Vegas Had the Best First Season of Any NHL Expansion Team

Every new team starts near the bottom

The trash can is the rule and Vegas the exception. The NHL has stocked a new franchise through an expansion draft 23 times since 1967, from the six teams of 1967 to Seattle in 2021. That is not many. Everything below rests on 23 first seasons in eight classes, and only one Vegas. Four more teams came over from the World Hockey Association (WHA) in 1979 with players of their own. They appear in the charts but stay off the curve.

Comparing 1967 with 2021 needs one scoring rule. A regulation win counts 1 and a regulation loss 0. Take a game tied after 60 minutes: in 1975 it stayed a tie, and in 2018 somebody won it in overtime or a shootout. Both count as half for each side. A team’s season average on that rule is its regulation share, and an average team scores .500 in every era. The average new team scored .371 in year 1. Year 2 was barely better, at .377. Then the climb starts: .431 in year 3, .445 by year 5, about .496 in year 7.

Winning seasons came late. Only two new teams finished above .500 in year 1, Vegas and the 1993-94 Florida Panthers. By year 3 it was about 30% of them, by year 6 about 46%. The typical team needed three seasons to reach its first playoffs and five or six to post its first winning season. One never had a winning season at all: the Oakland Seals, who became the California Golden Seals and then the Cleveland Barons, and stopped after 11 seasons.

The floor is 8-67-5

The worst season any NHL team has played belongs to a newcomer. The 1974-75 Capitals went 8-67-5. They took 13% of the points on offer (a points percentage of .131), the lowest of 1,719 team-seasons of 40 games or more. They scored 181 and allowed 446, 3.3 goals a game worse than their opponents. Their road record was 1-39-0.

They were not unlucky. A team that gets outscored that badly should win almost nothing: judged by goal difference alone, Washington would have finished with a regulation share of about .015. They managed .131. They were simply that outmatched. Nor was it only the draft: on our roster measure, explained below, Washington’s players were no worse on paper than other 1970s newcomers, and the team fell well short of what that roster predicted.

From Feb. 18 to March 26 they lost 17 straight. Then came Oakland. The Hockey Writers places the celebration on the ice: “there was a metal trash can on the visitors’ bench,” and “one player – possibly Tommy Williams – convinced his teammates to sign their names on the can with a marker.” The same piece warns: “The accounts of the event vary depending on the source.” They do. Russian Machine Never Breaks moves the scene indoors: “Inside the Capitals locker room, Tommy Williams picked up a trash can like it was the Stanley Cup and paraded around the room with it.”

Seventeen years later Ottawa came close. The 1992-93 Senators went 10-70-4 and took 14% of the points (.143), the third-lowest ever. In San Jose’s second season the Sharks went 11-71-2, a tie for that spot. Ottawa’s players picked up the nickname “Road Kill,” Roy MacGregor recalled in The Globe and Mail. Their only road win came on April 10, 1993, on Long Island, when they “had already set a new league record for consecutive road losses, 38,” and captain Laurie Boschman scored “a hat trick” in “a surprising 5-3 victory.”

Our game list counts that streak as 39. The difference is a game in Hamilton. On Oct. 20 the Senators lost to Toronto there, and our list files it as a road game. Ottawa’s management had pressed the league to leave its “two losses on neutral sites,” in Hamilton and Saskatoon, out of the road tally, and the league agreed. MacGregor calls it the “Asterisk Affair.”

Being the worst did not mean staying there. The 1972-73 New York Islanders went 12-60-6, then rose by .167 in their second season, more than any new team but one, and won the Cup in their eighth season. They won four in a row, from 1979-80 to 1982-83. The Capitals reached the playoffs in their ninth season and won their first Cup in 2017-18.

In 1967 the Original Six had to give up 20 players each

Some first-year records look far better than that. The first class shows how they can fool you, and it was also the strangest. On June 6, 1967, the Original Six stocked six new teams at once. As Sean McIndoe put it at Down Goes Brown, “each of the Original Six teams was going to have to cough up a total of 20 players.” The old clubs could protect 11 skaters and one goalie, and, as Sal Barry recounts in The Hockey News, “After losing a player, they could add another to their protected list.” Toronto’s Punch Imlach tried to hide Terry Sawchuk by listing his goalie among the skaters. “Well, he can skate, can’t he?” Imlach reasoned, in Barry’s telling, “unsuccessfully, to the other GMs in the room.” The Los Angeles Kings took Sawchuk first overall.

The 1967 Six were handed a place in the Final, and still lost to the old teams

The draft was only half the gift. Then the league put all six new teams in one division. “They were housed in their own division, called the West Division at the time,” The Hockey Writers notes, so the six spent much of their first season playing each other, and one of them had to reach the Cup Final. Five of them finished between .453 and .493 in regulation share, records that flatter them. Against the Original Six they took only .340 over 144 games.

The St. Louis Blues, with Glenn Hall in goal, came out of the West and reached the Final in each of their first three seasons. They lost all twelve games. “The Blues would reach the Final three years in a row,” The Hockey Writers wrote, and only after the 1970 expansion were the teams realigned so that “no expansion team was guaranteed a spot in the Cup Final from that point forward.” No new team reached the Final in its first season again until Vegas, in 2017-18.

By 2017 the drafts were handing over real NHL players

The usual explanation for Vegas is the draft. By 2017 the old teams were giving up far more than in the past. When the NHL last expanded, in 2000, “teams were allowed to protect a whopping nine forwards, five defencemen and a goalie, or seven forwards, three defencemen and two goalies,” Greg Beacham of The Associated Press wrote in 2018. For Vegas, “The NHL allowed its teams to protect only seven forwards, three defencemen and one goalie, or eight skaters and one goalie.” “The league also required teams to expose players with significant NHL experience who were under contract through next season,” Beacham added. “Third-line forwards and top-four defencemen were available from almost every team.”

“The (expansion draft) rules were favourable,” said George McPhee, the Golden Knights’ general manager. “Gave us something to work with, and gave this team an opportunity to be a good team.” The 1967 rules were not much stingier by count, McIndoe argued: they “were actually slightly more generous to the existing teams than the 2017 draft.” What changed was the pool. In 1967 each of six old teams had to give up 20 players. In 2017 there were 30 other teams to pick from, for one new one.

We can measure what the drafts handed over. Take every skater who played for a new team in its first season and ask whether he was in the NHL the season before. In the 1967-74 classes, players who had been in the NHL the season before got about 53% of the skater games. In 1991-2000 it was 82%; for Vegas and Seattle, 97%. A second measure, our roster index, asks how fast the year-1 players had scored in the NHL the season before. A roster that scored like an average NHL roster rates 1.00, and a roster of men who had all been outside the NHL rates zero. The index rose from 0.28 to 0.54 to 0.79. Both numbers are paper. They say how good the players looked before the first puck dropped.

Better rosters did not bring better first seasons

If the rules made new teams good, year 1 should have climbed with the rosters. It did not. Across the 23 debuts, a first season’s regulation share moved +.003 a decade. Over 54 years that adds up to about one regulation win a season. It is nothing. Try it this way: pin each team’s first season to a random year, so the 1974-75 Capitals might land in 2017 and Vegas in 1970, and redraw the trend. Do that 20,000 times and a rise at least that steep turns up 44% of the time. Take Vegas out and the line tilts down, −.012 a decade.

The 1990s teams show it most plainly. Those nine had far better paper rosters than the 1967-74 classes, yet they finished lower in the league. The average 1990s newcomer finished ahead of only about 15% of the league in year 1. The average 1967-74 newcomer finished ahead of about 22%. Against established opponents the 1990s teams did a little better, .347 against .293. Across all 23 debuts that measure rose +.021 a decade. Shuffle the years and 7 shuffles in 100 produce a rise as steep. That is too often to rule out chance. Without Vegas the rise shrinks to +.009 a decade.

Paper talent did help a little, on a loose leash. Florida in 1993-94 had one of the best paper rosters of any new team (an index of 0.61) and the second-best debut. San Jose in 1991-92 had nearly the same paper (0.58) and finished at .256. Across the 17 debuts from 1970 on, roster and result line up with a correlation of 0.48, where 0 would mean no link and 1 would mean results rose in perfect step with the roster. Without Vegas it is 0.39.

To put a price on a paper roster, we took 1,522 seasons of established teams since 1943-44 and checked how their roster index lined up with where they finished. That fit is our talent model. From paper rosters alone, it expected first seasons of about .253, .352 and .452 in the three eras. The 1990s teams landed almost exactly on theirs. Their rosters were better, the model expected more, and the standings barely moved.

The 1990s teams climbed slowest

The 1990s teams, with the better rosters, were also the slowest to climb. Making the playoffs is no fair test across eras, because the bar moved: four of the six West teams got in in 1968, 16 of 21 teams from 1979-80, 16 of 32 now. So we compare each team with an average team in the same seasons. In their first five years, the 1967-74 newcomers made the playoffs 28 times, where average teams would have made it about 37. The nine teams of 1991-2000 made it eight times, against about 27.

The typical 1990s team needed four seasons to reach the playoffs and six to post a winning season, against two or three and five for the earlier classes. Atlanta’s Thrashers needed seven seasons to reach the playoffs; Columbus needed eight, and 18 to win a round. The 1990s drafts also came with their own games. Barry recalls that in 1991 “the Canadiens listed Jean Beliveau, Yvan Cournoyer and Ken Dryden for the draft,” all long retired, and that before 1992 the Capitals “signed long-retired goaltender Bernie Wolfe solely for the purpose of making him available.” Wolfe’s worry, in Barry’s telling: “I said, ‘Jeez, I’m 40 years old. What if someone drafts me?’”

Vegas broke the pattern, mostly on merit

Into that history walked Vegas. The Golden Knights won their first game 2-1 at Dallas on Oct. 6, 2017, and won eight of their first ten. They finished 51-24-7, fifth of 31 teams, and won three playoff series before losing the Final 4-1 to Washington. Their regulation share, .592, is the best first season of all 27 new teams since 1967, ahead of Florida’s .512 in 1993-94.

The players called themselves the Golden Misfits, but, as Beacham wrote, “few of Vegas’ expansion draft selections were truly undesired by the clubs that lost them 11 months ago.” Some of them simply had the seasons of their lives. William Karlsson scored 43 goals after six for Columbus the year before. Erik Haula scored 29, after 15. Jonathan Marchessault went from 51 points to 75.

Of all 27 first seasons since 1967, only Vegas and Florida won more than they lost in regulation

Line every debut up, best to worst, and Vegas stands alone at the top. It finished first at .592, Florida second at .512, and nobody else reached .500. The 1974-75 Capitals are last at .131.

A debut like Vegas’s should come along about once in 53 tries

Luck explains only part of it. Start with how far Vegas stood from the pack. The other 22 debuts averaged .360, and they usually landed about .111 either side of that. The nine 1990s teams show what that spread looks like. Ottawa, the worst of them at .179, sat 1.6 of those steps below the average. Florida, the best at .512, sat 1.4 steps above. Vegas sat 2.1 steps above, or 2.5 if the 1967 Six, with their friendly schedule, are left out. Those steps are standard deviations.

If debuts spread out in the usual bell shape, a season 2.1 steps above the average comes along about once in 53 tries. The league has had 23 tries, which gave it about a 35% chance of seeing one. Rare, then, but chance could have produced it. Then allow for bounces. Over a full season, luck alone moves a team’s share by about .051 either way. Allow for that and Vegas’s first year still rates around .551, the best debut by a wide margin.

Only 18 teams since 1970-71 beat their paper rosters by more than Vegas

Against its paper roster Vegas was more remarkable. The talent model expected about .415, and it usually misses a team by about .084 either way. Vegas beat it by 2.1 times that, the 19th-biggest overachievement of 1,396 team-seasons since 1970-71. Only 18 teams beat their paper by more, led by the 1984-85 Philadelphia Flyers, the 1979-80 Philadelphia Flyers and the 1976-77 Montréal Canadiens. Some of it was shooting. Vegas scored on 10.0% of its shots, against 9.2% for the league, which was worth about 22 extra goals, and its goalies saved about 6 more than an average pair would have. That left little room for luck in close games. A team that outscored its opponents by 43 goals would be expected to reach about .577, near the .592 Vegas got.

And it lasted. Over years 2 to 5, Vegas averaged .557, behind only the Edmonton Oilers, who came over from the WHA (.634), among all 27 newcomers. In nine seasons Vegas has never dipped below .518 and has missed the playoffs once.

Seattle got the same rules and landed on the old curve

“It also seems improbable that the NHL would ever make it this easy to build a team again,” Beacham wrote in May 2018. Three years later the league did. Seattle’s 2021 draft was held “under the same rules for the Kraken as the Vegas Golden Knights in 2017,” NHL.com announced, with teams again allowed to protect “seven forwards, three defensemen and one goalie, or eight skaters (forwards/defensemen) and one goalie.”

The other general managers had learned something. Ron Francis, Seattle’s GM, “has been dealing with colleagues who are a little bit gun shy after the way Vegas Golden Knights counterpart George McPhee made off like a bandit in the 2017 expansion draft,” Adam Proteau wrote in The Hockey News the day after. Still, on paper Seattle got more than any newcomer ever had: 98% of its skater games went to men from last season’s NHL, and its roster ranked first of all 23 on our roster index. Vegas ranked second.

The Kraken lost their first game 4-3 at Vegas and went 27-49-6, 30th of 32. Their regulation share, .342, was 15th of the 23 debuts. Measured in the same kind of steps, Seattle sat −0.2 from the average of the other debuts, Vegas aside. That is almost dead on it. Its roster promised more. The talent model expected .489, and Seattle fell short by 1.8 times the model’s usual miss. Only 52 of the 1,396 team-seasons since 1970-71 fell further below their paper. Vegas and Seattle finished .250 apart. One season of luck usually puts about .069 between two teams, and this gap was 3.6 times that. Same rules, opposite results, and not mostly by chance.

Year 2 went the other way. Seattle jumped to .555, the biggest second-season rise of any new team (+.213), made the playoffs and won a round. It has not been back above .500 since.

Most new teams improve, but a fast start is no shortcut

Vegas stayed good. A strong first year usually promises less. The 1993-94 Panthers had the second-best debut ever, reached the Final in their third season, and were back at .378 by year 5.

Vegas and Seattle are one season each, and Florida is another. Florida opened at .512 and averaged .504 over the next four. The Islanders opened at .192, one of the worst debuts ever, and averaged .551. Across all 23 draft-built teams a better first season did go with a better next four, at a correlation of 0.49 (0.40 without Vegas). That is a loose link, and the Islanders broke it badly.

What year 1 does say is that teams improve. The average new team went from .371 in year 1 to .425 over years 2 to 5, a gain of .054, and 15 of the 23 got better.

As a forecast, year 1 barely beats a flat guess

A fan in a team’s first spring wants to know what the standings say about the next four years. We tried year 1 as a forecast. Hide one team, say the 1993-94 Panthers, and predict its years 2 to 5 from the other 22. Do it twice: once using its first season, and once with a flat guess that every new team lands at the same level. Then repeat for every team.

Year 1 barely beat the flat guess. The forecasts that used it missed a team’s years 2 to 5 average by about .076. The flat guess missed by .082. Over 82 games that difference is worth less than half a regulation win. With Vegas left out it shrinks further: .077 against .078.

Splitting by era gives no cleaner answer. A rule built from the 1967-74 teams helped forecast the later ones, WHA arrivals included: it missed by .080, against .089 for the flat guess. Run it the other way and it failed. A rule built from the later teams made the 1967-74 forecasts worse, .092 against .084.

Established teams are easier to read. Across 1,169 of their seasons since 1967-68, one season lines up with the next four at a correlation of 0.57, against 0.49 for first seasons. A first year is the noisier one.

Plan on six years, and treat a fast start as a bonus

• A new team should expect to be well below average for two seasons and to reach average around its sixth or seventh. That has held from 1967 to 2021.
• Generous draft rules give a new team a better roster on paper. On this evidence they do not lift it in the year-1 standings. Seattle, one team in one season, fits that.
• A strong first season says less about the next four than an ordinary season says about an established team’s. Vegas’s staying power is unusual in its own right.

Expansion also changes the teams that give up players. Our NHL GOAT series found that continuing players’ era-adjusted scoring jumped when the league doubled in 1967. It did not move when the WHA teams joined in 1979, because those players were already NHL calibre. That is one more reason the WHA four are kept apart here.

The trash can, and the Cup

Vegas’s owner, Bill Foley, had set what Beacham called “a public goal of bringing the Stanley Cup to Las Vegas within six years.” Six years is about how long the typical new team has needed just to become average. The Golden Knights won the Cup in 2022-23, their sixth season.

Washington took longer. Its Cup came in its 43rd season; count the cancelled 2004-05 season and it is the 44th year. “On June 7, 2018, after 44 years of trying, the Washington Capitals finally lifted the Stanley Cup,” The Hockey Writers begins its history of that 1974-75 team.

The team the Capitals beat for it was the first-year Golden Knights.

Frequently asked questions

More in Standings and results

• Verdict: Data story

  Does Scoring Depth Win Stanley Cups?

• Verdict: Data story

  Same Goals, Half the Blowouts

• Verdict: Data story

  Are the NHL’s Stars Pulling Away From Everyone Else?

• Verdict: Data story

  The Schedule Got Kinder. The Second Night Didn’t.

• Verdict: Data story

  Regular-Season Records Predict NHL Playoff Upsets Almost Exactly

• Verdict: Data story

  Home Ice Lost Almost Half Its Edge in 1992. The Power Plays Didn’t Budge.

All How Hockey Changed research →

Keep reading

• How Hockey Changed: NHL data stories →One chart-led story a day about how the NHL game, its players and its rules have changed.
• Does the regular season still matter? →How often the better team wins a playoff series, and why the gap between teams shrank.
• Home ice lost half its edge →The same regulation-share measure, applied to home teams since 1917.
• Who is the NHL GOAT? →Era-adjusted scoring since 1917, and why goalie talent compressed.
• NHL player pages →Every current player: stats, lines, ice time and form.

Sources and method

1. NHL stats (NHL.com), every season since 1917-18
2. NHL stats API: team season totals and every regular-season and playoff game since 1917-18
3. NHL stats API: skater season totals (for the roster measure)
4. NHL.com: Kraken 2021 expansion draft rules same as Golden Knights followed (July 20, 2021)
5. The Associated Press, Greg Beacham, via CBC Sports: Golden Knights were built for success thanks to favourable expansion terms (May 17, 2018)
6. The Hockey News, Adam Proteau: Seattle’s path to NHL greatness will be different than Vegas’ (July 22, 2021)
7. The Hockey News, Sal Barry: Expansion draft shenanigans from years past (June 20, 2017)
8. Down Goes Brown, Sean McIndoe: A brief history of NHL expansion drafts getting weird (June 14, 2017)
9. The Hockey Writers: Hockey History, the Second Six
10. The Hockey Writers: Capitals’ 1974-75 season, the worst ever
11. Russian Machine Never Breaks, Ian Oland: the Capitals since the Save the Caps campaign (Nov. 14, 2019)
12. The Globe and Mail, Roy MacGregor: Sens hoping to snap drought of back-to-back wins (Dec. 29, 2013)
13. Stats Crew: Ottawa Senators 1992-93 results, with neutral-site venues
14. 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.