Module Performance

How each prop module is actually performing on settled outcomes. The model can only learn from settlements — picks that have graded as W/L. As more games complete, this report sharpens.
Total Picks
Settled
Pending
Hit Rate
ROI %
By Market Family
Market
N
W
L
Hit %
Net Units
Hit Rate
Loading…
By Sport
Sport
N
W
L
Hit %
Net Units
Hit Rate
Loading…
By Source Module
Source
N
W
L
Hit %
Net Units
Hit Rate
Loading…
How learning works: Each settled pick (win/loss) feeds back into self_learn_weights.py, which adjusts the per-source shrinkage weighting. Modules with high settled hit rates get more weight in confluence ranking; bad sources get downweighted. With most picks still pending, weights are calibrating slowly. As outcomes accumulate, the brain gets sharper.