NHL Prediction Model Performance | Accuracy & Calibration Stats

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.

Game Predictions (Multi-Window)

Window	Start	End	Games	Accuracy	Brier	Log Loss	Avg Winner Prob	RMSE Total
last 30	2025-12-14	2026-01-13	227	53.3%	0.2703	0.7951	51.3%	2.531
season to date	2025-10-01	2026-01-13	734	51.6%	0.2609	0.7507	50.9%	2.412
multi season	2023-10-10	2026-01-13	3358	63.5%	0.2299	0.6647	54.4%	2.177

Totals (Over 5.5)

Window	Games	Accuracy	Brier	Log Loss	Avg Outcome Prob
last 30	227	56.4%	0.2527	0.6996	50.8%
season to date	734	55.3%	0.2511	0.6954	51.0%
multi season	3358	61.5%	0.2290	0.6493	53.2%

Playoff Game Performance

Start	End	Games	Accuracy	Brier	Log Loss
2024-04-20	2025-06-17	174	67.8%	0.2224	0.6981

Current Matchups

OTT vs VAN model odds PIT vs TBL model odds WSH vs MTL model odds NJD vs SEA model odds BUF vs PHI model odds NYR vs OTT model odds

Daily Performance (Last 14 Days)

Date	Games	Accuracy	Brier	Log Loss
2025-12-31	10	50.0%	0.2152	0.5960
2026-01-01	8	50.0%	0.2231	0.6386
2026-01-02	4	25.0%	0.5461	4.2034
2026-01-03	13	46.2%	0.2833	0.7841
2026-01-04	5	40.0%	0.2948	0.7876
2026-01-05	5	40.0%	0.3127	0.8239
2026-01-06	10	80.0%	0.1833	0.5383
2026-01-07	5	60.0%	0.3338	0.9241
2026-01-08	12	75.0%	0.2036	0.5917
2026-01-09	3	33.3%	0.2070	0.6025
2026-01-10	14	57.1%	0.2850	0.7667
2026-01-11	5	60.0%	0.2618	0.7189
2026-01-12	9	22.2%	0.3734	1.0122
2026-01-13	10	30.0%	0.4653	—

Calibration (Win Prob Deciles) — Last 30

Bin	Count	Mean Pred	Observed
4	41	48.9%	41.5%
5	89	53.4%	49.4%
6	69	68.8%	56.5%
8	19	86.7%	31.6%
9	9	100.0%	88.9%

Calibration (Win Prob Deciles) — Season To Date

Bin	Count	Mean Pred	Observed
0	1	0.0%	100.0%
4	148	49.2%	48.6%
5	343	53.1%	49.9%
6	185	67.3%	53.5%
7	10	74.7%	50.0%
8	36	85.7%	50.0%
9	11	100.0%	90.9%

Calibration (Win Prob Deciles) — Multi Season

Bin	Count	Mean Pred	Observed
0	13	0.6%	7.7%
1	1	17.5%	0.0%
2	1	27.5%	0.0%
3	3	35.2%	0.0%
4	834	49.0%	32.5%
5	1393	53.1%	54.5%
6	810	67.8%	66.5%
7	38	74.2%	84.2%
8	182	86.3%	78.6%
9	83	99.4%	95.2%

Calibration (Over 5.5) — Last 30

Bin	Count	Mean Pred	Observed
3	1	39.4%	100.0%
4	23	46.4%	47.8%
5	107	55.4%	57.9%
6	92	63.5%	58.7%
7	4	72.4%	0.0%

Calibration (Over 5.5) — Season To Date

Bin	Count	Mean Pred	Observed
3	4	36.4%	100.0%
4	78	46.7%	50.0%
5	372	55.6%	58.6%
6	256	63.7%	52.3%
7	20	72.8%	55.0%
8	3	82.7%	100.0%
9	1	91.4%	100.0%

Calibration (Over 5.5) — Multi Season

Bin	Count	Mean Pred	Observed
2	4	25.2%	0.0%
3	26	36.2%	19.2%
4	369	46.6%	26.0%
5	1682	55.7%	51.0%
6	1131	63.5%	68.3%
7	133	73.4%	91.7%
8	12	83.0%	100.0%
9	1	91.4%	100.0%

Team Calibration (Home, Top 15 by Volume)

Team	Count	Mean Pred	Observed	Bias
FLA	131	63.2%	62.6%	+0.6%
EDM	124	61.6%	63.7%	-2.1%
DAL	123	61.4%	62.6%	-1.2%
CAR	121	66.0%	68.6%	-2.6%
TOR	118	59.4%	59.3%	+0.1%
WSH	115	58.9%	56.5%	+2.3%
WPG	115	62.7%	64.3%	-1.6%
BOS	114	57.7%	53.5%	+4.2%
VGK	113	59.6%	61.1%	-1.5%
COL	112	65.9%	71.4%	-5.6%
NYR	111	61.2%	53.2%	+8.0%
NSH	110	58.3%	50.0%	+8.3%
STL	110	55.3%	56.4%	-1.1%
MIN	109	57.5%	51.4%	+6.1%
LAK	109	60.7%	56.9%	+3.8%

Team Calibration (Pred vs Observed)

Mean Pred Observed

Starter Calibration (Home)

Window	Starter Status	Games	Accuracy	Brier	Log Loss
last 30	Starter	227	53.3%	0.2703	0.7951
season to date	Starter	734	51.6%	0.2609	0.7507
multi season	Unknown	18	66.7%	0.2500	0.6931
multi season	Starter	3340	63.5%	0.2298	0.6646

Cross-Validation (Expanding Window)

Folds: 3 | Brier: 0.2496 | Log Loss: 0.6925 | RMSE Total: 2.405

In-Game Checkpoints — Last 30

Checkpoint	Games	Accuracy	Brier	Log Loss
end_p1	230	65.7%	0.2075	0.5968
end_p2	230	79.1%	0.1463	0.4498
ot_start	55	63.6%	0.2017	0.5745
p3_10	230	82.6%	0.1044	0.3302
p3_5	230	85.2%	0.0835	0.2632
pregame	230	54.3%	0.2483	0.6900

In-Game Checkpoints — Season To Date

Checkpoint	Games	Accuracy	Brier	Log Loss
end_p1	724	66.9%	0.2097	0.6034
end_p2	724	77.6%	0.1475	0.4489
ot_start	187	65.2%	0.1952	0.5583
p3_10	724	83.4%	0.1046	0.3267
p3_5	724	85.5%	0.0870	0.2720
pregame	724	51.5%	0.2493	0.6917

In-Game Calibration — Pregame (Last 30 Days)

Bin	Count	Mean Pred	Observed
4	49	48.9%	44.9%
5	139	52.7%	51.8%
6	42	68.7%	61.9%

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

Bin	Count	Mean Pred	Observed
0	32	5.1%	6.2%
1	7	15.4%	14.3%
2	15	26.1%	20.0%
3	20	33.1%	15.0%
4	18	45.1%	33.3%
5	22	53.4%	40.9%
6	23	64.0%	73.9%
7	20	74.3%	70.0%
8	31	85.3%	83.9%
9	42	94.9%	92.9%

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

Bin	Count	Mean Pred	Observed
0	46	2.2%	2.2%
1	22	15.8%	18.2%
2	16	21.9%	18.8%
3	1	37.3%	100.0%
4	13	44.7%	38.5%
5	21	53.2%	23.8%
6	6	62.7%	66.7%
8	32	86.6%	81.2%
9	73	97.4%	97.3%

xG Model Holdout

Train: 2023-10-10 – 2025-05-13 | Test: 2025-05-13 – 2026-01-06

Shots (test): 60376 | ROC AUC: 0.766 | Log Loss: 0.2271 | Brier: 0.0619

xG Splits — Strength State

Split	Shots	Goal Rate	AUC	Log Loss	Brier
Even	47811	6.2%	0.764	0.2052	0.0546
PP	10736	10.2%	0.691	0.3086	0.0874
PK	1329	6.4%	0.769	0.2088	0.0569
EmptyNet	500	53.6%	0.686	0.6261	0.2208

xG Splits — Shot Type

Split	Shots	Goal Rate	AUC	Log Loss	Brier
wrist	26708	6.7%	0.793	0.2069	0.0559
snap	14811	8.3%	0.751	0.2545	0.0715
slap	6869	5.0%	0.705	0.1865	0.0462
tip-in	5583	6.9%	0.676	0.2402	0.0628
backhand	4441	9.8%	0.792	0.2707	0.0786
deflected	988	11.7%	0.645	0.3405	0.0984
wrap-around	417	3.4%	0.609	0.1510	0.0332
bat	295	10.2%	0.701	0.3079	0.0866
poke	209	14.4%	0.598	0.4121	0.1264
between-legs	35	11.4%	0.742	0.3321	0.0962
nan	17	94.1%	0.406	0.3111	0.0805
cradle	3	0.0%	—	—	0.0161