Player Prop Sheet — NBA Oct 31 · NFL Nov 1–2
How the confidence meter works
Each pick gets a model chance of hitting, then a 1–10 score: score = 1 + (chance − 35%) × 20, so 55% ≈ 5, 65% ≈ 7, 80%+ = 10. Players with fewer than 6 logged games are capped at 5. The chance is half hit-rate history and half projection, plus small injury nudges.
Hit-rate history: last 10 games (45%), this season (30%), last season's most recent 20 games (25%). Projection: weighted average of this season, last season and two seasons ago, multiplied by a matchup factor (what this opponent allows to the player's position for this stat, from 2026 box scores, shrunk toward league average, ±15% max) and a usage factor (minutes for NBA; targets, carries or pass attempts for NFL over the last 3 games vs earlier, ±10% max). Injuries (ESPN report): Questionable −6 points, Day-To-Day −4; Out/Doubtful/IR players left off. A significant teammate newly out adds +3 points to overs (−3 to unders).
Best +500 parlay picks the combination the model rates most likely to land while paying at least +500, one leg per player. Model chance multiplies legs as if independent; same-game legs are correlated, so treat it as rough. Gaps: no NFL snap counts (targets/carries/attempts used instead); NFL defensive injuries not in matchups; NBA matchup samples only 4–6 games per team.
Sources: FanDuel Sportsbook public odds feed (pulled ); ESPN game logs, box scores, injury reports and rosters.