With the new season just around the corner, NHL training camps are in full swing. As at other times during the offseason, the writers who cover the league are looking for anything resembling a story until there are real games to write about. Predictably, we’ve already gotten our first update on Pittsburgh Penguins goaltender Marc-Andre Fleury. On the first day of camp, Fleury apparently allowed goals on the first three shots he faced in a team scrimmage, and later allowed an awful “own goal”. This has led to a rehashed discussion of whether Tomas Vokoun should be Pittsburgh’s starter, and of Fleury’s future with the Penguins.
What’s interesting to me about all this is how difficult it is to separate Fleury’s play as a goaltender from the narrative that’s surrounded his NHL career. It began when he was drafted #1 overall by Pittsburgh in 2003. Goaltending prospects are notoriously difficult to project – so much of the position is mental, and it’s not easy to tell whether an 18-year-old kid will grow up with the right psychological makeup to be an elite NHL netminder – and as such they’re generally taken in later rounds. Henrik Lundqvist, for example, was taken 205th overall in 2000 by the Rangers; the Predators’ Pekka Rinne was taken 258th overall in 2004. In using the #1 pick on Fleury (and passing on drafting such future stars as Ryan Suter, Eric Staal, Jeff Carter, Ryan Getzlaf, and Corey Perry), Pittsburgh made a very bold statement about Fleury’s chances of becoming a star goalie. Without, of course, any evidence that he was capable of being that good in the NHL.
He entered the 2006-2007 season with a new contract and the starting job in Pittsburgh, and though his 0.906 save percentage that year was nothing remarkable, the following season appeared to provide evidence that the Penguins’ faith in Fleury was justified. He missed time due to an ankle injury in early 2008, but was stellar in the team’s late-season surge and deep run in the playoffs. When considering Fleury’s career, it’s important to remember that the 2007-2008 Penguins were an awful defensive team; without his 0.933 performance in that year’s playoffs, it’s doubtful they’d have gotten so close to winning the Cup. And at the time, that save percentage was what everyone focused on. If you were looking for foreshadowing, you could recall that the Stanley Cup-winning goal was scored when Fleury tried to cover the puck and inadvertently pushed it into his own net in Game 6. But no one considered that an ominous sign. The Penguins believed they had their franchise goaltender, and rewarded Fleury with a 7-year, $35M contract extension. The following season, of course, ended with Pittsburgh drinking champagne on the ice at Joe Louis Arena. Fleury’s performance in those playoffs is best described as uneven – his 0.908 save percentage was nothing special, and he had some awful games, including Game 5 of the Finals (where he surrendered five goals on 21 shots in two periods of work). But mistakes are easily forgiven when your season ends with a championship; what everyone remembered that June was Fleury’s sterling play in Games 6 and 7 of the Finals. He now had the most precarious of reputations, that of a “clutch playoff performer”.
So how did Fleury go from being “clutch in the playoffs” to being viewed as a liability who actually falls apart in the postseason? On one hand, this seems like a simple question. The Penguins have entered the playoffs in the last two years as Stanley Cup favorites, and in both cases, Fleury’s play in the first round has been awful. In 2012, you could make a good argument that Fleury cost Pittsburgh their series against Philadelphia; in Games 2, 3 and 6 alone, he allowed a staggering 17 goals on just 80 shots (yes, that’s a 0.788 save percentage), and despite scoring 26 goals in six games and dominating play at even strength, the Penguins were done before May. This past year, Fleury allowed an embarrassing (and game-winning) own goal in Game 2 against the Islanders, and had a disastrous Game 4 (six goals on only 24 shots), after which Vokoun took over as the starter for the rest of the playoffs. He performed badly in relief during a Game 2 blowout against Boston, and his offseason was dogged by rumors that the Penguins would part ways with him via compliance buyout, as well as reports that he was working with a psychologist on the mental aspects of his game.
Still, if we want to know how good Marc-Andre Fleury actually is, we need to understand that the truth lies somewhere between the two narratives of “clutch playoff performer” and “playoff choker”. This requires doing something that’s very difficult for a sports fan to do: look past the things that stick out in your mind and try to take in the big picture. As I found in some statistical work I did on goaltender performance this summer, there’s really no such thing as a goalie that offers steady, high-level play every time he steps onto the ice. Even the best goalies have awful nights, and backup-caliber netminders have their share of great starts as well. With Fleury, even if you focus on playoff games, he’s a very mixed bag: the big picture is much more complicated than the good games in 2008 and 2009 and the bad games in 2012 and 2013. On the good side, the same goalie who lost his job this past May posted a strong shutout performance in Game 1 against New York, and was also very good in Game 5 versus Philadelphia the year before. On the bad side, his troubles hardly began in 2012: there were a lot of reasons the Penguins lost Game 7 to Montreal in 2010, but the four goals Fleury allowed on 13 shots surely didn’t help, and Tampa Bay’s comeback from 3-1 down in 2011 was aided along by Fleury’s awful performances in Games 5 and 6. But still, in the context of his career, neither the bad games nor the good games earlier on come close to telling the full story: we’re talking about a couple dozen games in a career that’s spanned nearly 550 NHL appearances.
When you compare Fleury’s numbers to those of other NHL goalies over the past six seasons, his save percentage is right at league average, as is the proportion of game appearances in which he’s performed at an 0.850 level or lower. When it comes to the proportion of strong appearances (i.e., those with a 0.925 save percentage or higher), Fleury is just a hair below league-average (43% vs. 44%). When we want to judge what to expect from Fleury in the coming season, this is probably where we want to orient our thinking. He might not be worth the $5M he’ll earn this season, and (barring injuries) he’s not the best goaltender on his team. But all narratives aside, Fleury will most likely give Pittsburgh a league-average number of good games and a league-average number of bad games.
This is also why Fleury’s reputation is largely out of his control. Remember: observers only care about whether he can stop pucks in playoff games. The probability distribution of his single-game save percentages can give you a good sense of how he’s likely to perform over a large number of future appearances. But the broad variability in that distribution suggests that predicting how he’ll do in a small number of specific games is impossible. If he’s lucky enough to have some good games in the postseason, we’ll see the broader opinion of him shift again; if he’s not, his reputation will probably get even worse. But the thing to remember is this: “playoff choker” is just a storyline. There’s no way of knowing when those bad games might happen, or whether a stretch of five or six games will be bad or good.