Skip to content

NHL model accuracy

National Hockey League

Every final NHL game, forecast with only what was known the day before and then scored. Lower Brier and log loss are better; a coin flip scores 0.250 Brier. All leagues

Scorecard

Accuracy of each model on final games
ModelGamesBrierLog lossAccuracySkillCalib. error
Game simulator (backtest)2410.2450.68355.6%+1.2%1.7%
Elo2410.2440.68157.3%+1.5%2.6%
Base rate2410.2480.68956.4%—1.0%
Published forecasts51small sample0.2450.68247.1%+4.2%14.0%

Calibration

When a model gives the home side a 60% chance, does it win about 60% of the time? Dots on the dashed line mean yes; whiskers show the 90% range the observed rate could land in by chance.

Brier score by month

Early-season forecasts lean on last season's regressed ratings, so expect them to improve as games accumulate.

Head to head

Game simulator (backtest) vs Elo

241 games both forecast

Brier
0.245 vs 0.244
Log loss
0.683 vs 0.681
Accuracy
55.6% vs 57.3%

Elo has the lower Brier score here.

Published forecasts vs Game simulator (backtest)

51 games both forecast

Brier
0.245 vs 0.249
Log loss
0.682 vs 0.690
Accuracy
47.1% vs 47.1%

Published forecasts has the lower Brier score here, but on too few games to call.

Market consensus isn't connected: it needs a licensed odds source, which an admin can import. Until then the base rate and Elo are the benchmarks.

How this is measured

The backtest replays the game simulator for every final game from the efficiency ratings and Elo as they stood before that date, the league's scoring and home advantage so far, rest and travel, the venue and the game's weather. Injuries and lineups of past games aren't on record, so the replay is team level; the published forecasts, which include them, are scored separately as they accumulate.

Brier = ½ Σ (probability − outcome)² over home, away. Log loss = −ln(probability given to the result), with probabilities clipped at 0.5%. Tied finals are left out. Full details are in the model methodology document.

Pick record

Every forecast the site published before a game, scored once it finished. A pick is the likeliest outcome. A well-calibrated model's 70% picks come in about 70% of the time, so compare each row's hit rate with its confidence.

Record

24–27 47%

Average confidence 55%

Hit rate by confidence
Confidence PicksHit
Under 60%4743%
60%–70%4100%
Latest picks
GamePickResult
Nov 6 CBJ @ BUFBUF 56%2–5
Nov 6 SJS @ MINMIN 63%1–4
Nov 6 ANA @ CGYCGY 54%5–3
Nov 1 WPG @ CARCAR 61%0–7
Nov 1 MIN @ WSHWSH 52%5–1
Nov 1 UTA @ PHIUTA 52%3–6
Oct 31 CHI @ BOSBOS 62%4–5
Oct 31 STL @ DETDET 55%5–4
Oct 31 EDM @ NYINYI 51%4–0
Oct 31 TBL @ NSHNSH 54%4–5
Oct 31 OTT @ SJSSJS 51%5–1
Oct 31 BUF @ LAKLAK 55%3–2
Oct 31 TOR @ VANTOR 54%4–2
Oct 31 PIT @ MTLMTL 56%4–5
Oct 31 CBJ @ DALDAL 58%1–3
Oct 31 CGY @ COLCOL 59%2–5
Oct 30 NJD @ VGKVGK 55%5–2
Oct 30 WPG @ FLAFLA 58%2–5
Oct 30 SEA @ CGYCGY 51%2–6
Oct 29 CBJ @ CARCAR 60%2–3

Data as of Source: NHL.com

Cookies

Cookies keep you signed in. Optional ones remember your settings. No ads or tracking. Cookie policy