How Hockey ChangedThe Vanishing Game
The Last Goalie Who Played Every Night
A rule ended the every-night goalie in the mid-1960s, and the 1990s brought him back. Today teams split the net between two goalies, a tandem meant to keep them fresh. The tiredness it is supposed to prevent barely shows up in the data.
By Edgehalla ResearchPublished 8 min read
In early November 1962, Glenn Hall reached down to fasten his leg pads and felt a twinge in his lower back. On Nov. 7 he started against the Boston Bruins anyway, barely able to bend. He “missed a puck he normally would have stopped,” as NHL.com tells it, and gave the net to Denis DeJordy.
“I didn’t want to give up the net,” Hall said. He had started and finished 502 straight regular-season games without a mask: seven full seasons, plus the first 12 games of an eighth. He died in January, at 94.
In 2025-26, just one NHL goalie started 60 games. Back in 2001-02, 16 of the league’s 30 teams had a goalie who did.
On this page
A rule retired the iron man in the 1960s
The goalie who plays every night has disappeared twice since Hall: first by rule in the 1960s, then again over the last decade. Our data shows a heavy workload costs a goalie very little, so tiredness can’t explain the second exit.
For most of the NHL’s first 45 years, a team’s goalie played nearly every game. From 1950-51 the only backup was an emergency goalie: a spare in full gear, sitting in the building, who could go in for either team. In the six-team 1950s, three to five goalies a season played 65 or more of the 70 games, and two or three usually started all 70. Hall did it seven seasons running, from 1955-56 to 1961-62, two for Detroit and then five for Chicago.
The last man to start every game of his team’s season was Eddie Johnston, for a last-place Boston team in 1963-64. He won 18 times and faced 2,454 shots, the most in the league, without a mask. His season ended on March 22, 1964, with a 4-3 loss to Chicago at Boston Garden.
Workhorse goalies thinned out after 1965, came back in the 1990s and faded again
Share of NHL teams with a goalie who made 60 or more starts, every regular season since 1917-18. Shorter schedules are scaled up to 82 games
Show data
| Season | Teams with a 60-start goalie |
|---|---|
| 1917-18 | 75% |
| 1918-19 | 100% |
| 1919-20 | 75% |
| 1920-21 | 100% |
| 1921-22 | 100% |
| 1922-23 | 100% |
| 1923-24 | 100% |
| 1924-25 | 83% |
| 1925-26 | 100% |
| 1926-27 | 90% |
| 1927-28 | 90% |
| 1928-29 | 100% |
| 1929-30 | 90% |
| 1930-31 | 80% |
| 1931-32 | 88% |
| 1932-33 | 78% |
| 1933-34 | 89% |
| 1934-35 | 89% |
| 1935-36 | 88% |
| 1936-37 | 75% |
| 1937-38 | 100% |
| 1938-39 | 86% |
| 1939-40 | 86% |
| 1940-41 | 86% |
| 1941-42 | 100% |
| 1942-43 | 83% |
| 1943-44 | 50% |
| 1944-45 | 83% |
| 1945-46 | 67% |
| 1946-47 | 83% |
| 1947-48 | 100% |
| 1948-49 | 100% |
| 1949-50 | 100% |
| 1950-51 | 83% |
| 1951-52 | 100% |
| 1952-53 | 83% |
| 1953-54 | 100% |
| 1954-55 | 67% |
| 1955-56 | 100% |
| 1956-57 | 83% |
| 1957-58 | 67% |
| 1958-59 | 83% |
| 1959-60 | 67% |
| 1960-61 | 67% |
| 1961-62 | 67% |
| 1962-63 | 50% |
| 1963-64 | 83% |
| 1964-65 | 33% |
| 1965-66 | 33% |
| 1966-67 | 33% |
| 1967-68 | 17% |
| 1968-69 | 42% |
| 1969-70 | 42% |
| 1970-71 | 14% |
| 1971-72 | 14% |
| 1972-73 | 0% |
| 1973-74 | 25% |
| 1974-75 | 17% |
| 1975-76 | 28% |
| 1976-77 | 17% |
| 1977-78 | 22% |
| 1978-79 | 12% |
| 1979-80 | 10% |
| 1980-81 | 19% |
| 1981-82 | 24% |
| 1982-83 | 14% |
| 1983-84 | 5% |
| 1984-85 | 10% |
| 1985-86 | 5% |
| 1986-87 | 5% |
| 1987-88 | 14% |
| 1988-89 | 14% |
| 1989-90 | 10% |
| 1990-91 | 14% |
| 1991-92 | 32% |
| 1992-93 | 33% |
| 1993-94 | 35% |
| 1994-95 | 46% |
| 1995-96 | 27% |
| 1996-97 | 35% |
| 1997-98 | 50% |
| 1998-99 | 59% |
| 1999-00 | 43% |
| 2000-01 | 40% |
| 2001-02 | 53% |
| 2002-03 | 47% |
| 2003-04 | 40% |
| 2004-05 | – |
| 2005-06 | 27% |
| 2006-07 | 40% |
| 2007-08 | 43% |
| 2008-09 | 33% |
| 2009-10 | 43% |
| 2010-11 | 43% |
| 2011-12 | 43% |
| 2012-13 | 43% |
| 2013-14 | 27% |
| 2014-15 | 40% |
| 2015-16 | 33% |
| 2016-17 | 37% |
| 2017-18 | 29% |
| 2018-19 | 26% |
| 2019-20 | 13% |
| 2020-21 | 13% |
| 2021-22 | 16% |
| 2022-23 | 22% |
| 2023-24 | 9% |
| 2024-25 | 16% |
| 2025-26 | 3% |
- 1962-63: Hall’s streak ends
- 1965-66: Two goalies in uniform every game
- 1994-95: 48-game season
- 2012-13: 48-game season
Then the rules changed. In 1964-65 the NHL scrapped that emergency-goalie system and made playoff teams keep a second goalie in uniform. From 1965-66, every team had to have two goalies in uniform for every regular-season game. Five of six teams had a 60-start goalie in 1963-64, two a year later. In Toronto, Terry Sawchuk started 36 games and Johnny Bower 34, and they shared the Vezina Trophy, the goaltending award, the first season the league allowed it. Two seasons later the Leafs split the regular season three ways (Sawchuk 24 starts, Bower 24, Bruce Gamble 20) and won the Stanley Cup. The next season the league doubled to 12 teams, and only two of them had a 60-start goalie.
The 1980s already shared the net
The usual picture, of goalies who once played every night and are coddled now, skips two decades. Through the 1970s and 1980s, workhorses were as rare as they are now. No team had a 60-start goalie in 1972-73, and only one did in 1983-84, 1985-86 and 1986-87.
Coaches talked about it the way they do now. In 1982-83, Glenn “Chico” Resch started 30 of the New Jersey Devils’ first 34 games. “He has a right to be tired,” said Billy MacMillan, the Devils’ general manager and coach. “All I can do is ask him before the game if he’s tired, and if he says yes, I don’t use him.”
No. 1 goalies start three games in five again, as in the mid-1980s
Share of all starts made by the league’s busiest goalies (as many goalies as there are teams), every season; dashed line: each team’s own No. 1, from our game-by-game records since 2017-18
Show data
| Season | Busiest goalies, as many as there are teams | Each team’s own No. 1 (since 2017-18) |
|---|---|---|
| 1917-18 | 93% | – |
| 1918-19 | 96% | – |
| 1919-20 | 88% | – |
| 1920-21 | 96% | – |
| 1921-22 | 98% | – |
| 1922-23 | 100% | – |
| 1923-24 | 97% | – |
| 1924-25 | 95% | – |
| 1925-26 | 97% | – |
| 1926-27 | 94% | – |
| 1927-28 | 95% | – |
| 1928-29 | 97% | – |
| 1929-30 | 93% | – |
| 1930-31 | 90% | – |
| 1931-32 | 90% | – |
| 1932-33 | 91% | – |
| 1933-34 | 93% | – |
| 1934-35 | 92% | – |
| 1935-36 | 95% | – |
| 1936-37 | 86% | – |
| 1937-38 | 100% | – |
| 1938-39 | 90% | – |
| 1939-40 | 91% | – |
| 1940-41 | 88% | – |
| 1941-42 | 93% | – |
| 1942-43 | 91% | – |
| 1943-44 | 74% | – |
| 1944-45 | 86% | – |
| 1945-46 | 79% | – |
| 1946-47 | 92% | – |
| 1947-48 | 95% | – |
| 1948-49 | 98% | – |
| 1949-50 | 96% | – |
| 1950-51 | 91% | – |
| 1951-52 | 96% | – |
| 1952-53 | 93% | – |
| 1953-54 | 94% | – |
| 1954-55 | 81% | – |
| 1955-56 | 93% | – |
| 1956-57 | 89% | – |
| 1957-58 | 82% | – |
| 1958-59 | 88% | – |
| 1959-60 | 82% | – |
| 1960-61 | 75% | – |
| 1961-62 | 81% | – |
| 1962-63 | 78% | – |
| 1963-64 | 87% | – |
| 1964-65 | 67% | – |
| 1965-66 | 66% | – |
| 1966-67 | 62% | – |
| 1967-68 | 63% | – |
| 1968-69 | 68% | – |
| 1969-70 | 69% | – |
| 1970-71 | 59% | – |
| 1971-72 | 60% | – |
| 1972-73 | 59% | – |
| 1973-74 | 64% | – |
| 1974-75 | 63% | – |
| 1975-76 | 64% | – |
| 1976-77 | 61% | – |
| 1977-78 | 64% | – |
| 1978-79 | 58% | – |
| 1979-80 | 59% | – |
| 1980-81 | 58% | – |
| 1981-82 | 63% | – |
| 1982-83 | 60% | – |
| 1983-84 | 56% | – |
| 1984-85 | 60% | – |
| 1985-86 | 58% | – |
| 1986-87 | 59% | – |
| 1987-88 | 61% | – |
| 1988-89 | 59% | – |
| 1989-90 | 58% | – |
| 1990-91 | 61% | – |
| 1991-92 | 65% | – |
| 1992-93 | 64% | – |
| 1993-94 | 68% | – |
| 1994-95 | 73% | – |
| 1995-96 | 66% | – |
| 1996-97 | 68% | – |
| 1997-98 | 70% | – |
| 1998-99 | 68% | – |
| 1999-00 | 68% | – |
| 2000-01 | 68% | – |
| 2001-02 | 73% | – |
| 2002-03 | 69% | – |
| 2003-04 | 67% | – |
| 2004-05 | – | – |
| 2005-06 | 65% | – |
| 2006-07 | 69% | – |
| 2007-08 | 71% | – |
| 2008-09 | 66% | – |
| 2009-10 | 69% | – |
| 2010-11 | 69% | – |
| 2011-12 | 70% | – |
| 2012-13 | 70% | – |
| 2013-14 | 63% | – |
| 2014-15 | 68% | – |
| 2015-16 | 64% | – |
| 2016-17 | 67% | – |
| 2017-18 | 65% | 65% |
| 2018-19 | 62% | 62% |
| 2019-20 | 61% | 60% |
| 2020-21 | 59% | 58% |
| 2021-22 | 61% | 60% |
| 2022-23 | 59% | 58% |
| 2023-24 | 60% | 59% |
| 2024-25 | 63% | 61% |
| 2025-26 | 60% | 59% |
- 1965-66: Two goalies in uniform every game
Counting starts shows the same thing. The old season totals lump a traded goalie’s teams into one line, so each team’s starter is hard to pin down. Instead we took the league’s busiest goalies, as many as there were teams. In 1983-84 that meant the 21 goalies who started the most games. Together they made about 56% of the league’s starts, the lowest share in any season since 1917-18. Do the same for 2001-02, with 30 teams, and the busiest goalies made near 73%, the highest since teams had to carry two goalies in 1965-66. The No. 1 goalie of the early 2000s started about seven games in ten. Today’s No. 1 starts about three in five, as in the mid-1980s.
Then Brodeur’s era brought the workhorse back
The workhorse came back in the 1990s. From 1991-92 to about 2018-19, at least a quarter of teams had a 60-start goalie each season. At the 1998-99 peak, more than half did: 16 of 27. Grant Fuhr’s 1995-96 shows how far a coach would push it. Mike Keenan’s St. Louis Blues started him in their first 76 games. Keenan “appeared to have every intention of starting the veteran in all 82 games,” Sean McIndoe wrote in Grantland, “until a strained knee snapped the streak in early April.” Fuhr came back from the knee and played 79 games in all, 78 of them as the starter. Early in Game 2 of the first round, a collision at his net tore a ligament in his knee and ended his season. That collision tells us nothing about whether 76 straight starts had worn him down.
Nobody carried the era like Martin Brodeur. He made 70 or more starts in eleven seasons, the most of any goalie since expansion in 1967. In 2006-07 he started 78 games and won 48, one more than Bernie Parent’s 47 in 1973-74, though ten of Brodeur’s came in shootouts, which did not exist in Parent’s day. Brodeur thought the load suited the right kind of goalie: “Goalies who are athletic rather than robotic can handle a 70-game or a similarly large workload,” he said.
Nobody since 1967 matches Brodeur’s 11 seasons of 70 starts
Goalies with three or more 70-start seasons since expansion (1967-68 to 2025-26)
Brodeur had 11 seasons of 70+ starts, Miikka Kiprusoff 6 and Roberto Luongo 4.
Then the 70-start seasons stopped, with little notice. Cam Talbot started 73 games for Edmonton in 2016-17, and nobody has started 70 since.
The 65-game season went the same way. Eight goalies played 65 or more games in 2000-01, and eight did again in 2010-11. In 2015-16 it was six. The last time anyone got there was 2021-22, when two goalies did. In 2025-26 nobody played more than 64.
Utah rode an average goalie to 63 starts
Today the workhorse is nearly gone again, with one exception. At the Olympic break in February, Karel Vejmelka led the league with 44 starts. “He’s played a lot of hockey going into the Olympics,” said his coach, André Tourigny. He finished with 63; the next busiest, Nashville’s Juuse Saros, had 59. His save percentage was .897. The league’s was .896. Utah has leaned on him before. From Feb. 23 to April 12, 2025, Vejmelka started 23 games in a row, the longest run in our game-by-game records, which start in 2017-18. Next is Darcy Kuemper’s 22 for Arizona in 2019.
Everyone else split the net, including Vegas, the team that would end Utah’s season. Vegas’s busiest goalie, Akira Schmid, started 29 games, and Adin Hill 27. Six teams had no goalie with half their starts. The league used 98 goalies. In 2000-01, with 30 teams to today’s 32, it used 89.
Then his spring went wrong. From October through January, Vejmelka let in 108 goals. Hand an average goalie the same shots, from the same spots, the same tip-ins and slap shots and backhands, and he would have let in about 120. That figure is his expected goals against. Vejmelka beat it by about 12, so he was about 12 goals saved above expected. In March and April Vejmelka let in 52, over 17 starts. The average goalie would have let in about 45. Utah led Vegas 2-1 in the first round, lost Games 4 and 5 in overtime and went out in six. Vejmelka started all six. It looks like a goalie who wore out. But it is one goalie over 17 starts, and the rest of the league tells a different story.
The obvious fatigue test fools itself
So did Vejmelka wear out? To find out, we looked at about 19,000 starts from 2017-18 to 2025-26, by every goalie who made at least 20 starts in a season. For each start we asked how busy the goalie had been going in, and how well he played once we allowed for how dangerous the shots were. Our shot-by-shot records start in 2017-18, so this test can’t be run on Hall’s seasons or Brodeur’s.
Start with the obvious test. Line up each start against the goalie’s own season average. By that test, a heavy run of five or six starts in two weeks costs about 1.7 goals per 1,000 unblocked shots. The count takes in every shot that gets past the skaters in front of him, including the ones that miss the net. A goalie faces about 41 a night, so 1,000 of them is about 24 starts’ worth.
Here is why that number can’t be trusted. The same test says a goalie who has been hot for five starts plays about 3 goals per 1,000 worse next time, and a cold one about 4 better. That comes from the arithmetic. Vejmelka had a hot run in January. From Jan. 5 to 15, he stopped 125 of 132 shots in five starts, a .947 save percentage. Those five starts sit inside his season average and pull it up. Over his next five, Jan. 17 to 29, he stopped 125 of 141 (.887), back below the .897 he finished at. Yet an average goalie would have let in about 19 goals on those shots, and he let in 16.
Goalies believe in the roll, Brodeur among them. “The mental game is so huge and it can be draining,” he said. “But when you’re playing every night, you can overcome a bad game by going on a roll.” Coaches give the net to a goalie who is rolling, so heavy runs tend to go to goalies who have been playing well. That is the same trap, and it makes heavy runs look costlier than they are.
Next to his season average a hot goalie seems to cool off. Next to his later starts, he doesn’t
Goals saved above expected per 1,000 unblocked shots in the next start, by how the goalie played over his previous five starts, 2017-18 to 2025-26. Below 0 = worse
Against the season average, hot goalies seem to fade (−3.2) and cold ones bounce back (+3.9); against the goalie’s later starts both are near zero.
- Vs his season average
- Vs his later starts
In each pair, the first bar is the obvious test and the second is the fairer one we switched to. Against the season average, cold goalies bounce back and hot ones fade. Against their own later starts, both shrink to under half a goal per 1,000, too small to tell from chance.
A heavy run costs about one goal every 22 starts
So we compared each start with the same goalie’s later starts that season. Those later starts don’t include the streak that came before, so the streak can’t tilt the comparison. A goalie who had made five or six starts in the past two weeks still did a little worse: about 1.1 extra goals per 1,000 unblocked shots, down from the 1.7 the obvious test found. At about 41 of those shots a start, that is one extra goal every 22 starts, during those busy stretches. Resampling the goalies puts the likely size of the dip between 0.3 and 2.0 extra goals per 1,000, over 4,519 busy starts, so even the low end is above zero. The mirror image shows up too. Goalies with two or fewer starts in the past two weeks did a little better than in their later starts, by about 0.8 goals per 1,000 (likely range 0.2 to 1.4).
We split the seasons in two to check it. The dip shows up in both the earlier seasons (2017-21) and the later ones (2021-26). In the earlier seasons alone, though, the dip is too small to rule out luck.
Busy goalies did a little worse than in their own later starts, in both halves of the data
Goals saved above expected per 1,000 unblocked shots (shots not blocked by a skater), compared with the same goalie’s later starts that season. 0 = an ordinary start; below 0 = worse. Earlier seasons (2017-21) and later seasons (2021-26) side by side
Five or six starts in the previous 14 days was −0.8 in 2017-21 and −1.4 in 2021-26; the 7th to 10th straight start was −1.5 and then −5.4.
- 2017-21
- 2021-26
Each pair of bars is one kind of start, earlier seasons first. The goalies with the lightest workloads sit above zero in both halves, and so do goalies starting right after another goalie started the team’s last game. The busiest sit below zero in both. The deepest bar, the 7th to 10th straight start in the later seasons, rests on only 237 starts.
And the busy goalies weren’t facing easier shots, or harder ones. The danger of the shots barely changed from one group to the next, by less than half a goal per 1,000.
The longest runs look worst, and there are too few of them to call. Goalies in their 7th to 10th straight start gave up about 3 extra goals per 1,000, over just 536 starts. In the earlier seasons the dip was about 1.5 and could easily be luck. In the later seasons it was about 5, on only 237 starts.
Rest and the schedule show no sign of wear
Nothing else that should show tiredness held up. Raw save percentage barely moved with days of rest: .907 with one day off, .906 with two, .906 with three to five and .907 after six or more. Against his own later starts, no rest group was clearly worse than an ordinary start.
No amount of rest made a clear difference
Goals saved above expected per 1,000 unblocked shots vs the same goalie’s later starts that season (0 = an ordinary start), 2017-18 to 2025-26
No rest group was clearly below an average start; playing the day before was +2.1, on 514 starts.
Bars above zero mean better than the goalie’s own later starts. Goalies who had played the night before came out a little ahead, by about 2 goals per 1,000, but on only 514 starts that could be luck.
Our goalie study found no clear drop either in starters playing the second of two nights in a row, and our back-to-back study traced tired teams’ second-night losses to their skaters, not their goalie. Coaches pick those nights, so they may be steering tired goalies away from them. If goalies feel fatigue on second nights, it hasn’t shown up in the goals against.
The team’s schedule is a cleaner test, because the league sets it and the coach cannot change it. If fatigue matters, a packed schedule should hurt. Instead, goalies did slightly better than usual after their team had played seven or eight games in 14 days, by about 0.9 goals per 1,000. The bump is small. Its likely range stays above zero, but in the next check it is one of the results luck alone could produce. Either way, tired goalies should do worse here, so it points the wrong way for fatigue.
One more check. We cut the starts 24 ways: by rest, by recent workload, by straight starts, by the team’s schedule, by recent form and by point in the season. Even if fatigue meant nothing, luck alone would usually make two or three of the 24 stand out. Five stood out. Two are busy goalies doing a bit worse: five or six starts in two weeks, and the 7th to 10th straight start. Two are rested goalies doing a bit better: two or fewer starts in two weeks, and a start right after another goalie started the team’s last game. Those four are really two findings seen from both ends. Every group is measured against the average start, so if the busiest goalies sit below it, other starts have to sit above it, and the rested goalies are where that shows. The fifth is the packed-schedule bump, which points the wrong way. All five pointed the same way in both the earlier and later seasons, though four of them were too small in one half to rule out luck. Together they hint at a small cost of fatigue.
The season-average test also says goalies fade late in the season, by about 2 goals per 1,000 in their 41st to 55th starts. The fairer test puts that fade at about 0.1. The likely reason is who is still playing. Coaches ride a goalie who starts the season well, so a goalie who reaches his 41st start often has a strong first half lifting his season average. His spring starts then look worse next to it, tired or not. That leaves Vejmelka’s bad March and April, 17 starts by one goalie, proving little on their own.
Tiredness can’t explain the tandem
Utah rode Vejmelka for 63 starts. Across the league, a stretch like his busiest ones costs about one extra goal every 22 starts. Tiredness may be real, but that is far too small to explain how much teams now split the net.
Other reasons may matter more: injury risk, playoff freshness and backups who are closer to starters in quality. None of them can be measured from save percentage, and we found no coach on the record explaining why teams split the net now.
The last workhorse is a backup now
The last man to start 70 games in a season spent 2025-26 as a backup. Cam Talbot was in Detroit. John Gibson started 57 games. Talbot started 25.
Hall never made much of his record. “It’s just a number,” he told The Hockey News. “It doesn’t mean that much. Just something that happened.”
Share this article
Goalie tools
Live Edgehalla tools on the same subject.
Show the work
- Tests written down
- Not in advance (exploratory)
- Results computed
Goalie starts for every regular season from 1917-18, from the NHL’s goalie totals (starts sum to two a game in every season but 1917-18), scaled to 82 games: a goalie counts as a workhorse at 60 starts on that scale, so 43 starts in the 48-game 2012-13 count as about 73. Named goalie-seasons, 70-start counts and the every-game list come from the same totals; the run of straight starts and Vejmelka’s months come from our game logs. Workload test: 19,446 starts by goalies with 20+ starts in a season (502 goalie-seasons, 124 goalies), 2017-18 to 2025-26, out of 22,102, with goals and expected goals per start from our play-by-play. For each start we recorded days of rest, starts in the previous 14 days, the run of straight team starts, the season start number and the team’s games in the previous 14 days. Expected goals are recalibrated per season and month. They account for where shots come from and what kind they are, not who takes them. Each start is compared with the same goalie’s later starts that season. 90% ranges come from resampling goalies, and 2017-21 vs 2021-26 is the split-half check. Not pre-registered: the first baseline (season average) was replaced after it failed the hot/cold placebo.
| Test | Result | Verdict |
|---|---|---|
| No goalie played 65 games in 2025-26 | Most games 64; one goalie with 60+ starts | Holds |
More tests below
| Test | Result | Verdict |
|---|---|---|
| Workhorse goalies are at a historic low | 1 of 32 teams in 2025-26; also 1 team in 1983-84, 1985-86, 1986-87; none in 1972-73 | Reframed: back to 1980s levels |
| Season-average baseline is fair (placebo: form over the last 5 starts) | Goals saved above expected per 1,000 unblocked shots in the next start. Vs season average: cold +3.9, middling −0.4, hot −3.2 (a goalie hot over his last five starts plays about 3 goals per 1,000 unblocked shots worse next time, a cold one about 4 better). Vs later starts (relative to the average start, 0): cold +0.4, middling +0.1, hot −0.5. Hot = 20+ goals saved above expected per 1,000 unblocked shots over the previous five starts, cold = 20+ below; starts: cold 2,210, middling 12,589, hot 2,137. Why it fails: a hot run counts in the season average it is measured against, pulling that average up, so whatever follows looks worse; coaches ride the roll, so heavy runs go to goalies playing well and get the same unfair comparison | Fails; replaced |
Continues below
| Test | Result | Verdict |
|---|---|---|
| Naive test: a heavy run (5-6 starts in 14 days) costs save % vs the season average | −1.7 goals saved above expected per 1,000 unblocked shots (about 1.7 goals per 1,000 lost), across 19,446 regular-season starts, 2017-18 to 2025-26 | Biased by the baseline (see placebo) |
| Save % fades on a heavy run (5-6 starts in 14 days) | −1.1 (90% range −2.0 to −0.3), n 4,519; about 1.1 extra goals per 1,000 unblocked shots × about 41 unblocked shots a start = one extra goal every 22 starts; in the 2017-21 half alone the range includes zero; halves −0.8 (90% range −2.1 to +0.4) / −1.4 (90% range −2.4 to −0.3); light load (0-2) +0.8 (90% range +0.2 to +1.4) | Small, same sign in both halves |
| Save % fades deep into a start streak (7th-10th) | −3.2 (90% range −5.8 to −0.7) (n 536); halves −1.5 / −5.4 | Too thin to call |
Continues below
| Test | Result | Verdict |
|---|---|---|
| Days of rest change save % | No rest group below an average start; raw save % .907 (1 day off), .906 (2), .906 (3-5), .907 (6+); played the day before +2.1 (90% range −0.3 to +4.9), n 514 | Not detected |
| A packed team schedule hurts the goalie | 7-8 team games in 14 days: +0.9 (90% range +0.2 to +1.6) | Not detected |
| Goalies fade late in the season (starts 41-55) | Naive −1.9; fair −0.1 | Selection, not fatigue |
| Shot quality differs on heavy runs | Expected save rate moved less than 0.5 goals per 1,000 across the load and streak groups shown | No |
| Vejmelka wore down in 2025-26 | Oct-Jan 108 goals on 120.3 raw expected (42 starts); Mar-Apr 52 on 44.5 (17 starts) | One goalie: a scene, not tested |
Frequently asked questions
Who played the most games in a season as an NHL goalie?
Grant Fuhr played 79 games for St. Louis in 1995-96, with 78 starts. Martin Brodeur also started 78 for New Jersey in 2006-07.
How many NHL goalies played 65 games in 2025-26?
None. The most was 64 games, and only Karel Vejmelka reached 60 starts (63). Eight goalies played 65 or more in 2000-01 and eight in 2010-11; the last season anyone did was 2021-22.
Which goalie has the most 70-start seasons?
Since expansion in 1967-68, Martin Brodeur, with 11. Miikka Kiprusoff had 6 and Roberto Luongo 4. The last 70-start season was Cam Talbot’s 73 for Edmonton in 2016-17.
What was Glenn Hall’s consecutive games streak?
Hall started and finished 502 straight regular-season games, from the start of 1955-56 with Detroit to 12 games into 1962-63 with Chicago, without a mask. It ended on Nov. 7, 1962, against Boston, when a back injury forced him out. The last goalie to start every one of his team’s games in a season was Eddie Johnston in 1963-64.
When did NHL teams start using two goalies?
From 1965-66 teams had to have two goalies in uniform for every regular-season game; the playoffs required a backup in uniform from 1964-65. The share of teams with a 60-start goalie fell from 83% in 1963-64 to 33% in 1964-65 and 17% in 1967-68.
Does goalie fatigue lower save percentage?
Only a little, if at all. From 2017-18 to 2025-26, after allowing for how dangerous the shots were and comparing each start with the same goalie’s later starts, a goalie who had made 5 or 6 starts in the previous 14 days gave up about 1.1 extra goals per 1,000 unblocked shots. At about 41 unblocked shots a start, that is roughly one extra goal every 22 starts. Days of rest and the team’s schedule showed no sign of fatigue.
More in The Vanishing Game
- Verdict: Data story
Did Garbage Goals Take Over the NHL?
- Verdict: Data story
For 19 Seasons, an NHL Hit Depended on the Rink
- Verdict: Data story
Hainsworth’s 22 Was Mostly Arithmetic
- Verdict: Data story
The Slapshot Went From One NHL Goal in Four to One in Twelve
- Verdict: Data story
Is the Rush Dying in the NHL?
- Verdict: Data story
NHL Fights Fell to 0.18 a Game in 2018-19, Then Climbed Past the 0.20 of Gordie Howe’s Era
Sources and method
- NHL stats (NHL.com), every season since 1917-18
- NHL stats API: goalie season totals (games played and started), every season since 1917-18
- Scouting the Refs: NHL historical rule changes (two goalies in uniform from 1965-66)
- NHL.com, Bob Verdi: Glenn Hall, 100 Greatest NHL Players (the streak and its end)
- The Hockey News, Brian Costello: Best of the Books, consecutive games played by a goalie
- The New York Times obituary, via The Boston Globe: Glenn Hall dies at 94 (Jan. 8, 2026)
- NHL.com: This date in NHL history, March 22 (Eddie Johnston’s last iron-man season)
- The Hockey News: Top 100 goalies, No. 87 Eddie Johnston
- The Canadian Encyclopedia: Vezina Trophy (shared from 1964-65)
- Wikipedia: 1967 Stanley Cup Final
- The Hockey News, Glenn Dreyfuss: NHL history murky about when goalie use becomes overuse (MacMillan, Brodeur)
- Grantland, Sean McIndoe: 10 facts about the 1995-96 St. Louis Blues
- The Hockey Writers, Chase Beardsley: Vejmelka embracing Olympic opportunity (Feb. 14, 2026)
- NHL API reference (community documentation of the stats endpoints)
Research and analysis, not betting advice. Past results do not guarantee future returns; bet only what you can afford to lose.