Success is the NHL playoffs is generally attributed to a number of things, the specifics of which vary from team to team and year to year. Rarely, if ever, is "being a really good hockey team" invoked as a reason that a team has a successful run in the playoffs. Typically, things like "heart" and "courage" and just "knowing what it takes to win" are mentioned as keys to winning in the second season.
It's unquestionable that some teams exceed expectations when they win the Stanley Cup, or even just by advancing to a conference final. But just because a team exceeds expectations over X number of games, does not mean that this performance level is their "true" performance level in that particular situation. Any one NHL team can beat any other on any given night, and this variance has a much stronger effect in the playoffs, when a short run of poor performance, even if it's due entirely to chance, can end a great team's season prematurely.
The ultimate question is, do some teams really perform better in the playoffs, or are the performances variances instead the result of chance? Is there something about these teams that allows them to excel in the postseason, even if they were merely good in the regular season? Surely, if a team is a "playoff team" one year, it will generally be the same the following year. If being a playoff team is the result of something about the makeup of a team, that should not suddenly disappear from one season to another. Here we will be examining team playoff results to see if we can shed any light on this question.
If we examine playoff results from 1980 (the first season of 16-team playoffs) to 2010, we have 30 seasons worth of playoff data to work with. In order to determine which teams exceed and which teams fail to meet their expectations, we calculate their expected wins in the playoffs, based on their games played against each of their opponents and using the Log5 Method based on regular-season winning percentages, including an adjustment for the number of home games the team played in those playoffs. Comparing a team's actual wins to its expected wins, on a per-game basis, tells us which teams exceeded expectations the most. Using this method, the team that most exceeded its expectations was the 1995 New Jersey Devils. The Devils are certainly "known" to be a playoff team, so perhaps the idea itself has some merit:
Team Year GP W E(W) Diff Diff/GP
New Jersey Devils 1995 20 16 8.15 7.85 .393
Edmonton Oilers 1988 18 16 10.05 5.95 .331
Pittsburgh Penguins 1992 21 16 9.35 6.65 .317
Detroit Red Wings 1997 20 16 9.86 6.14 .307
New York Islanders 1980 21 15 8.66 6.34 .302
Anaheim Ducks 2003 21 15 8.77 6.23 .297
Toronto Maple Leafs 2001 11 7 3.96 3.04 .276
Minnesota North Stars 1981 23 14 7.67 6.33 .275
Anaheim Ducks 2007 21 16 10.43 5.57 .265
New York Islanders 1983 20 15 9.81 5.19 .260
Of course, just looking at successful teams in the playoffs can skew our view of things, so perhaps we should look at the bottom of the list as well. Here are the team that fell short of expectations by the most, with a minimum of five games played:
Team Year GP W E(W) Diff Diff/GP
New Jersey Devils 2004 5 1 2.55 -1.55 -.310
Dallas Stars 2004 5 1 2.51 -1.51 -.302
Chicago Black Hawks 2002 5 1 2.48 -1.48 -.296
Minnesota Wild 2007 5 1 2.45 -1.45 -.290
Boston Bruins 1996 5 1 2.42 -1.42 -.284
Pittsburgh Penguins 2007 5 1 2.39 -1.39 -.278
Toronto Maple Leafs 1990 5 1 2.36 -1.36 -.272
Buffalo Sabres 1989 5 1 2.30 -1.30 -.260
Chicago Black Hawks 1988 5 1 2.25 -1.25 -.250
Phoenix Coyotes 2002 5 1 2.24 -1.24 -.248
only a year after having won the 2003 Stanley Cup , the Martin Brodeur-led Devils turned in one of the worst playoff performances ever. Expected to win 2.55 games out of five with the Flyers, they won only a single game. That's a bit odd, considering that the Devils are the posters boys for playoff teams.
Of course, you might say "that's an awfully small sample size", and you'd be right. But we're dealing with a claim most often put forward by those who ignore sample size all the time, dealing with "streaks" of five games or so all the time. But regardless, we're not going to stop here (although you should keep "small sample size" in your mind as we progress). Let's have a look at the best playoff team from each season, and see how they performed in the following year's playoffs:
Best playoff team, one postseason later
Team Year Diff/GP(N) Diff/GP(N+1)
New York Islanders 1980 .302 .126
Minnesota North Stars 1981 .209 -.385
Vancouver Canucks 1982 .166 -.230
New York Islanders 1983 .260 .022
Montreal Canadiens 1984 .257 -.051
Edmonton Oilers 1985 .233 -.149
Montreal Canadiens 1986 .226 .039
Quebec Nordiques 1987 .168 -
Edmonton Oilers 1988 .331 -.020
Chicago Black Hawks 1989 .217 -.038
Edmonton Oilers 1990 .213 .074
Minnesota North Stars 1991 .275 .106
Pittsburgh Penguins 1992 .317 -.118
St. Louis Blues 1993 .239 -.465
Vancouver Canucks 1994 .224 -.031
New Jersey Devils 1995 .393 -
Colorado Rockies 1996 .209 -.044
Detroit Red Wings 1997 .307 .182
Buffalo Sabres 1998 .186 .153
Buffalo Sabres 1999 .153 -.162
New Jersey Devils 2000 .193 .016
Toronto Maple Leafs 2001 .276 -.052
Ottawa Senators 2002 .143 .007
Anaheim Ducks 2003 .297 .071
Philadelphia Flyers 2004 .140 N/A
Edmonton Oilers 2006 .233 -
Anaheim Ducks 2007 .265 -.190
Pittsburgh Penguins 2008 .182 .208
Pittsburgh Penguins 2009 .208 -.034
Philadelphia Flyers 2010 .152 -.177
There are a number of repeat teams on this list, including some that were the best relative playoff teams in consecutive seasons: Buffalo in 1998 and 1999, and Pittsburgh in 2008 and 2009. The Ottawa Senators, though, make an appearance as the best playoff team in 2002, despite their reputation as one of the greatest playoff chokers of all time.
If the positive variance these teams achieved were merely a result of chance, what would we expect the following season? We would not, in fact, expect them to have a strong negative performance variance the following season. Instead, if the variance were random, we would expect their variance the following season to be near zero, which is the mean performance variance of all teams.
In the above table, we see some teams have consecutive positive variances, some have a negative variance the subsequent season, and some are near zero. The mean of all the subsequent season performances above is -.038. It's a small negative variance, surely small enough to not be considered significant. As such, the best playoff teams in a particular year tend to perform exactly as you would expect them to in the subsequent playoffs, based on their regular season performance that season.
In general, then, it seems that playoff teams do not exist. However, it should be noted that there remains the possibility that while there is no general trend, there may be specific teams that buck that lack of trend, so to speak. There are several repeat teams on the above list, after all; perhaps they are true playoff teams, while the others just got lucky. This is something we'll look at in the next and final installment, by examining the distribution of all playoff performances in our data set.