Post-Week-3 player ratings -> subject-resolved tossup-contest simulation on each team's most-recent actual lineup -> home-court category ban -> Normal margin/total. Model-only unless a live per-game moneyline is open (then blended 70/30); the conference-winner FUTURES fallback was DROPPED after the Round-3 backtest showed it hurt (model-only beat the blend on direction 83% vs 50%). Spreads on odd integers, totals on .5 (never push).
Week 4
cryo@(Stitch)^2
| cryo | (Stitch)^2 (home) |
|---|
| Win prob | 55% | 45% |
| Spread | cryo -7 |
| Total (O/U) | 162.5 |
home most-likely bans Chemistry
Week 4
mingle/Yunyi@dan.k.memes
| mingle/Yunyi | dan.k.memes (home) |
|---|
| Win prob | 44% | 56% |
| Spread | dan.k.memes -9 |
| Total (O/U) | 158.5 |
home most-likely bans Math
Week 4
czz@jonathan
| czz | jonathan (home) |
|---|
| Win prob | 52% | 48% |
| Spread | czz -3 |
| Total (O/U) | 162.5 |
home most-likely bans Math
Week 4
James W@ChessFun
| James W | ChessFun (home) |
|---|
| Win prob | 54% | 46% |
| Spread | James W -5 |
| Total (O/U) | 161.5 |
home most-likely bans Chemistry
Week 4
xpoes@sumin
| xpoes | sumin (home) |
|---|
| Win prob | 40% | 60% |
| Spread | sumin -15 |
| Total (O/U) | 155.5 |
home most-likely bans Earth/Space
Week 4
GidTheKid2@Fez_Keyreb
| GidTheKid2 | Fez_Keyreb (home) |
|---|
| Win prob | 61% | 39% |
| Spread | GidTheKid2 -15 |
| Total (O/U) | 154.5 |
home most-likely bans Physics
Method
Post-Week-3 player ratings -> subject-resolved tossup-contest simulation on each team's most-recent actual lineup -> home-court category ban -> Normal margin/total. Model-only unless a live per-game moneyline is open (then blended 70/30); the conference-winner FUTURES fallback was DROPPED after the Round-3 backtest showed it hurt (model-only beat the blend on direction 83% vs 50%). Home removes one tossup of the category that most helps it; soft softmax over bans (lambda=1, a prior until bans are observed).
Caveats
- LOW CONFIDENCE. Over Weeks 2-3 the model's game-level direction was ~58% (n=12) -- barely better than a coin flip. These are a sharpened prior, not a proven edge.
- Each team is projected with its MOST RECENT actual lineup (who dressed last game), not the full drafted roster -- captures no-shows. Unmapped-but-present players are padded with roster slots so identity gaps don't fake a no-show; if you know a lineup change, set MANUAL_LINEUP and re-run.
- Model-only this week (no per-game moneylines open yet). The conf-winner futures blend was dropped: on Round 3 it dragged model picks toward season favorites (czz, anurag) who then lost.
- 3 weeks of game data; ratings still lean partly on the combine+draft seed. Byes this week: Connor Chang (ponnor), anurag (seanjay).
- Real NSBA margins are huge (avg ~45) and hard to predict directionally, so spreads are kept modest on purpose -- widening a coin-flip just amplifies wrong picks. The gain (2.5x) only avoids pick'em.
- Totals nudged +4 vs the raw model: they ran 8-4 OVER in Weeks 2-3 (median actual 161 vs line 156), so the baseline was centered on the mean but low on the median. Totals still lack matchup dispersion (all land ~155-160) -- can't yet tell a shootout from a grind.