Edgehalla

How Hockey ChangedThe Vanishing Game

Shots on Goal Fell. Shooting Didn’t.

NHL games now have eight fewer shots on goal than in 2017-18, the fewest since 2001-02, while scoring rose above six goals a game. The obvious story is that teams stopped shooting. Every attempt in our play-by-play says otherwise.

Verdict: Data story

By Edgehalla ResearchPublished 7 min read

More takeaways below

  • Part of it is probably how shots are recorded. The drop is concentrated in 2023-24 and 2024-25, and recorded misses rose in 29 of 30 rinks. League save % (.889) is the lowest since 1992-93 partly for the same reason.
Use the toolNHL goalie hubStarters, save percentage and workload for every goalie.
On this page

Did NHL teams stop shooting?

No. Count every attempt, on goal, missed or blocked, and a 2025-26 game had 116.6 of them, against 117.8 in 2017-18. What changed is where those attempts ended: 63.9 reached the goalie in 2017-18 and 55.7 in 2025-26.

Attempts held steady; shots on goal fell and misses rose

Per game, both teams, all situations, NHL regular season (our play-by-play)

Show data
Attempts went from 117.8 to 116.6 a game from 2017-18 to 2025-26 while shots on goal fell from 63.9 to 55.7 and misses rose from 24.4 to 29.8.
SeasonAll attemptsOn goalBlockedMissed
2017-18117.8 a game63.9 a game29.5 a game24.4 a game
2018-19115.3 a game62.9 a game28.5 a game23.8 a game
2019-20113.5 a game62.8 a game27.4 a game23.3 a game
2020-21108.2 a game60.0 a game26.0 a game22.2 a game
2021-22113.3 a game63.3 a game27.0 a game23.1 a game
2022-23116.4 a game62.5 a game29.4 a game24.5 a game
2023-24121.6 a game60.6 a game34.1 a game27.0 a game
2024-25119.0 a game56.6 a game33.4 a game29.0 a game
2025-26116.6 a game55.7 a game31.2 a game29.8 a game
Every shot attempt in the NHL play-by-play, shootouts excluded: 11,052 regular-season games.

Missed shots rose by about 5.3 a game and blocked shots by 1.7. With about 1.2 fewer attempts, that accounts for the 8.2 missing shots on goal. The NHL’s own five-on-five shot attempt counts, published from 2009-10, tell the same story: attempts rose to about 95 a game in 2023-24 and 2024-25, while the share on goal fell.

The share of attempts on goal broke in 2023-24

At five on five, between 52.9% and 55.2% of attempts reached the goalie in every season from 2009-10 to 2022-23. Then it dropped to 49.0% in 2023-24 and 46.6% in 2024-25, and stayed there in 2025-26. Fourteen flat seasons and then a two-season cliff is not how playing styles usually change.

Share of five-on-five attempts that reach the goalie

NHL regular season, from the first season the league published five-on-five attempts

Show data
The share of five-on-five attempts on goal held between 52.9% and 55.2% from 2009-10 to 2022-23, then fell to 46.6% in 2024-25.
SeasonOn goal ÷ attempts, 5-on-5
2009-1054.4%
2010-1153.7%
2011-1253.6%
2012-1352.9%
2013-1453.5%
2014-1553.1%
2015-1653.3%
2016-1753.5%
2017-1853.7%
2018-1953.9%
2019-2054.8%
2020-2154.9%
2021-2255.2%
2022-2353.1%
2023-2449.0%
2024-2546.6%
2025-2646.6%
  • 2023-24: Teammate blocks counted
NHL five-on-five shot attempts (SAT) and shots on goal, all teams.

Two analysts found the same break and both point at the scorekeeping. Hockey Analysis showed the crease miss rate roughly doubling after 2021-22 and noted the NHL began counting shots blocked by the shooter’s own teammates in 2023-24. Puck Over The Glass found 2023-24 had by far the biggest one-year change in where shots were recorded, and suspects the league’s puck-tracking system began feeding the play-by-play, a rollout it thinks likely started in 2022-23. We found no NHL announcement of a change.

Where the misses come from

Split unblocked attempts by distance and the on-goal share fell at every range, by 5 to 12 points. Most of the extra misses come from close to the net: from inside 20 feet, 75.8% reached the goalie in 2017-18 and 65.8% in 2025-26. Misses from there went from 5.94 to 9.16 a game.

Fewer close attempts reach the goalie

Share of unblocked attempts on goal, by distance from the net

From inside 20 feet the on-goal share fell from 75.8% to 65.8%; from 40 to 60 feet it fell from 69.6% to 64.7%.

0-20 ft
75.8%
65.8%
20-40 ft
71.4%
65.5%
40-60 ft
69.6%
64.7%
60+ ft
75%
63.3%
  • 2017-18
  • 2025-26
Our play-by-play, regular season, shootouts excluded. Distance from the recorded shot location to the goal line centre.

Some of it may be shot choice. Unblocked attempts from 40 to 60 feet fell from 27.2 to 20.7 a game, a slide that began well before 2023-24, while attempts inside 20 feet rose from 24.6 to 26.8. Our expected goals per unblocked attempt rose from 0.066 to 0.073, and actual goals per unblocked attempt from 6.63% to 7.21%, most of it before 2023-24. Teams take slightly better shots. Shot locations come from the same feed, though: attempts from beyond 60 feet jumped in 2023-24, the season of the recording break.

Is it the home scorer?

Shots are logged by staff in each building, and rinks differ: our arena scorer bias work shows how much hits vary by home scorer. If a few buildings changed habits, the misses would rise in a few rinks. Instead, missed shots recorded per game rose in 29 of 30 rinks between 2017-19 and 2024-26, by a median of 5.1 a game.

Almost every rink now records more misses

Missed shots per game recorded in each home rink, both teams

Show data
Missed shots per game rose in 29 of 30 home rinks from 2017-19 to 2024-26, by a median of 5.1 a game.
Missed shots per game, 2017-19Missed shots per game, 2024-26
PIT2131
NYI2131
CHI1928
COL2130
VAN2230
FLA2331
BUF2330
MTL2330
CBJ2330
WPG2431
VGK2330
NYR2329
BOS2430
DET2329
NJD2429
TBL2328
MIN2428
ANA2933
STL2226
EDM2428
LAK2731
NSH2327
DAL2427
PHI2427
WSH2731
CAR2931
OTT2427
SJS2527
CGY2728
TOR3028
  • No change: from (19, 19) to (34, 34)
Each dot is one home rink (82 games in each window: 2017-18 and 2018-19 against 2024-25 and 2025-26). Dots above the line record more misses now. Seattle and Utah (no 2017-19 games) and Arizona (none in 2024-26) are left out.

A league-wide shift in every building fits either explanation: players everywhere missing more, or one league-wide change in how misses are logged. Toronto is the outlier that shows rinks still differ. It already logged about 30 misses a game in 2017-19 and was the only building to log fewer since.

What it does to save percentage

Save percentage is the share of shots on goal that are not goals. If attempts that used to be logged as saves are now logged as misses, save percentage falls even if goalies are just as good. League save % was .911 in 2013-14 and .889 in 2025-26, the lowest since 1992-93. (This counts empty-net goals; the goalie-only figure, .896, is the lowest since 1993-94.)

League save percentage since 1959-60

1 − goals ÷ shots on goal, empty-net goals included, NHL regular season

Show data
League save percentage was .911 in 2013-14 and 2014-15 and fell to .889 in 2025-26, the lowest since 1992-93.
SeasonSave %
1959-600.907
1960-610.907
1961-620.905
1962-630.907
1963-640.915
1964-650.908
1965-660.904
1966-670.906
1967-680.908
1968-690.907
1969-700.911
1970-710.902
1971-720.901
1972-730.895
1973-740.895
1974-750.889
1975-760.889
1976-770.889
1977-780.887
1978-790.881
1979-800.881
1980-810.874
1981-820.871
1982-830.874
1983-840.871
1984-850.873
1985-860.872
1986-870.878
1987-880.878
1988-890.877
1989-900.878
1990-910.884
1991-920.886
1992-930.883
1993-940.893
1994-950.898
1995-960.896
1996-970.902
1997-980.904
1998-990.906
1999-000.902
2000-010.901
2001-020.905
2002-030.906
2003-040.909
2004-05–
2005-060.899
2006-070.903
2007-080.906
2008-090.906
2009-100.909
2010-110.910
2011-120.911
2012-130.909
2013-140.911
2014-150.911
2015-160.910
2016-170.910
2017-180.908
2018-190.905
2019-200.905
2020-210.903
2021-220.902
2022-230.899
2023-240.898
2024-250.894
2025-260.889
  • 2023-24: Shot-recording break
Shots on goal are published from 1959-60 in this feed; counts before 1983-84 are less reliable. The 2004-05 season was cancelled.

Show the work

Tests written down
Not in advance (exploratory)
Results computed

League shots on goal for every season since 1959-60 and five-on-five attempts since 2009-10 from the NHL stats API, checked against every attempt in our play-by-play (11,052 regular-season games, 2017-18 to 2025-26). The rink check compares each home building’s recorded misses in 2017-19 and 2024-26.

Every test we graded
TestResultVerdict
Total shot attempts fell along with shots on goal117.8 to 116.6 a gameDoes not hold
The miss rise is a few rinks29 of 30 rinks roseDoes not hold
Shot quality rosexG per unblocked attempt .066 to .073Holds, modestly

Frequently asked questions

Why are NHL shots on goal down?

Not because teams shoot less. Shot attempts were 117.8 a game in 2017-18 and 116.6 in 2025-26. More attempts are recorded as misses, mostly from close range, and the break in 2023-24 points at least partly to a change in how shots are logged.

How many shots on goal per game are there in the NHL?

About 55.7 a game for both teams in 2025-26, down from 63.9 in 2017-18. That is the fewest since 2001-02.

Why is NHL save percentage going down?

League save percentage (empty-net goals included) was .889 in 2025-26, the lowest since 1992-93. Part of the fall is shot quality, and part is that fewer attempts are recorded as shots on goal, which lowers save percentage even if goaltending is unchanged.

Did the NHL change how it records shots?

We found no NHL announcement of a change. The data shows a sharp break in 2023-24, recorded misses rose in 29 of 30 rinks, and analysts have linked it to puck-tracking data and to counting blocks by the shooter’s teammates from 2023-24.

More in The Vanishing Game

All How Hockey Changed research

Keep reading

Sources and method

  1. NHL stats (NHL.com), every season since 1917-18
  2. NHL stats API: team summaries and five-on-five shooting (SAT) totals
  3. Hockey Analysis: are changes in NHL shot tracking hiding better defending? (2025)
  4. Puck Over The Glass: something strange is going on with NHL shot data

Research and analysis, not betting advice. Past results do not guarantee future returns; bet only what you can afford to lose.