Penalty Kill Save Percentage
by Scott Reynolds on Dec 8, 2009 11:30 AM MST
in Analysis
Email
PrintIf you're an avid reader of someone like Gabriel Desjardins, Tyler Dellow or Jonathan Willis you've already heard quite a lot about players and teams regressing to the mean. One such study was done by Tyler Dellow who looked at goaltending while on the penalty kill and found that goalies who had a high save percentage one year tended to regress in the following season. About this time last season I looked at goalies who had jumped out to a very good save percentage while penalty killing and predicted that they would, as a group, regress toward the mean and by the end of the year, each of the eight goaltenders who had started the year with a save percentage above .900 had seen their PK save percentage fall. Now, some fell further than others but a very high save percentage on the PK is an indication that a goaltender's overall save percentage has a good chance of declining as the season goes on and a very low save percentage on the PK is an indication that a goaltender's overall save percentage has a good chance of increasing as the season goes on. So which goalies have benefited by a high save percentage while penalty killing so far this season? We'll look at that after the jump.
For this study I looked at goalies who had faced at least 60 shots (mostly so that the study would be maximally relevant to the Oilers. 60 shots really isn't very many). One thing Tyler noted in his study was that a historical average for save percentage on the PK is about .866. I like to calculate "goals saved above/below average" and the number I'm using for "average" is that .866 figure. The two tables below are the goalies with a PK save percentage above .900, followed by goalies with a PK save percentage below .840. We should expect these goaltenders to regress somewhat toward .866 as the season goes on:
Now, I'm not trying to say is that there is no skill at all involved in save percentage while penalty killing. Surely there are some goalies who are good at it and others who are poor. Nevertheless, there are very few goalies who are this good and very few goalies who are this bad and there have been quite a few goaltenders who have jumped from one category to the other.
Last season only five goaltenders finished the year with a save percentage over .900 while facing over 100 shots. There were five others in 2007-08. On the other side, only eight goalies were worse than .840 over at least 100 shots last season and only six were worse than .840 over in 2007-08. Four goalies (Dan Ellis, Patrick Lalime, Alex Auld and Mathieu Garon) were on the "best" list one season and on the "worst" list in the other. So far this season we're seeing a very similar phenomenon. Jonas Hiller was on the "best" list in 2007-08 and is on the "worst" list so far this year. Brian Elliott was on the "worst" list last season but is on "best" list so far this year. There's a huge amount of variance, likely because the goalies are facing so few shots. This makes it very difficult to determine a goaltender's true skill, especially since it's quite difficult to know the exact impact the rest of the penalty killing unit has on the goaltender's performance.
So how will regression impact the Oilers? Well, Jeff Deslauriers' EV save percentage on .916 is currently tied for 33rd in the league (minimum of 100 shots) and if that ends up being a true indicator of his skill level, the Oilers aren't likely to be very happy. But at least they're not the Thrashers. Atlanta has really been leaning on Ondrej Pavelec so far this season and his overall save percentage of .914 looks respectable. However, as we can see in the table above, that's largely a product of his save percentage while penalty killing. At EV his save percentage of .908 is tied for 42nd. The Thrashers will likely need to have better goaltending than that if they hope to make the second season since, as a team, they've regularly been outshot.
6 comments
|
0 recs |
Do you like this story?
Comments
Well, Jeff Deslauriers’ EV save percentage on .916 is currently tied for 33rd in the league (minimum of 100 shots) and if that ends up being a true indicator of his skill level, the Oilers aren’t likely to be very happy.
I’m not exactly sure how this .916 overall sp is going to make the Oilers unhappy. All things being equal, the Wall has an overall sp of .909 and a higher GAA. To be fair, Bulin’s played in 6 more games, but I don’t think any of us were expecting JDD to have had this many at bats so early into the season. JDD has effectively played nearly half the games we were assuming he would get a mere 30 games in. With the reports out of the team suggesting that Khabi’s back is still screwed, the Oilers are clearly going to be riding JDD and this surprisingly solid (IMO) sp.
If the Oilers were going to be unhappy with JDD, I would imagine it would be more for his strange positioning, his inconsistency at covering the top corners at times, and his propensity to wander and lose his stick or the puck while outside the blue paint.
The PK sp is a strange indicator; what I mean is that I don’t totally buy it. The Oilers as a team have been brutal on the PK. Sure the goalie has to be the best player on the ice in that situation, but the PK system is a bit of a mess on its own. What are Bulin’s numbers like in comparison, both last year and this? How much of these apparent dropoffs have to do with the teams in front of the goalies. The Hiller example-don’t you think that the loss of CFP might be impacting the ducks PK just a little? Like I said, I really don’t by this stat as a real indicator of the goalie’s skill level as much as it is an indicator of overall team systems play
that other regular writer for bringing back the glory...
It’s actually a .916 EVSV%, tied for 37th in the NHL with Jonathan Quick, ahead of (among others) Dwayne Roloson, Marc-Andre Fleury and Cristobal Huet.
A posse ad esse.
The Copper & Blue|OilersNation|Hockey or Die!
Twitter: @JonathanWillis
Mail: jonathan.willis@live.ca
by Jonathan Willis on Dec 8, 2009 2:46 PM MST up reply actions
The only thing this is supposed to indicate is that anyone with a very high PK Sv% now will very likely see that regress over time and that regression will impact their overall Sv%. So, right now JDD’s overall number is .915 which is very good. Had he been average on the PK and allowed, say, three extra goals then his overall number drops to .907, a pretty significant drop. I don’t think the Oilers would be disappointed in JDD putting up .907 but it’s clearly a step back from what he’s done so far. That said, if the Oilers were making specific complaints to JDD himself I doubt save percentage would even come up since telling him, “Hey, your save percentage needs to be better” really isn’t going to help anything. I agree that they’re far more likely to focus on things like positioning or rebounds.
I think you’re right in saying that the defence will have an impact on how the goalies perform. As I said in the article, it’s very difficult to disentangle the goaltending skill from some of the other factors. But the biggest problem in evaluating goaltender skill with PK save % is sample size. The guy with the most shots against in either chart is Evgeni Nabokov with 147, which is kind of like five or six games. After five or six games there are often a couple of goalies in the NHL with an overall save percentage in the .950s or .960s and, on the other side, in the .800s or .810s. We know that they’re bound to regress toward the mean over time. It’s the same thing here (and even moreso for the guys at the top like Pavelec, Miller and Rask). We’ll take a look again at the end of the year but my prediction is that all of the vast majority of guys in both charts that get to 200 shots will have a final PK save percentage closer to .866 than what they have now, maybe even all of them.
by Scott Reynolds on Dec 8, 2009 3:28 PM MST up reply actions
I’m not exactly sure how this .916 overall sp is going to make the Oilers unhappy.
He isnt talking .916 overall Sv% He is talking about .916 EV strentght sv%. Consider the fact that his Sh sv% is going to drop, according to the analysis Scott did. To be a decent goalie in the NHL, his EV will have to be better.
Like I said, I really don’t by this stat as a real indicator of the goalie’s skill level as much as it is an indicator of overall team systems play
Now as you said our PK is bad, but our goalie is in the upper tier of NHL goalies in this stat. So then this shows excellent skills by JDD on the pk right? Consider how much worse our PK would have been if JDD had been average (about .860 sv%).
What are Bulin’s numbers like in comparison, both last year and this
This season Bulin is around .847
Last season his was .866
So we can expect him to be around .850-.860 range by the end
He isnt talking .916 overall Sv% He is talking about .916 EV strentght sv%. Consider the fact that his Sh sv% is going to drop, according to the analysis Scott did. To be a decent goalie in the NHL, his EV will have to be better.
My whole thing, misinterpreting the numbers or not, is based in the material validity of this particular statistical category. Maybe it’s my own limitations with mathematics coming into play, but the academic in me is simply having a tough time trying to see the “so what?” of this particular correlation. I’m just having a hard time with this, particularly based on the sample size as well as the subjective outliers (like team defense and systems play, shots allowed and games played to name a few) for being able to say with certainty that even if his pksv% drops towards the mean that JDD is not holding his own overall. I am just having a hard time understanding the purpose of this category is all….
Now as you said our PK is bad, but our goalie is in the upper tier of NHL goalies in this stat. So then this shows excellent skills by JDD on the pk right? Consider how much worse our PK would have been if JDD had been average (about .860 sv%).
Not to rip on your study Scott, I just don’t see it. I wish I could, as I am sure the advanced stat realm provides a lot of excellent insight. Yet even with what SumOil said above, I am not entirely convinced in the validity of the argument, again based on the really small sample size, especially in comparison to nearly all the other goalies in question. Like for example, where does the pk shots allowed compare in relation to total shots allowed? There are too many other variables the affect pk shots allowed to place the burden of potential suck on a goalie’s shoulders alone.
that other regular writer for bringing back the glory...
I guess what I’m trying to say is that, so far, JDD, among others, have a very high PK save percentage. It’s high enough that we can say with reasonable confidence it will go down. Is that JDD’s fault? We don’t know. Mayb eit is, maybe it isn’t, but the fact is that it will very likely go down.
What we do know is that goalies in general have a worse save percentage when shorthanded than they do at EV. We also know that PK save % does not repeat well. Goalies that have high PK save percentages one year do not have a tendency to maintain that the following season. I think this is likely because of sample size, i.e. goalies tend to face very few PK shots. We also know that one statistic used very frequently when evaluating goaltenders is overall save percentage and that the margin between “very good” and “about average” is paper thin. In JDD’s case the whole margin in his overall save percentage between “very good” (he’s currently at .915) and “about average” (he would be at .907 if his PK save percentage was average) is found in his superior PK performance. I’m not trying to say that JDD has been terrible or that he will be terrible. I am saying that JDD’s PK performance is hiding his actual ability if we’re using save percentage as a primary evaluative tool; in other words, it’s likely that his PK save percentage will normalize and that he’s more likely to be about average in terms of save percentage than he is very good going forward. And that’s what we see when we look at his EV save percentage which I think is a lot more reliable for predicting future performance because I think it’s a better (though by no means perfect) indicator of a goaltender’s true ability.
by Scott Reynolds on Dec 8, 2009 5:52 PM MST up reply actions
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
Northwest Standings
Oilers Schedule
Next 5 Games
@
Detroit
|
Wed 02/08 | 5:30 PM MST |
|
@
Ottawa
|
Sat 02/11 | 12:00 PM MST |
|
Toronto
|
Wed 02/15 | 8:00 PM MST |
|
Colorado
|
Fri 02/17 | 7:30 PM MST |
|
Vancouver
|
Sun 02/19 | 7:00 PM MST |
Last 5 Games
|
@
Toronto
|
Mon 02/06 | L 3 - 6 |
|
Detroit
|
Sat 02/04 | SO - W 5 - 4 |
|
Chicago
|
Thu 02/02 | W 8 - 4 |
|
Colorado
|
Tue 01/31 | W 3 - 2 |
|
@
Vancouver
|
Tue 01/24 | SO - L 2 - 3 |
In Association With
Latest Puck Daddy - NHL from Yahoo! Sports
- Puck Previews: Capitals, Panthers in Southeast showdown; Winter Classic announcement day
- Tribute to Owen Nolan: 10 moments that best celebrate his hockey legacy (Video)
- Puck Headlines: Canucks split Sedins; Ovechkin’s $4M house; Coyotes booster club folds
- The building challenges for Seattle, Quebec City as NHL markets
- Marek Vs. Wyshynski Radio: Sharks’ Drew Remenda, Owen Nolan tributes, accidental head-butts
Latest stories and headlines from NHL.com
- Former San Jose Sharks captain Owen Nolan retires after 18 season
- Owen Nolan announces his retirement
- Malkin takes the helm again as Penguins await return of Crosby, Staal
- Michigan regents call special meeting about renting the Big House to NHL for Winter Classic
- Michigan regents call meeting about Winter Classic
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
