NHL Prediction Model Performance & Calibration

See how our NHL prediction model performs: accuracy, Brier scores, calibration, and error metrics across thousands of games.

To understand each metric, read Understanding Performance Metrics . To apply this in practice, view today's NHL predictions, our NHL playoff odds, and in-game win probability charts.

How to Read These Metrics

Accuracy: The percentage of games where the predicted winner (team with >50% win probability) actually won. Simple but incomplete — it ignores how confident the model was.
Brier Score: Measures the mean squared error of probability predictions (0–1 scale, lower is better). A coin-flip baseline yields 0.25; our model targets values below 0.24. Brier score rewards well-calibrated confidence levels, not just picking the right side.
Log Loss: A logarithmic scoring rule that heavily penalises confident wrong predictions. Assigning 90% to a team that loses costs far more than assigning 55%. This keeps the model honest about uncertainty.
Calibration: Shows whether stated probabilities match real outcomes. In a well-calibrated model, games given a 70% win probability should be won about 70% of the time. The calibration tables below group predictions into decile bins so you can verify this directly.
RMSE (Root Mean Squared Error): For goal total predictions, RMSE measures how far predicted totals are from actual totals on average, in goal units.

Why calibration matters most: For probabilistic predictions, calibration is more important than raw accuracy. A model that says "55%" every game can be 55% accurate but useless for decision-making. A well-calibrated model tells you how much to trust each prediction. Learn more in our analytics guide and methodology.

Game Predictions (Multi-Window)

Window	Start	End	Games	Accuracy	Brier	Log Loss	Avg Winner Prob	RMSE Total
last 30	2026-01-27	2026-02-26	100	51.0%	0.2576	0.7092	50.5%	2.906
season to date	2025-10-01	2026-02-26	928	51.8%	0.2527	0.6988	50.6%	2.430
multi season	2023-10-10	2026-02-26	3552	55.5%	0.2407	0.6740	51.8%	2.211

Totals (Over 5.5)

Window	Games	Accuracy	Brier	Log Loss	Avg Outcome Prob
last 30	100	55.0%	0.2361	0.6640	52.6%
season to date	928	56.1%	0.2510	0.6956	51.1%
multi season	3552	60.7%	0.2312	0.6542	53.0%

Playoff Game Performance

Start	End	Games	Accuracy	Brier	Log Loss
2024-04-20	2025-06-17	174	55.2%	0.2378	0.6682

Current Matchups

BOS vs CBJ model odds MTL vs NYI model odds OTT vs DET model odds FLA vs BUF model odds WSH vs VGK model odds UTA vs MIN model odds

Daily Performance

Show:

Date	Games	Accuracy	Brier	Log Loss

Prediction Recap Highlights

Definition: High confidence means the model assigned the predicted team a win probability well above 50%. Edges that hit are the highest-confidence correct calls, misses are the highest-confidence incorrect calls, and surprise results show the largest absolute gap between win probability and the actual outcome.

Show:

Biggest Model Edges That Hit

Date	Matchup	Win Prob	Outcome

Biggest Misses (High Confidence)

Date	Matchup	Win Prob	Outcome

Surprise Results

Date	Matchup	Surprise	Outcome

Calibration (Win Prob Deciles) — Last 30

Bin	Count	Mean Pred	Observed
5	45	57.1%	51.1%
6	55	63.5%	50.9%

Calibration (Win Prob Deciles) — Season To Date

Bin	Count	Mean Pred	Observed
4	8	48.5%	37.5%
5	590	55.6%	49.3%
6	330	62.7%	56.1%

Calibration (Win Prob Deciles) — Multi Season

Bin	Count	Mean Pred	Observed
4	57	48.9%	10.5%
5	2257	55.8%	47.8%
6	1233	62.7%	68.0%
7	5	70.5%	100.0%

Calibration (Over 5.5) — Last 30

Bin	Count	Mean Pred	Observed
4	7	46.9%	57.1%
5	49	55.8%	42.9%
6	41	64.3%	68.3%
7	3	73.6%	100.0%

Calibration (Over 5.5) — Season To Date

Bin	Count	Mean Pred	Observed
3	8	35.2%	62.5%
4	85	46.8%	51.8%
5	445	55.8%	58.9%
6	355	63.6%	54.6%
7	34	72.5%	58.8%
8	1	89.6%	100.0%

Calibration (Over 5.5) — Multi Season

Bin	Count	Mean Pred	Observed
1	1	18.4%	0.0%
2	3	26.1%	0.0%
3	34	36.1%	20.6%
4	356	46.6%	29.2%
5	1822	55.7%	50.5%
6	1203	63.3%	69.4%
7	122	72.8%	86.9%
8	10	84.6%	100.0%
9	1	91.0%	100.0%

Team Calibration (Home, Top 15 by Volume)

Team	Count	Mean Pred	Observed	Bias
FLA	137	59.7%	60.6%	-0.9%
EDM	133	60.0%	62.4%	-2.4%
DAL	129	59.3%	63.6%	-4.3%
CAR	128	61.2%	68.8%	-7.6%
TOR	123	58.4%	56.9%	+1.4%
WSH	121	57.8%	57.9%	-0.1%
WPG	120	58.5%	63.3%	-4.9%
VGK	120	58.5%	60.8%	-2.3%
BOS	120	57.5%	55.8%	+1.7%
COL	119	60.3%	68.9%	-8.6%
VAN	117	57.6%	45.3%	+12.3%
NSH	116	58.4%	50.0%	+8.4%
NYR	116	58.3%	51.7%	+6.6%
STL	116	56.8%	56.0%	+0.7%
MIN	115	57.6%	52.2%	+5.4%

Team Calibration (Pred vs Observed)

Mean Pred Observed

Starter Calibration (Home)

Window	Starter Status	Games	Accuracy	Brier	Log Loss
last 30	Starter	100	51.0%	0.2576	0.7092
season to date	Starter	928	51.8%	0.2527	0.6988
multi season	Unknown	18	66.7%	0.2500	0.6931
multi season	Starter	3534	55.5%	0.2407	0.6739

Cross-Validation (Expanding Window)

Summary: 3 folds | Brier: 0.2517 | Log Loss: 0.6971 | RMSE Total: 2.410

Show fold details

Fold	Train N	Val N	Brier	Log Loss	RMSE
Fold 1	701	2,097	0.2542	0.7025	2.449
Fold 2	1,399	1,399	0.2540	0.7020	2.396
Fold 3	2,103	695	0.2468	0.6868	2.386

In-Game Checkpoints — Last 30

Checkpoint	Games	Accuracy	Brier	Log Loss
end_p1	92	70.7%	0.1990	0.5870
end_p2	92	77.2%	0.1663	0.5179
ot_start	21	71.4%	0.1598	0.4703
p3_10	92	85.9%	0.1016	0.3314
p3_5	92	87.0%	0.0802	0.2697
pregame	92	58.7%	0.2365	0.6656

In-Game Checkpoints — Season To Date

Checkpoint	Games	Accuracy	Brier	Log Loss
end_p1	916	66.7%	0.2097	0.6050
end_p2	916	77.7%	0.1533	0.4662
ot_start	234	63.2%	0.2008	0.5694
p3_10	916	84.0%	0.1048	0.3303
p3_5	916	85.4%	0.0878	0.2766
pregame	916	52.3%	0.2514	0.6960

In-Game Calibration — Pregame (Last 30 Days)

Bin	Count	Mean Pred	Observed
5	59	57.8%	50.8%
6	33	61.7%	72.7%

In-Game Calibration — End P2 (Last 30 Days)

Bin	Count	Mean Pred	Observed
0	12	5.3%	33.3%
1	3	17.2%	66.7%
2	9	25.2%	0.0%
3	11	33.4%	36.4%
4	2	41.6%	0.0%
5	7	56.7%	57.1%
6	6	64.9%	50.0%
7	13	75.8%	69.2%
8	9	85.3%	88.9%
9	20	94.2%	100.0%

In-Game Calibration — P3 10 (Last 30 Days)

Bin	Count	Mean Pred	Observed
0	17	2.9%	11.8%
1	11	15.4%	18.2%
2	8	23.1%	12.5%
4	9	45.0%	55.6%
5	9	55.2%	66.7%
6	2	63.2%	100.0%
8	7	87.3%	100.0%
9	29	96.8%	100.0%

xG Model Holdout

Train: 2023-10-10 – 2025-10-28 | Test: 2025-10-28 – 2026-02-05

Shots (test): 64343 | ROC AUC: 0.763 | Log Loss: 0.2287 | Brier: 0.0623

xG Splits — Strength State

Split	Shots	Goal Rate	AUC	Log Loss	Brier
Even	51324	6.2%	0.763	0.2059	0.0548
PP	11119	10.4%	0.676	0.3162	0.0896
PK	1365	6.9%	0.773	0.2181	0.0605
EmptyNet	535	52.5%	0.703	0.6293	0.2204

xG Splits — Shot Type

Split	Shots	Goal Rate	AUC	Log Loss	Brier
wrist	27938	6.8%	0.787	0.2094	0.0565
snap	16061	8.3%	0.756	0.2531	0.0711
slap	7487	5.2%	0.701	0.1935	0.0485
tip-in	6046	6.9%	0.673	0.2375	0.0618
backhand	4739	10.0%	0.782	0.2767	0.0806
deflected	1055	10.7%	0.640	0.3243	0.0922
wrap-around	440	4.5%	0.628	0.1813	0.0429
bat	311	9.6%	0.739	0.2900	0.0814
poke	195	12.8%	0.621	0.3662	0.1078
between-legs	44	13.6%	0.675	0.3678	0.1071
nan	22	95.5%	0.405	0.2556	0.0636
cradle	5	0.0%	—	—	0.0180

Monthly Performance Trends

Track how model performance varies month-to-month across the season.

Month	Games	Accuracy	Brier	Log Loss
2023-10	140	58.6%	0.2467	0.6866
2023-11	213	54.0%	0.2404	0.6735
2023-12	219	58.9%	0.2328	0.6580
2024-01	208	53.4%	0.2363	0.6644
2024-02	172	51.2%	0.2384	0.6690
2024-03	228	59.2%	0.2264	0.6445
2024-04	132	58.3%	0.2326	0.6573
2024-10	166	58.4%	0.2436	0.6803
2024-11	220	54.5%	0.2402	0.6732
2024-12	214	57.9%	0.2279	0.6476
2025-01	224	56.7%	0.2347	0.6618
2025-02	122	53.3%	0.2497	0.6923
2025-03	234	60.3%	0.2352	0.6630
2025-04	132	60.6%	0.2356	0.6638
2025-10	180	53.9%	0.2485	0.6901
2025-11	225	52.0%	0.2522	0.6976
2025-12	226	50.4%	0.2547	0.7027
2026-01	240	52.1%	0.2528	0.6988
2026-02	57	49.1%	0.2603	0.7149

Playoff Model Performance

Game-level and series-level accuracy across playoff rounds.

Playoff Games

Round	Games	Accuracy	Brier	Log Loss
All Rounds	86	58.1%	0.2348	0.6624
Round 1	47	63.8%	0.2282	0.6492
Round 2	23	60.9%	0.2333	0.6594
Round 3	10	30.0%	0.2627	0.7186
Round 4	6	50.0%	0.2454	0.6839

Playoff Series

Round	Series	Accuracy	Brier	Log Loss
All Rounds	15	53.3%	0.2386	0.6702
Round 1	8	50.0%	0.2390	0.6709
Round 2	4	75.0%	0.2265	0.6459
Round 3	2	50.0%	0.2486	0.6904
Round 4	1	0.0%	0.2643	0.7218

Playoff Calibration (Pred vs Observed)

Mean Pred Observed