Marc-Andre Fleury, Team Defense & Scoring Chances

A topic that interests me a lot is evaluating team defense. I've written about this previously in the context of scoring chances in the playoffs. This issue is amplified for the Penguins because it's a key component in evaluating Fleury's numbers. Fleury has put up average to below average numbers during his tenure in Pittsburgh--whether measuring by even strength save percentage or overall save percentage. Making matters worse is the fact that he's been really bad in the playoffs over the last five years.

His save percentage is powerful evidence that the Penguins have been paying far too much for a goalie. But a routine defense of Fleury's play is that he plays in an aggressive system that prioritizes offense and doesn't focus on limiting quality chances against. The result under this theory is that Fleury's save percentage is submarined by the fact that he faces higher quality shots on average than his counterparts on other teams.

We don't see an analytical basis for these claims. I read a great article the other day that looked at goalies throughout the league and sorted them by how often they saw shots from 0-10 feet, 10-20 feet, 20-30 feet, etc. Using distance as a proxy for shot quality is a quick way to see if teams really are giving up higher quality shots on average. You can see the table in the article, but I'll spoil the conclusion: the Penguins gave up shots in each range (0-10 ft, 10-20 ft, etc) nearly at league average rates. They weren't a bad defensive team last year.

A different way to present this data is to look at scoring chance percentage across all teams over the regular season last year. I am pulling data from Greg Sinclair's excellent super shot ninja website. What I am doing is taking the total number of shots each team gave up at even strength that are classified as scoring chances, and dividing that number by the total number of shots against at even strength (scoring chances and non-scoring chances). A lower scoring chance percentage means that a team is giving up less scoring chances as a percentage of total shots. This implies good team defense.

Data on total even strength shots against was pulled from War on Ice. The chart below contains the total number of raw scoring chances against for every team last year.

One caveat before diving into this. Some of the location data is not reliable, but given that these are thousands of entries for 30 teams, many of the errors become marginalized given the size of this sample. This still isn't perfect, but it's big enough where we can draw some moderately strong conclusions.

The numbers tend to conform, I think, with who we traditionally think of as good and bad defensive teams. Super stingy teams like LA, NJ, and STL are toward the top, and teams known for poor defensive play like BUF and TOR are near the bottom. The great thing for us is that the Penguins are sixth on this list. Given all of their injuries last year, they did an exceptional job of limiting the raw number of scoring chances against.

I've written about how the Penguins might have changed their system after the 2011 season to become more defensive. Whether that is true or not, the Penguins put up great defensive numbers last year when measured by raw scoring chances against.

The chart below looks at scoring chance percentage (the % of total shots against which were scoring chances).

This is a weird list. Toronto gives up fewer scoring chances as a percentage of total shots than New Jersey, which at first glance might suggest that Toronto is a better defensive team than New Jersey. But I'm confident enough in all of the other evidence we have to conclude that that's a ridiculous proposition.

What's going on here is that while Toronto gives up a lot of scoring chances, they give up so many total shots that the sheer volume of their shots against dilutes their scoring chance percentage. That isn't something to be celebrated; Toronto has serious problems that they need to fix. I am comfortable saying in this instance that with a season-long sample, the raw scoring chances against data is more informative than looking at the percentage of total shots against that are scoring chances.

Back to the Penguins, though. Even assuming the usefulness of this table, the Penguins once again rank very well. Only 44.6% of their shots against were scoring chances, which is nearly a top five number in the NHL. Again, with all of the injuries they had last year, this is very impressive.

It bears repeating that this data has some quality-control concerns. Nevertheless, it's a massive sample, and I think we can confidently say that in terms of allowing scoring chances against, the Penguins were very good last year. That's a credit to the players and the coaches for doing solid defensive work.

The darker note is that this doesn't bode well for Fleury's numbers. The charts above indicate that the Penguins did a really good job at limiting scoring chances against, and thus made Fleury's life easy at even strength. Nevertheless, out of the 31 goalies who played 1500 or more 5v5 minutes last year, Fleury's 5v5 save percentage of 91.93% is 24th on that list. That is comfortably below average. Since this is a contract year for Fleury, he has a lot to prove. Barring a stellar performance in the regular season and the playoffs that leads to (or very close to) a Cup win, the Penguins need to part ways with Fleury this summer.