Counts the analyzed fixtures, leagues, countries, stored predictions with at least 50 confidence, and fixtures with bookmaker odds. Top markets are ranked by average model confidence; confidence is not the same as historical accuracy.

Uses the highest matching stored bookmaker price. Implied probability is 100 ÷ decimal odds, and Edge is the model probability minus that implied probability. Only rows with confidence of at least 55, valid odds, and positive edge are shown. Bookmaker margin and model calibration still matter.

Displays average stored model confidence by league and normalized market family. The number under each cell is the prediction sample. Darker cells mean higher average confidence, not proven profitability or settled accuracy.

Consolidates duplicate feeds per bookmaker, converts decimal odds to implied probabilities, and removes each bookmaker market's overround by normalizing across its available outcomes. Consensus is higher when distinct bookmakers' normalized probabilities are close together. It measures price agreement, not outcome certainty.

Uses the upcoming fixture's explicit league season and grades the highest-probability pick per supported market from settled fixtures in the previous 365 days. The score combines settled accuracy (55%), Brier-based calibration (30%), sample size up to 100 picks (15%), and data freshness. Unsupported or incomplete markets are excluded. Season baseline statistics are displayed for context.