Methodology

The scanner

Each week, the pipeline screens the equity universe for stocks exhibiting the structural characteristics of historical “big movers” — stocks that went on to gain 100% or more within 30 weeks of a qualifying setup. Candidates are matched against a library of named profiles derived from those historical events.

The model

A proprietary machine learning model trained on those historical matches ranks each week's candidates. The training objective is ranking — the model is not predicting return magnitude, but ordering candidates by their probability of belonging to the top tier of outcomes.

The production model is version v9 (French). A new candidate model must beat the live model's Precision@5 by at least 5 percentage points before it is promoted — ties keep the incumbent.

Outcomes and the 12-week window

A candidate's outcome is not written until 12 weeks have elapsed. The metric is maximum favourable excursion (MFE) — the highest return the stock reached at any point within 30 weeks of the scan date. Outcome files are written once and never edited, enforced at the code level.