Making Sense Of Petr Mrazek’s Career Performance

*Note this piece was written before the weekend games, so all stats are through Feb. 19th.

Earlier in the year, I wrote a lengthy piece describing how Petr Mrazek was having a sensational first full season. I closed out the piece with the following quote:

He’s been a winner at every level he’s played at. While I can’t 100% tag him with the “elite” moniker just yet, all signs point to Detroit having found their franchise netminder

Since this piece was written, we’ve seen Petr soar to great heights and then struggle mightily in his past few starts. He started 2016 blazing hot, posting a 10-2-1 record with a 1.31 GAA and a .951 SV%. However, his last three games have seen him go 1-1-1 with a 4.20 GAA and a .840 SV%. What gives? Is this the first crack in the armor and a sign of future struggles? Or is this a minor blip in the ascension of a future elite goaltender? For that matter, when can we even call a goaltender elite?

Mrazek Reminds Me Of A Familiar Someone…

Thus far, Mrazek’s career has been incredible, posting a 2.12 GAA and a .924 SV% through 80 career games. In fact, he seems to vaguely remind of someone else…

Yes, Mrazek’s numbers seem to compare very favorably to one James Tiberius Howard III through his first 80 career games. Right now, Howard seems to be everyone’s favorite punching bag (and I’m guilty as well). However, we can see that there aren’t really a whole lot of differences between Howard and Mrazek aside from 5v5 SV% and high-danger SV% (remember this). Does this mean Mrazek is doomed to see his numbers tank moving forward? After all, since Howard’s first 80 games, he’s done the following:

As you can see, Howard was unable to sustain the numbers he posted in the early part of his career and actually fell off in a big way in terms of SV%, dropping a full percentage point. Given this, how long until we know whether or not we can call Mrazek elite?

Can Mrazek Repeat His Performance?

At what point should we feel comfortable that Mrazek will be able to maintain his place in the top-10? Well, Don’t Tell Me About Heart (@DTMAboutHeart) took a look at how repeatable save percentage is on a year-to-year basis. He found that there was very little correlation between a goaltender’s overall, adjusted, low-danger, and medium-danger save percentages from season-to-season, and that the highest correlation came from high-danger save percentage, but even that was minimal. However, when I expand the sample to look at goalies who faced >2000 shots in two consecutive three-year spans, I get the following graph:

As you can see, the correlation is strengthened significantly as we increase the sample size. This should be intuitive, but I felt that it would still be useful to show it. There are some limitations of this data, particularly that selection bias exists. This means that only goaltenders good enough to face 2000 shots in the first three-year sample are likely to be the ones that see 2000 shots in the second three-year sample. However, that selection bias should impact both samples equally, minimizing its impact on the data. Keeping that caveat in mind, many have suggested that a sample size of >2000 or even >3000 shots may be necessary to have some degree of confidence that a goaltender will be able to repeat their performance from the previous sample.

What Does This Mean For Mrazek?

What I’ve shown above basically suggests that we can’t have a huge amount of confidence that Mrazek will be able to repeat his early career results, even though he has seen 2112 shots. However, there are a couple of promising factors that I’ve identified that make me think we’ll see a different career trajectory than Howard.

Going back to the comparison graph between Howard and Mrazek, you see that one of the areas where Mrazek was better than Howard was in High-Danger Save Percentage. As shown in the second section, HD SV% has been shown to be the most repeatable of the different measures of SV% and therefore is likely the best tool that we can use to evaluate a goaltender. Mrazek’s High-Danger SV% of 84.29% since his 1st NHL start does rank 9th out of the 46 goaltenders who have played >4000 minutes since 02/07/2013, which is a promising sign. However, check out some of the names on this list:

Names like Cory Schneider, Corey Crawford, and Braden Holtby have been thrown around as being members of the NHL elite this year, but you also see guys like Martin Jones, Michael Neuvirth, Thomas Greiss, and Cam Talbot on this list. Still, this does make me feel a little bit better when considering Mrazek’s future prospects.

Secondly, Eric Tulsky, formerly of Broad Street Hockey and now currently a member of the Carolina Hurricanes organization, found that as a goaltender racks up starts, 5v5 SV% starts to become a better predictor of future performance than overall save percentage. Specifically, he found that as a goaltender surpassed 100-150 starts, 5v5 SV% was found to be more predictive of future performance than overall SV%.

http://www.broadstreethockey.com/2012/1/25/2730816/goalie-save-percentage-projections-even-strength

Well it just so happens that the other area where Mrazek held a sizeable advantage over Howard was in 5v5 SV%. In fact, Mrazek’s 5v5 SV% of .9351 through 80 career games (71 career starts) ranks #1 among all goaltenders who have played >3000 minutes at 5v5 since the start of the 2007-2008 season.

Aside from Thomas Greiss, all of these goaltenders have been considered among the NHL’s elite at one point or another in their career. Mrazek currently tops this list, although it’s very important to note that he’s seen significantly fewer 5v5 shots than the other goaltenders on this list.

Bottom line, it’s still too early to make a confident determination as to whether or not Mrazek will be able to sustain his incredible performance so far. Just because he had two or three bad starts doesn’t mean he’s going to be the worst goalie in the world, just like he was never going to be able to sustain his hot start to 2016. He’s somewhere in between, and when it comes to the numbers that have been shown to be most predictive and/or repeatable Mrazek’s a high positive in those. I think that’s where he’s different from Jimmy Howard and why I still feel that he’ll eventually be regarded as one of the league’s elite goaltenders for years to come.

For more information about low-, medium-, and high-danger save percentage, please check out the War-On-Ice glossary. All data presented here is from War-On-Ice.

Will Petr Mrazek be an elite goaltender in the NHL?