Hi all,

The best soccer prediction approach is not vibes, it is process. I start with a quality prior like ClubElo, then adjust toward the mean because leagues and continental competitions have different baselines. Next I bring in real expected-goals numbers (xG and xGA) because goals are noisy but expected goal models actually describe repeatable quality.

Finally, I weight two things far more than most casual predictors: squad availability (injuries, suspensions, red cards) and travel load (distance + rest days). When I combine those layers, my predicted edges become explainable, testable and improvable. A method you cannot formalize is not a prediction method.