We describe Ringside Analytics, an end-to-end machine-learning system trained on 40+
years of professional wrestling data — 482,166 matches and 731,133 wrestler-match par-
ticipations spanning WWE, AEW, WCW, ECW, NXT, and TNA. Pro wrestling outcomes are
scripted, which makes the prediction target an artifact of human creative decisions rather
than athletic measurement. We discuss how this kayfabe problem shapes feature engineer-
ing, evaluation, and interpretation; report honest test-set performance for both XGBoost
(AUC 0.718) and a logistic-regression baseline (AUC 0.698); and document a 25-point vali-
dation→test AUC gap that reveals temporal autocorrelation in storyline arcs. The dataset,
trained model, and source code are released under permissive licenses.