A Deep Look At Fleury's 2014 Season

Fleury can be a lightning rod for Penguins fans. As long-time readers know, I've been critical of his play in the past. But I want to take a deep look in this article into what's behind Fleury's seemingly improved play this year, and most importantly, whether it's sustainable.

It's helpful to break hockey down by game state: powerplay, shorthanded, and even strength (5v5). Goalies only factor into shorthanded and even strength play to a meaningful extent because when their team is on the powerplay, they face few shots against. Consequently, to get a better picture of his real talent, I'm going to look at Fleury's play during even strength and penalty kill game states. But before we get to that, here is a graph of Fleury's even strength, unadjusted save percentage from the 09-10 season to the present, represented as a 25-game moving average (courtesy of War on Ice; all stats in this article current as of January 31).

You can see that his unadjusted save percentage has been inching up over time. Some people have made the claim that Mike Bales has helped stabilize Marc-Andre Fleury. This graph makes that proposition dubious. Fleury's recent decline is more extreme than some of his fluctuations in previous seasons, especially the 2012 season. If we're talking about stability, Fleury hasn't been all that different this year.

It is also important to separate what Mike Bales has done from Fleury's natural peak age. Goalies tend to play their best hockey around 28-29 years old (Fleury is 30), and as you can see from the graph above, Fleury's play has naturally become better and more consistent over time. If we're looking for coaching effects, we'd like to see a substantial change at the point Mike Bales took over. If we don't see that, it becomes very difficult to disentangle age effects from coaching effects.

Now let's look at the same graph but this time for shorthanded save percentage.

Fleury has routinely put up a stellar save percentage while shorthanded (12-13 season notwithstanding). It's obvious that there is far more variability in this graph compared to the even strength one above, but that's because PK time represents only a tiny fraction of total ice time for goalies. This small sample magnifies the randomness of goal-scoring, leading to higher peaks and deeper troughs. But this shouldn't take away from what Fleury has been able to accomplish while shorthanded over the past 5+ seasons.

We can break this data down to a more granular level by looking at Fleury's numbers year to year. The table below contains adjusted and unadjusted save percentages for both even strength and shorthanded ice time. It goes back to the start of the 09-10 season. (Note that adjusted save percentage from War on Ice accounts for shot location).

Two things stand out. First, Fleury's even strength play this year has been within career levels, both in terms of adjusted and unadjusted save percentage. Fleury has put up better numbers in individual seasons at least twice in past years, so if we think Fleury has improved, it's not while teams are skating five on five.

Second, Fleury's PK numbers are at a career high, both in terms of adjusted and unadjusted save percentage. And it's not because the team in front is doing anything differently. The Penguins this year give up the fifth-most shot attempts against while on the PK, the worst it's been during this time frame. So Fleury has been stopping more shots on average while also facing longer barrages than he has in previous seasons.

All of this leads to the big point: to the extent we think Fleury has improved, it's been entirely on the penalty kill. The question then becomes whether this is sustainable.

Fleury's PK prowess should be no surprise: he's got the second-best shorthanded save percentage among all goalies to play at least 800 shorthanded minutes since 2009. But trying to figure out how reliable these numbers are is thorny work. Fleury has only played about 1,800 4v5 minutes since the start of the 2009 season. He played nearly that many 5v5 minutes in just the lockout-shortened 2012-13 season alone. Given that we know goalies have outlier years (just like forwards and defensemen), and given that his total PK sample would be just a portion of one season's worth of 5v5 play, should we put much stock into Fleury's PK save percentage to date?

There is, unfortunately, no obvious answer. But research has shown that PK save percentage is not entirely random. It in fact shows some repeatability year to year, and does a very good job (relative to other metrics) in terms of predicting penalty kill success. Given Fleury's history of dominance while his team is shorthanded, it's reasonable to think that his current PK save percentage of 92% is close to his true talent.

But that doesn't mean he won't regress some; Fleury's shorthanded save percentage this season, for example, is currently better than Sergei Bobrovsky's even strength save percentage. We should expect Fleury to come down somewhere around his career average while shorthanded, which is about 89%. Still excellent, but not the level we've seen thus far.

One final point worth exploring is whether Fleury has maintained this level of play through his first 40 starts in prior years (the number of starts he has this season). The chart below contains that data, going back to the 2009-10 season (I skip the lockout-shortened year because there weren't enough games).

Nothing interesting here. Fleury's overall save percentage is in line with where he was the past two 82-game seasons. And it's not as high as the numbers he put up in the 2010-11 season. So this level of play from Fleury 40 games into the campaign shouldn't be considered unusually good or unusually bad.

***

The takeaway from this season is that Fleury hasn't put up numbers that are all that different from where he's been in the past, and to the extent he is playing better, it's been exclusively on the penalty kill (where additional regression should be expected). More broadly, there also doesn't seem to be a lot of evidence that Fleury has been more stable under Bales, especially when accounting for age effects.

Overall, we had about ten years of data on Fleury coming into this season, and a betting woman would have predicted that Fleury would be right around his career numbers. That prediction looks pretty good right now. When the Penguins re-signed Fleury for the next four years, I hope that no one in management operated under the assumption that Fleury would magically morph into a different goalie. But that is really neither here nor there. As we all know, Fleury will be roundly cheered or aggressively trashed depending on what happens in the playoffs.