Preseason vs. Regular Season Success
The NHL is in the middle of its preseason, a time when young players try to prove themselves, coaches get a glimpse of what they’re working with, and fans get a preview of their team before meaningful games begin in October.
But does preseason matter? Can it provide accurate expectations for a team in the coming year?
The above graphs show each team’s points percentage in both the preseason and regular season in 2018-19.
The blue line denotes the point where both values are equal. Teams above the line performed better in the preseason and teams below the line performed better in the regular season.
The red line shows the points percentage required to make the playoffs in each conference, so any team to the right of the line qualified for the postseason last year.
The Western Conference shows mixed results. While some teams like the Anaheim Ducks, Colorado Avalanche, and Winnipeg Jets were near the even line, other organizations didn't come close to it.
One of those teams was the Edmonton Oilers, who earned 85.7 percent of the points possible in their seven preseason games. Even though that number was unsustainable over an entire year, it could have pointed to them being a strong club.
Instead, they earned just 79 points (48.2 percent), making them the 25th ranked team in the NHL.
The Eastern Conference shows a little more correlation, but still has its outliers in the Tampa Bay Lightning, Detroit Red Wings, and Carolina Hurricanes.
So, while there are occasions when a team’s regular season reflects their preseason record, it’s not something that happens consistently. In fact, only 14 of the league’s 31 organizations had preseason point percentages within 10 percent of their regular season results.
The reason this happened can be summed up in three factors:
Small Sample Size
Some of you might think using one season of data is too little to get an accurate conclusion, and you’re correct. But that proves my point of one of the issues with preseason results.
Teams only play between 5 and 10 games in September, which makes for unstable outcomes. If you want a case study on how far off small sample sizes can be, even in games that matter, look no further than the Buffalo Sabres.
The Sabres finished the month of November tied for first in the NHL after going on a 10-game winning streak and earning a 17 - 7 - 3 record. But, they would proceed to win just 16 more games that year and finish 27th in the league.
On that same date when Buffalo topped the NHL, the St. Louis Blues and Pittsburgh Penguins were ranked in the bottom quarter of teams. Both would go on to make the playoffs, and St. Louis would go on to win the Stanley Cup.
In these cases, a nearly 30-game sample size wasn’t enough to accurately project where the teams would end up. If we can’t make accurate predictions off of that, then we can’t use outcomes from seven meaningless games in September to do so either.
The preseason is a time for coaches to see how their up-and-coming prospects perform against tougher opponents, meaning undeveloped 18-year-old players get a hefty share of ice time compared to established NHL talent.
Joe Pavelski, for example, played only three of the San Jose Sharks’ six preseason contests in 2018. Meanwhile, unestablished center Antti Suomela appeared in five games.
During the preseason, organizations ice lineups that aren’t even close to what we’ll see when games start to count. It’s a mix of veterans, promising draft picks, and players that may never see time in the NHL.
Mid-Season Roster Changes
It’s impossible to know what will happen to a franchise’s roster during the season, but trades, injuries, and goaltending changes can have a massive impact on where the team finishes at the end of the year.
Last season, the St. Louis Blues were last in the league on January 2, behind even the Ottawa Senators. On that date, their chances of going to the postseason were roughly seven percent.
But on January 7, goaltender Jordan Binnington took over the starting role. He would go on to post a 24 - 5 - 1 record along with a .927 save percentage en route to securing his team a playoff berth and leading them to a Stanley Cup victory.
This mid-season change influenced an outcome that 37 games of data helped predict. If incidents like this can render larger sample sizes obsolete, then that further diminishes the meaning of preseason results.
The preseason is far too early, far too experimental, and far too meaningless to make predictions off of. What's important is seeing how players have developed over the summer and making a decision on who is good enough to make the NHL roster.
As long as everyone is progressing well, winning games doesn't matter.