One thing that a lot of casual hockey fans seem absolutely convinced of is that the playoffs are a completely different animal from the regular season. That perspective no doubt contributes to the way that goaltenders tend to get judged on team success, and even goalies with strong regular season credentials will often be severely downgraded if their teams don't win much in the postseason (See: Luongo, Roberto).
There are some differences between how the game is played in the regular season compared to the playoffs, but most of them have little impact on the netminders. There are some extra things a goaltender has to deal with, such as more traffic in the crease, longer overtime games, and starting every game in constant high-pressure situations, which we would perhaps be expected to make the goalie's job tougher in the playoffs. That said, the numbers show that the average playoff save percentage is routinely higher than the average regular season save percentage.
The problem with evaluating playoff samples is always sample size. Plenty of Carolina fans are convinced that Cam Ward is an exceptional clutch performer and will perform the rest of his playoff career well above his regular season rates. It is possible that some of the 'Canes front office thought the same thing, given the contract that they agreed to with Ward that seemed to overpay given Ward's regular season performance to that point.
It turns out that Ward's continually improving regular season play has largely justified his signing, but with the Hurricanes again falling short of the playoffs, Ward's postseason sample remains stuck at two seasons and just 41 games played. Ward has had some great moments in his playoff career, but half a season's worth of games is really not enough to say much about Ward's postseason play with a high degree of confidence. History is littered with goalies that put up great numbers for half a season and then were barely heard from again.
A statistical analyst would typically use what we know about the population at large to inform his assessment, which is that most goalies have similar numbers in the regular season and playoffs, and declare that it is unlikely that Ward has special clutch skill and that he was probably just fortunate to hit a couple of well-timed hot streaks in 2006 and 2009. For a traditionalist fan who believes that playoffs reveal a player's true mettle, this is an unconvincing line of argument.
If there is one goalie that is universally agreed to be a clutch performer, it has to be Patrick Roy. With four Stanley Cups and three Conn Smythe Trophies, many consider Roy to be the greatest postseason goaltender ever, if not the greatest postseason performer at any position. Roy also has the largest playoff sample size of any goalie in history. As a result, nobody is better suited for a regular season vs. playoff comparison than St. Patrick.
A look at Roy's game-to-game save percentage results throughout his regular season career, comparing to the same numbers during the postseason, should therefore provide some insight into whether the two contexts appear to be similar or substantially different. If Roy's save percentage variance is roughly equivalent in the regular season compared to the playoffs that is at least one bit of evidence to suggest that we don't need to differentiate between those two competitive environments.
Appearances where Roy played a period or less, or faced fewer than 10 shots against, were excluded. Here is the distribution:
Save percentage Regular Playoffs
.799 or less 5% 4%
.800 to .850 9% 10%
.850 to .870 9% 6%
.870 to .890 10% 7%
.890 to .910 14% 12%
.910 to .930 15% 15%
.930 to .950 13% 14%
.950 to .975 18% 21%
.976 or more 7% 11%
Overall, those numbers are strikingly similar. Roy did have a few more mediocre games during the regular season (.850-.890) and a few more great games during the playoffs (.950+). That could be evidence of a small clutch effect, or perhaps may have something to do with there being tighter games and better defensive play in the postseason.
However, there is another possible explanation. It should be kept in mind that Roy played most of his playoff games in his eight seasons in Colorado, when overall league scoring was much lower than it was in the 1980's, which means his playoff record should be skewed slightly towards the higher save percentage numbers. As a point of comparison, Roy posted 78% of his playoff shutouts with Colorado, even though just 56% of his regular season shutouts came in an Avalanche uniform. That factor likely explains a lot of the discrepancy at the top end, and means that the underlying distributions are perhaps even more similar than they appear.
Roy still had some absolutely awful games in the playoffs. Some of them came at the worst possible times for his teams, including perhaps his most famous meltdown in Game 7 of the 2002 Western Conference Finals. Roy's rate of sub-.850 games was identical in the season compared to the playoffs (occurring 14% of the time). Even for the all-clutch St. Patrick, awful games happened at the same rate in meaningless regular season games as they did in all-important playoff matches.
Goaltenders always get blamed for bad playoff performances, based on the rationale that they should have raised their game in such an important situation. This attitude shows a simple unawareness of the fact that all athletes have unexpected bad games, and that luck plays a role. Sometimes a goalie is playing well enough but the other team is getting great chances and making their shots, and when the game ends he has four goals against on 25 shots and ends up having to shoulder the blame for essentially not managing to pull off the impossible.
For the most part, goalies with a high number of playoff games tend to see their playoff numbers fall closely in line with their regular season numbers. If Patrick Roy had almost the same distribution in meaningful games as he did during the first 82, that is further evidence against separating regular season and playoff performance. This means that there is probably little reason to differentiate between regular season and playoff results for goaltenders (i.e. to claim that an individual is great in the regular season and poor in the playoffs just because a small sample of results seems to indicate that), and that rather than considering both situations separately it should be possible in many cases to simply add together regular season and playoff results to increase the sample size for evaluating goalies.
There may be a few NHLers who simply can't withstand the pressure and are more likely to play poorly when it counts, but the vast majority of talk about netminders as playoff saviors or playoff chokers is probably nothing more than observers being fooled by randomness. In the long run, the best goalies in the regular season are also the best goalies in the playoffs.
Philip Myrland is an author of Hockey Prospectus.
You can contact Philip by clicking here or click here to see Philip's other articles.