NBA · Research
BenchIQ — Bench Production
Per-team bench scoring, rebounding, assists, and minutes share over the last 14 days, plus a depth score (bench points per 100 bench minutes vs league average). Deep benches cover spreads on back-to-backs; thin benches collapse late.
Sport NBA
MLB · Coming as BullpenIQ
What this is. Sort teams by their bench's actual contribution. Depth score is the headline column — bench points per 100 bench minutes, vs league average. Positive means efficient bench production; negative means the bench struggles to score even when on the floor. Not a betting recommendation; not picks.
BenchIQ isn't published yet. The BenchIQ feed for the NBA has not been written. This page fills in once the first daily build lands.
How this works
For each game in the 14-day window, each team's players are sorted by minutes played. The top 5 are treated as starters by minutes-rank; players ranked 6th and below are bench. Bench season totals are summed across all valid games, then normalized per-game.
- Bench definition — minutes-rank within each game, not the box-score "starter" flag. A starter who plays only 8 minutes in foul trouble still counts as a starter; what matters here is "who came off the bench tonight and produced."
- Depth score — bench points per 100 bench minutes. Normalizes for teams whose bench gets fewer minutes (good benches with low minutes vs bad benches with high minutes both surface honestly).
- Minutes share — bench minutes / total team minutes. Higher share = team relies more on bench; lower = top-heavy.
- Distinct — unique bench-rotation players across all logged games. Higher = deeper bench rotation; lower = same few players coming off the bench every night.
- Completeness pill — High (5+ games), Medium (3–4), Low (1–2), Insufficient (0).
Refreshes with each NBA pipeline cycle. Source feed: public/nba/bench.json, built by scripts/basketball/bench.py --league nba.