Marc-Andre Fleury and Save Percentage: Part I

As a relatively new Penguins fan, one of the first games I ever watched was Game 5 of the 2008 Stanley Cup Final. As I'm sure many of you remember, that was the game where Fleury shed his mortality for about two hours and stopped everything he saw in overtime, giving the Penguins a chance to win it in triple OT. At the end of that game, for better or worse, I became forever endeared to the guy we call "Flower."

This year, Fleury unfortunately wasn't able to replicate the success he had in that Game 5. He was inconsistent throughout, and he posted his lowest single-season save percentage this year. The purpose of this series of posts is to try and figure out why. To do this, I'm going examine his play and divide it into two parts: even strength save percentage and power play save percentage. To clarify, I refer to power play save percentage as Fleury's save percentage while the other team is on the power play (this is also how NHL.com reports the statistics). I'm not going to examine Fleury's performance while the Penguins were on the power play because it's a minuscule segment of his overall play.

I'll be looking at save percentage to the exclusion of GAA, wins, and shutouts because it is the best measure of a goalie's performance irrespective of the team in front. GAA is dependent on shot volume against while wins are dependent on goals scored, neither of which a goalie has any control over. Shutouts happen too infrequently to draw anything meaningful from them.

In this post, I'm going to be looking at Fleury's power play save percentage this year. To compare him to other goalies, I used the top 30 goaltenders in power play shots against this year as my sample. Analysis after the jump...

Fleury finished this season with an .891 power play save percentage, good for 10th in this sample. The sample average was .879 and the distribution of power play save percentages was very close to normal (or in other words, a bell curve). In terms of power play shots against, Fleury was 7th out of 30 with 340. It makes sense that Fleury would have seen so many shots against on the powerplay because the Penguins spent the 3rd most time this season (508 minutes) playing 4-on-5 hockey. 

 

While I was looking at the shots against for goalies on the power play, I realized that it might be helpful to also examine the ratio of power play shots against to even stregnth shots against. I did the math for all 30 goalies in the sample [(PP SA)÷(ES SA)] and ranked them from highest to lowest. Fleury's PP SA ratio was 24.6%. Here's some info about the stat:

 

	PP SA Ratio
Mean	22.47
St. Dev.	2.89
Range	13.7
Minimum	15.8 (Brodeur)
Maximum	29.5 (Nittymaki)

 

The important thing at this point is to figure out in which context we want to analyze these numbers. We can compare the Fleury of this year to the Fleury of year's past, or we can compare the Fleury of this year to the 29 other goalies of this year in the sample. I think both contexts have merit.

I'll start by comparing Fleury against himself. Fleury posted the best power play save percentage of his career this year (.891). However, he also faced the least amount of power play shots against this year, as he saw 386 in 2005, 413 in 2006, and 361 in 2008 (he saw 184 in 2007 but he only started 35 games; it would translate to 352 in 67 games). His PP SA ratio also was the lowest of his career, as it was 37.6% in 2005, 29% in 2006, 26.9% in 2007 and 25.4% in 2008. In terms of the other goalies in the sample, Fleury's 24.6% PP SA ratio was 8th. Among goalies in the sample who played in at least 50 games, Fleury's 24.6% was 4th out of 20.

Fleury's collection of shots was one of the most difficult in the league this year because such a high percentage of them came on the power play. The average power play shot is of a higher quality than the average shot at even strength. Because of this, a goalie's power play save percentage is always considerably lower than his even strength or total save percentage. Here are the differences for the sample:

	EV SV%	TOT SV%
Mean Decrease	.042	.034
St. Dev.	.024	.019
Range	.096	.08

[For example, the average goalie's pp SV% is .042 lower than his even strength SV%.]

 

As a final thought, this analysis got me thinking about how the PP SA ratio might affect a goalie's total save percentage. Goalie's who have a higher ratio would have lower total save percentages (due to the higher quality of more of their shots). The correlation coefficient for these two sets of data (PP SA ratio and TOT SV%) was r = -0.28. While the data seems to be negatively related (my initial thought), the correlation coefficient is much too weak to draw any conclusions. Shrinking the sample to goalies with 50 or more games played actually weakens the correlation. Yet after I did this, I visually realized that there were three goalies who were seriously bucking the trend: Ilya Bryzgalov, Mikka Kiprusoff, and Evgeni Nabokov. They all had high total save percentages and high PP SA ratios. Once I removed these three goalies from the sample of those who played in 50 or more games, the correlation coefficient doubled to r = -0.4. Here's a scatter plot showing the 17 data points:

 

[Note: While this is extremely tentative evidence, I think it's encouraging enough to trigger more research. I'm going to collect data from more seasons to see if the correlation is stronger over a bigger sample.]

 

Overall, Fleury had a very good year as far as power play save percentage is concerned. His .891 was well above the sample average even though he faced the 7th most shots against and had the 8th highest PP SA ratio. His teammates made his job much harder this year by taking all the penalties they did, and forcing Fleury to make saves in situations in which shot quality is at its peak. But Fleury performed admirably and set a career high for himself in the process. Ultimately, Fleury's power play save percentage was not the problem this year. In Part II, I'll look at Fleury's even strength save percentage and the circumstances surrounding it, and hopefully be able to provide an acceptable explanation as to what happened this year to Marc-Andre Fleury.