🏒 NHL Model v1 2026-10-16 · 2 games

📋 Model season record 11 games graded

Moneyline: 9-2 (82% · avg prob 58%)
Puck line: 4-7 (36%)
Model's side pick per game, logged before puck drop and graded on finals. For calibration, not betting advice.

💎 Value board model's confident disagreements with Vegas

💎 SEA@TBL away — model 42% vs DK +225 (30%) · gap 12%
💎 BUF@TOR away — model 58% vs DK -102 (48%) · gap 10%
💎 VAN@FLA away — model 36% vs DK +240 (28%) · gap 8%
💎 NYI@OTT away — model 48% vs DK +136 (41%) · gap 7%
💎 CGY@VGK away — model 36% vs DK +230 (29%) · gap 7%
💎 PHI@DET home — model 62% vs DK -142 (56%) · gap 6%
Value board record: 2-4 (33% · avg prob 47%)
Only games where our win prob beats DK's no-vig implied prob by 5+ pts, ranked by gap. Dogs and favorites both qualify — confidence is the size of the disagreement. Experimental until ~100 graded plays.

📈 Line movement DK moneyline trail, snapshotted every ~2h to puck drop

PIT@COL (+0.0pp to home since 09:39)
+160/-192 09:39
away/home moneyline at each snapshot
BOS@ANA (+0.0pp to home since 09:39)
-118/-102 09:39
away/home moneyline at each snapshot
PIT @ COLmodel: COL 59%
Win prob (home)
59%
Vegas lines (DK)
ML: PIT +160 · COL -192
Puck: PIT +1.5 (-130) · COL -1.5 (+110)
O/U 6.5 (o -118 / u -102)
Cover −1.5
H 36% · A 24%
Goalies
TBD
Drivers
xG diff -0.046 · goalie +0.16
BOS @ ANAmodel: BOS 53%
Win prob (home)
47%
Vegas lines (DK)
ML: BOS -115 · ANA -105
Puck: BOS -1.5 (+210) · ANA +1.5 (-258)
O/U 6.5 (o +105 / u -125)
Cover −1.5
H 29% · A 31%
Goalies
TBD
Drivers
xG diff +0.004 · goalie -0.66
Methodology (honest version): P(home win) from 5v5 xG differential, goalie quality (goals saved above expected), back-to-back flags, and home ice — logistic, walk-forward fitted on 2022–26 (3,854 games, 56.6% accuracy vs 54.3% baseline). Puck line P(cover −1.5) = P(win) × empirical P(win by 2+ | win) — 40% of NHL games end exactly 1-0/2-1/etc., and empty-netters make 3-goal margins more common than 2-goal ones. Early-season xG is regressed 8 games toward last season; goalie numbers are shrunk hard in October. Totals are excluded — team xG rates failed to beat the league average in backtest. Model-vs-DK gaps are shown for information, not as bets.