RESEARCH NOTE

Live Scanner vs Hindsight Backtest

A live scanner acts on the data that was actually available at that moment. A later backtest can accidentally use cleaner, revised or differently sampled data.

What matters

This creates a research-integrity problem: the historical test may not represent what the product really showed in real time. The practical fix is point-in-time logging. Save the exact timestamp, candidate list and factor values that the live process observed, then evaluate later outcomes from those frozen observations.

How Money Radar uses this

Money Radar now preserves verified scanner snapshots before measuring later +1h, +4h, +8h and +12h outcomes. Historical outcomes remain separate from the anomaly score.