What is expected goals (xG) in hockey?

Expected goals (xG) estimates the likelihood a shot becomes a goal based on factors like location, angle, shot type, and game state.

How should I use xG to compare teams?

Compare xG for and against per game to see which teams create and allow the best shot quality, alongside finishing rates.

Why can goals differ from xG?

Goals can outpace xG due to hot shooting, goalie performance, or small-sample variance, especially in short windows.

Expected Goals (xG) in Hockey | NHL

Expected Goals (xG) Analysis

Expected Goals (xG) measures shot quality by predicting the probability each shot becomes a goal based on distance, angle, shot type, and game situation. Our xG model is trained on thousands of NHL shots. The tables below use our neutral xG model, which evaluates shot quality independent of the current score — the best lens for comparing teams and players on a level playing field.

Key Metrics:
• xG/G: Expected goals per game (offensive threat)
• xGA/G: Expected goals against per game (defensive exposure)
• Net xG/G: xG/G minus xGA/G (overall xG dominance, >0 is good)
• xG%: Team's share of total xG (xGF / (xGF + xGA) × 100)
• Finishing Rate: Goals / xG (>1.0 = above expectation)
• GSAx: Goals Saved Above Expected (xGA - GA, positive = good goaltending)

Team xG Rankings (2025-26)

Click column headers to sort. Net xG/G >0.3 in green, <0 in red. GSAx >0 = good goaltending.

Team	GP	xG/G	xGA/G	Net xG/G	xG%	Finish	GSAx

Shot Quality by Shot Type (2025-26)

Shot Share and xG Share are league-wide percentages for each shot type.

Shot Type	Shots	Shot Share	xG	xG Share	xG/Shot	Goal Rate

Shot Quality by Zone (2025-26)

Zone codes follow NHL play-by-play conventions (OZ, NZ, DZ).

Zone	Shots	Shot Share	xG	xG Share	xG/Shot	Goal Rate

Player xG Leaderboard (2025-26)

Click column headers to sort. xG/Shot = total xG / total shots (shot quality).

Player	Team	GP	Shots	xG	Goals	S%	xG/Shot

S% = Shooting Percentage | xG/Shot = Expected Goals per shot attempt

Team Finishing Rates (2025-26)

Finishing Rate = Goals / xG. >1.1 = overperforming (green), <0.9 = underperforming (red). GSAx = Goals Saved Above Expected (positive = good goaltending).

Team	Goals	xG	Finishing Rate	GSAx	GSAx/G

Player Finishing Leaderboard (2025-26, min 50 shots)

Δ = Goals − xG. Positive (green) = overperforming xG, negative (red) = underperforming. Click column headers to sort.

Player	Team	G	xG	Δ	Sh%	xG/Shot	GP

Player Spotlight

🌟 xG Leaders

Player	Team	Metric	Value
Nathan MacKinnon	COL	xG	51.8
Connor McDavid	EDM	xG	51.7
Cole Caufield	MTL	xG	47.4

⇧ Overperformers

min 50 shots

Player	Team	Metric	Value
Cole Caufield	MTL	G−xG	+13.6
Cutter Gauthier	ANA	G−xG	+13.4
Morgan Geekie	BOS	G−xG	+12.9

⇩ Underperformers

min 50 shots

Player	Team	Metric	Value
Tomas Hertl	VGK	G−xG	-14.1
Anders Lee	NYI	G−xG	-12.7
Brady Tkachuk	OTT	G−xG	-11.3

Top 15 Most Dramatic Games (Comeback Index)

Comeback Index = max win-probability swing from a team's low point to final. Max Swing = largest single-event change in home win probability. Higher = more drama.

Date	Teams	Comeback	Max Swing	Result
2025-11-03	PIT @ TOR	0.985	0.635	TOR 4–3 PIT
2026-01-22	OTT @ NSH	0.978	0.640	NSH 5–3 OTT
2025-10-25	STL @ DET	0.978	0.882	DET 6–4 STL
2026-03-08	BOS @ PIT	0.977	0.604	PIT 5–4 BOS
2026-01-17	MTL @ OTT	0.977	0.813	MTL 6–5 OTT (away win)
2026-02-02	STL @ NSH	0.973	0.598	NSH 6–5 STL
2026-01-01	WPG @ TOR	0.972	0.844	TOR 6–5 WPG
2025-12-30	MTL @ FLA	0.970	0.553	MTL 3–2 FLA (away win)
2026-01-31	CAR @ WSH	0.969	0.575	WSH 4–3 CAR
2025-11-15	BUF @ DET	0.966	0.532	BUF 5–4 DET (away win)
2026-03-04	VGK @ DET	0.966	0.719	VGK 4–3 DET (away win)
2026-01-29	UTA @ CAR	0.965	0.798	CAR 5–4 UTA
2026-04-14	PIT @ STL	0.963	0.751	STL 7–5 PIT
2026-03-08	TBL @ BUF	0.962	0.789	BUF 8–7 TBL
2026-01-04	PIT @ CBJ	0.962	0.706	PIT 5–4 CBJ (away win)

Based on in-game win probability curves from the 2025–26 NHL season. View in-game dashboards →

Download Data

Game xG Summary (CSV) Player xG Leaderboard (CSV) Team xG Summary (CSV) Shot Type Breakdown (CSV) Zone Breakdown (CSV)

About the xG Model: Our expected goals model uses gradient boosting on shot distance, angle, shot type, game situation (even/power play), rebounds, and rush attempts. The leaderboards above use the neutral variant, which excludes score context so that shot quality comparisons are not skewed by game state. A separate contextual variant (which factors in score differential) is shown per-game on the recent games xG page and on the model performance page.

View: Today's Predictions | Model Performance | Playoff Odds