How value is built
1 · Per-game rates. Games played range from 2 to 9, so season totals would just rank whoever showed up most. Rates level the field first.
2 · Reliability. Rates regress toward replacement level — the 25th percentile of the pool — by sample size. A 30-point average over three games is a weaker claim than the same average over nine. The baseline is replacement level rather than the league mean on purpose: regressing toward the mean quietly rewards a below-average player for missing games, which punishes exactly the people who keep showing up.
adj = (GP·rate + 2·replacement) / (GP + 2)
3 · Composite. Ten categories, standardized across the pool so a rebound and an assist carry comparable weight, then weighted below. Deflections and Net PPP (on-court point differential per possession) come from Hoopsalytics' advanced views rather than the box score; turnovers are scored as TO% (turnovers per 100 plays) instead of a raw per-game count, since raw count just rewards low usage.
4 · Availability. Attendance against the 9 games of recorded stats, scaled into the final value.
×1.12 at 9/9 ×0.96 at 5/9
×0.84 at 2/9 ×0.80 at 1/9
Teams forfeit when they can't field a roster, so showing up carries real value on its own — but it's deliberately tuned as a secondary factor. It moves players between tiers without letting a role player outrank an elite producer.
What availability changed
Rank shift versus a production-only board:
Jeffrey Luman 7 GP↑ 4
Samim Jabarkhail 7 GP↑ 4
Jon Fiorillo 7 GP↑ 4
Josh Carter 2 GP↓ 7
Dan Greif 3 GP↓ 6
The top five move by at most one spot — production still decides the premium picks.
Data caveat
Every team has played 8 games and the stat table carries all 8 — no lag this snapshot, unlike the prior one. Per-event box scores are still published empty, so every rate here comes from season totals divided by games played.