Player props · value 18+
For each line (e.g. “Pts+Reb Over 13.5”) the model computes the probability from the player’s expected minutes and per-minute rate (weighted towards recent games). When a key player who played last game is out, about 45% of his minutes (as the backtest showed) go to teammates, mostly in the same position. Because the market knows things we don’t, the probability we use is half model, half no-vig market. Value = the price beats the fair price (1/probability): we show only those with EV of at least +8%.
Gambling is for adults 18+ only. Play responsibly and only with money you can afford to lose. If gambling stops being fun, get help — begambleaware.org
Odds may have changed. Predictions are statistical estimates and are not a guarantee of profit. Only operators licensed by UK Gambling Commission are shown.
No player props from licensed bookmakers yet for the upcoming games. They usually appear about a day before tip-off.
Methodology
Minutes: exponential average of games played (half-life 3 games). Per-minute rate: time-weighted (half-life 8 games), lightly shrunk towards the league average. Distribution: negative binomial, with variance that includes minutes uncertainty. Parameters were chosen by backtesting every game this season (each forecast using only earlier data) and probabilities come out calibrated: when the model says 85%, the over landed about 84%. Market probability is the implied probability without the bookmaker’s margin. The model does not know about the opponent, game pace or late news beyond the injury list: a “big disagreement” flag often means the market knows something.