the metric, target line, historical sample and minimum evidence required to create a signal.
2
Read Opportunities as all eligible historical cases and Signals as the subset where the prior evidence met your signal rule.
3
Compare signal hit rate with the baseline hit rate. Lift is the difference between them in percentage points.
4
Check the chronological holdout: development data comes first in time and validation data comes later and remains untouched when the rule is formed.
5
Prefer repeated evidence that remains stable through time and in validation over a very high hit rate built from only a few signals.

Key terms

Signal

A historical opportunity where the evidence available before the target game met the configured rule.

Hit rate

The share of signals whose target outcome was reached in the following game.

Baseline

The target outcome rate across all eligible opportunities, not only the opportunities that produced a signal.

Lift

Signal hit rate minus baseline hit rate, expressed in percentage points.

Positive lift is evidence of historical separation, not proof that the rule will remain profitable or predictive.

Holdout validation

The final chronological portion of the data that is evaluated separately from the earlier development sample.

StatMindX backtesting validates statistical signal behaviour. It does not use betting odds and does not make profitability claims.

Player Backtesting

Validate player-level rules across outfield and goalkeeper cohorts with no look-ahead.

Validate team For, Against and Match Total rules across FT, 1H and 2H evidence.

Resume saved player and team validation workflows from the Research Board.