During live soccer games, ShotLab displays a separate in-play expected final shots total based on verified current shots plus a pace update. It never presents a pregame number as an in-play forecast. A 2-shot first half is meaningful evidence against a high pregame forecast. The model is not calibrated and may miss tactical changes, red cards, and score effects.

The NBA, NFL, NHL, NCAA basketball, and MLB winner estimates use recent completed games, weighted scoring margins and a modest home advantage. The soccer model uses recent goals scored and conceded to estimate goal rates, then computes home/draw/away outcomes with a Poisson model. Soccer team shots / shots on target, NHL team shots on goal, and basketball field-goal attempts are projected from box-score data where available, including combined totals. NFL has no literal shots market, and PGA uses tournament leaderboards rather than team moneylines. Missing source data never becomes a fabricated projection.

Feedback log: Logged prediction errors measure historical performance; they do not automatically update model coefficients or prove future accuracy. Old/manual entries cannot verify that a number was recorded before the match.

Samples are limited to games before the selected matchup, so future games do not leak into a retrospective projection. Matchup strength, injuries, starting goalies, lineups, travel, weather, and bookmaker market adjustments are not fully modeled. Probabilities have not been validated out of sample. Never treat high displayed numbers as a reason to bet or chase losses.

Data: unofficial ESPN site data endpoints, accessed directly. Feeds can be blocked, go offline or change format. The app will display the error rather than fabricate a result. Most actions need an active internet connection.