Is it Possible to Determine Goaltender True Talent on the PK?
by Hawerchuk on Mar 5, 2010 9:00 AM EST
Email
PrintThere's a huge amount of year-to-year variation in goaltender performance. The general consensus is that save percentage converges to true talent levels after approximately four full seasons, though goaltender true talent is rarely stable for four seasons.
A significant portion of the variation in overall save percentage is variation in short-handed save percentage (and, to a much lesser extent, variation in PP save-percentage.) If a goalie posts a very high SH save percentage one season, should we read anything into it? Not likely. This chart shows the regression to the mean of ES and SH save percentage for pairs of one season and two seasons:
| Years | ES | SH |
| 1-2 | 78% | 96% |
| 1/2-3/4 | 64% | 93% |
Basically, Even-Strength save percentage, which is measured over a large number of shots, is a significant predictor of future performance, though it must be heavily-regressed to the mean save percentage. SH save percentage, on the other hand, is essentially random. Your best guess of next year's SH save percentage is the league average. Indeed, last year's even-strength save percentage is a much better predictor of next year's SH save percentage than this year's SH save percentage. No surprises here - lots of shots, even at even-strength, give us a better estimate of a goalie's talent level than a small number at any other strength.
15 comments
|
1 recs |
Do you like this story?
Comments
90%+! That’s a pretty wicked regression to the mean there.
All things Thrashers + stats: www.birdwatchersanonymous.com
I got it from NHL.com. You can figure it out at 5-on-5 from the goalie sheets on my site, but nobody ever looks at the goalie stats, so I haven’t put it in there automatically…One of the many things I should do!
Even-strength save percentage is the best measure of a goalie’s ability. I like to use 3 or 4 years of data…
ES sv% is better than overall sv%?
I’m not questioning, just curious, since you’re reducing your sample size. It would be very interesting, with regards to the Habs’ goaltending situation, if EV save% was the best predictor.
ES sv% is better than overall sv%?
I would imagine if everyone played the same ratio of EV-to-SH minutes then you would use overall SV%, but since some goalies just get shelled more because their team takes more penalties, overall SV% isn’t too illumintating.
by R O on Mar 5, 2010 3:06 PM EST up reply actions
Even if the EV/SH ratio was constant, the variation from year to year in SHSV% is much larger than EVSV%.
I think that’s because of small sample size.
EVSV% also varies substantially, just not as much as SV%.
It’s all a sample size problem why PKSV% alone can’t be trusted to reflect a goalie’s abilities on the PK. But if every goalie played the same amount of PK minutes then overall SV% might become more useful. Just a bigger sample.
Certainly there might be questions about shot quality on the PK affecting the stat. But shot quality doesn’t appear to look all that big an effect at EV, it’s possible we overestimate its effect on the PK as well.
by R O on Mar 6, 2010 4:36 PM EST up reply actions
Tyler Dellow covers the randomness of SHSV% (or PPSV%, depending on your perspective) quite nicely, and Scott Reynolds adds to the discussion in a couple of posts.
The variance does come down to small sample size, but because the SHSV% varies so much more than EVSV%, even if the ratio were the same for every goalie EVSV% would still be more stable from year to year and, as this post shows, a better predictor of future performance.
But if the ratio was the same it would make overall SV% a better apples-to-apples comparison among goalies for a given year, so I guess it depends on what you want to do with it, judge how good of a year a goalie had or predict how he’ll do in the future.
I look at goalie stats! Bring on the save percentage columns!
Is there a any way to extrapolate the goalie part of PK from ES performance and assign the remainder to the PK unit in front of him. I’m thinking specifically of Vokoun. ES his save percentage computes to 0.938. PK I get 0.847. What “should” his PK be?
by DoctorMyBrainHurts on Mar 5, 2010 9:23 PM EST up reply actions
Does this mean Vesa Toskala gets pardoned for his ridiculously bad PK performance earlier this year?
I’m anticipating the answer: Toskala shouldn’t be pardoned for anything.
Bettman's Nightmare: A Blog Where Hockey Aficionados Dismantle That Mighty Empire, One Balsillie at a Time
http://bettmansnightmare.blogspot.com/
by Bettman's Nightmare on Mar 5, 2010 5:24 PM EST reply actions
Toskala’s still a pretty bad EV goaltender, although some of that has to be that his team has an asinine strategy for defending leads.
If this is a roundabout way to diss the McElhinney-Toskala trade though… well you’re about two days late to the bandwagon doncha think?
by R O on Mar 5, 2010 5:32 PM EST up reply actions
Toskala got traded?
J/k. No, I really could care less about the trading of backup goaltenders. But it’s funny to hear that some other people did.
The only reason I brought it up is because I recalled Gabe’s previous post on the Leaf’s and their PK.
Bettman's Nightmare: A Blog Where Hockey Aficionados Dismantle That Mighty Empire, One Balsillie at a Time
http://bettmansnightmare.blogspot.com/
by Bettman's Nightmare on Mar 5, 2010 6:44 PM EST up reply actions
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
