In all sports, teams want to start a season hot. It's simple logic. If you're hot while others falter, you might be able to establish an insurmountable lead by the midpoint of a season, which affords you flexibility headed towards the postseason.
Last season, the Colorado Avalanche used their blistering start to set the tone for a surprising playoff year. Just a few weeks into this season, the Toronto Maple Leafs and Dallas Stars have blitzed to 4-0-1 and 4-1-0 respective records out of the starting gates. Is it time to start thinking playoffs in Toronto and Dallas? Our study says fans should rein in those expectations for now.
The Stars and Leafs have gotten off to hot starts this year, but on Oct. 20 last season the Atlanta Thrashers were 4-1-1, the Edmonton Oilers 5-2-1 and the Columbus Blue Jackets 5-2-0. This recent history doesn't mean the Stars and Leafs are destined for the bottom like those three, but it does show how short samples can misrepresent a team's true talent level.
A hot start is great, and points in the standings in October count the same as those in March, but the months in between are as important as the first and last of the season. An early winning streak isn't everything.
To analyze the importance of the season's first month, we first took each team's Pythagorean Expectation (defined here) as of Nov. 1. This is a way to calculate an expected winning percentage based on goals scored and goals allowed, and it correlates well with success dating back to the second world war (93.5 percent). Then, we correlated those early-season expectations to their Pythagorean Expected Win Percentage at the end of the season.
For each season, we determined a correlation coefficient and an 'R-squared.' The correlation coeffiecient measures how much one variable -- in this case, the expected win percentage as of Nov. 1 -- correlates to a second variable -- in this case, the expected win percentage at the end of the season. 'R-squared' talks about how they relate -- how much variation in the first variable can cause a change in the second.
Last season for example, there was a 62 percent positive correlation (and a .39 R-squared) between Nov. 1 expected win percentage and end of season expected win percentage. This has been moderately volatile over the last five seasons, although in four of the last five seasons the correlation has not exceded 62 percent.
Overall, we see some correlation here -- but not enough to think a hot start is the make or break factor in reaching the playoffs. Over a six-month season, talent levels come into play and teams will regress to the mean.
So is a hot start irrelevant? Are people getting overly excited about Dallas and Toronto? Probably.
We went back to the first season after the lockout and looked at the biggest differences in Nov. 1 expected winning percentage and end-of-season expected winning percentage. We used a differential of .20 as a cut off (fairly high to pick out the biggest early overachievers).
Team Pyth% Nov 1 Pyth% Final Diff Conf Rank Nov 1 Conf Rank Final
Montreal (08-09) 0.70 0.50 0.20 2nd 8th
Ottawa (07-08) 0.75 0.53 0.22 1st 7th
Columbus (07-08) 0.67 0.44 0.23 3rd 13th
Carolina (06-07) 0.69 0.48 0.21 2nd 11th
Detroit (05-06) 0.84 0.62 0.22 1st 1st
We found just seven teams with a differential of .20 or higher and three of those squads made the playoffs -- the 2008-09 Montreal Canadiens, the 2007-08 Ottawa Senators and the 2005-06 Detroit Red Wings. That last team might even be considered an outlier. Most expected the Wings to be a strong team, but they started with an insane 12-1 record with a plus-31 goal differential. That profound success is impossible to sustain, even for a 58-win Wings team.
Some other squads, like the 2007-08 Columbus Blue Jackets ( third in the conference on Nov. 1), completely fell back to Earth (finished 13th in the conference that season). The 2006-07 Carolina Hurricanes were similar. At second in the conference as of Nov. 1, they dropped to 11th by the end of the year.
That brings us to last season's Avs, a team that had 10 wins by Nov. 1. They didn't make our study because the differential between expected wins as of November and end of season was .12 (below the .20 cut-off). That differential means they weren't playing significantly over their talent level. The Avs made the playoffs -- thanks in part to the early accumulation of points -- and while the hot start certainly helped them hold on late, their early luck ran out. They lost 10 of their final 13 games. The three W's they did manage to grab required OT.
The results underscore the notion that teams can outplay their talent levels for days, weeks or even a full month of games at a time. It's about the bigger picture of the whole season. The teams with the most talent and best situational adjustments tend to win out, even if they must overcome a few games in the standings against the hot starters.
A version of this story originally appeared on ESPN Insider .
Follow Corey on Twitter at @coreypronman.
Corey Pronman is an author of Puck Prospectus, runs the statistical hockey site The Hock Project and is a writer for Premium Scouting. You can contact him at CPronman@gmail.com.
Corey Pronman is an author of Hockey Prospectus.
You can contact Corey by clicking here or click here to see Corey's other articles.