Post-Week-2 player ratings -> subject-resolved tossup-contest simulation -> home-court category ban -> Normal margin/total, then blended 70/30 model/market. The market leg uses an OPEN per-game moneyline if one exists, otherwise the conference-winner FUTURES (Bradley-Terry on the standings-aware futures odds) -> the market's read on team strength enters every line. Spreads on odd integers, totals on .5 (never push).
Week 3
(Stitch)^2@Connor Chang
| (Stitch)^2 | Connor Chang (home) |
|---|
| Win prob | 43% | 57% |
| Spread | Connor Chang -11 |
| Total (O/U) | 161.5 |
home most-likely bans Math
Week 3
sumin@anurag
| sumin | anurag (home) |
|---|
| Win prob | 52% | 48% |
| Spread | sumin -3 |
| Total (O/U) | 155.5 |
home most-likely bans Biology
Week 3
ChessFun@jonathan
| ChessFun | jonathan (home) |
|---|
| Win prob | 58% | 42% |
| Spread | ChessFun -11 |
| Total (O/U) | 158.5 |
home most-likely bans Biology
Week 3
xpoes@GidTheKid2
| xpoes | GidTheKid2 (home) |
|---|
| Win prob | 40% | 60% |
| Spread | GidTheKid2 -15 |
| Total (O/U) | 151.5 |
home most-likely bans Earth/Space
Week 3
dan.k.memes@Fez_Keyreb
| dan.k.memes | Fez_Keyreb (home) |
|---|
| Win prob | 56% | 44% |
| Spread | dan.k.memes -7 |
| Total (O/U) | 151.5 |
home most-likely bans Biology
Week 3
cryo@czz
| cryo | czz (home) |
|---|
| Win prob | 46% | 54% |
| Spread | czz -5 |
| Total (O/U) | 158.5 |
home most-likely bans Biology
Method
Post-Week-2 player ratings -> subject-resolved tossup-contest simulation -> home-court category ban -> Normal margin/total, then blended 70/30 model/market. The market leg uses an OPEN per-game moneyline if one exists, otherwise the conference-winner FUTURES (Bradley-Terry on the standings-aware futures odds) -> the market's read on team strength enters every line. Home removes one tossup of the category that most helps it; soft softmax over bans (lambda=1, a prior until bans are observed).
Caveats
- Each line = 70% model / 30% market. Market leg = open per-game moneyline if any, else the conference-winner futures (the market's standings-aware team-strength read) via Bradley-Terry, shrunk toward 50% since season odds overstate one-game edges.
- (Stitch)^2 @ Connor Chang already played (market resolved); it is shown for completeness off the pre-game numbers.
- Each team is projected with its MOST RECENT actual lineup (who dressed last game), not the full drafted roster -- this captures no-shows. xpoes is short with Rohan G OUT (confirmed out again Week 3) -> the GidTheKid2 line is wide by design. Unmapped-but-present players are padded with roster slots so identity gaps don't fake a no-show.
- FIRST LIVE WEEK of this model version (projected lineups + 70/30 futures-market blend). It fixes the prior version's under-dispersion, but the blend weight and lineup-continuity assumption are not yet backtested -- treat as a sharpened prior, not a proven edge.
- 2 weeks of game data: player ratings still lean partly on the combine+draft seed.
- The model leg is widened by a confidence factor (gain 2.5x) so lines aren't pick'em; totals are NOT widened. Win/loss STANDINGS enter only through the futures blend (and via player-rating updates), not as a direct record-based feature.