What do NHL GMs think is Replacement-Level Goaltending?
by Hawerchuk on Feb 24, 2011 7:00 AM EST
in Statistics
Email
PrintThe market for NHL goaltending has undergone a significant shift over the last few years as total goaltender salary has dropped over 20%. Teams have been able to find quality goaltenders like Jonas Hiller and Niklas Backstrom for nothing and are no longer willing to pay top dollar to average- to below-average goaltenders like Evgeni Nabokov and Marty Turco. Based on what GMs have been willing to pay for goaltending, what do they perceive replacement-level goaltending to be?
In order to figure this out, we need to know two things: 1) team replacement level; 2) replacement-level save percentage. This is non-trivial - team replacement level is the subject of much debate; is it 56 points, the worst single-season total since the lockout? Is it 39, Atlanta's total in its first season? It turns out that it doesn't much matter:
| Repl | Marg Pts | Marg Goals | 8% | 10% | Marg Sv% |
| 56 | 26 | 78 | 6.2 | 7.8 | 909.3 |
| 39 | 43 | 129 | 10.3 | 12.9 | 906.4 |
Remember that 3 goals = 1 point in the NHL standings. If we set replacement at 56 points and goaltending at 8%, we find that a league-average goaltending tandem is 6.2 goals better than a marginal one. If we set it at 39 points and 10%, then goaltending is 12.9 goals above replacement. The true answer falls somewhere in the middle, which implies that replacement-level save percentage is approximately 907.8.
What does this say about goaltender salaries? The best goaltender in the league might have a true talent save percentage of 920. If he could consistently play 74 full games in a season, he'd be approximately 25 goals above replacement, which translate to just slightly over four wins, and approximately $11M on the free agent market. (No such goalie exists.)
Could replacement save percentage be higher? I'm open to this idea, and the downward pressure on goaltender salaries suggests that GMs are too. And could replacement level be significantly lower? This doesn't seem likely - it would imply that 1) goaltenders are systematically underpaid; 2) GMs should have difficulty finding league-average goaltending and 3) many goaltenders should be paid the league maximum because the value of their talents exceeds $7M. That's simply not true - ask Evgeni Nabokov if he felt like there was a lot of demand for his services in the off-season!
The bottom line: goaltenders drive something like 8-10% of winning in the NHL. You could probably even come up with a system that said they were worth 1/6 of all wins. But significantly more than that, say 27%? Not a chance.
19 comments
|
1 recs |
Do you like this story?
Comments
This is interesting. I’m surprised that replacement level is that high. Tyler’s study is pretty old now, but from 1998-99 to 2007-08, he had the average Sv% as .907, which is actually a bit lower than what you have as replacement level. Where would you set average, and what is that based on?
Abney, Abney, oh why TF did we have to pick Abney?
I got avg = 912, but that varies y2y. Repl = avg – 4 or so
by Hawerchuk on Feb 24, 2011 5:09 PM EST via mobile up reply actions
Is that number weighted to allow for the fact that 30 of the league’s 60 goalies are playing the vast majority of the games?
Maybe I’m not following, but it seems to me that there’s probably a wide gap between league-average goaltending and league-average backup goaltending based on GP.
A posse ad esse.
The Copper & Blue|OilersNation|Hockey or Die!
Twitter: @JonathanWillis
Mail: jonathan.willis@live.ca
by Jonathan Willis on Feb 24, 2011 11:42 PM EST up reply actions
So how did you calculate average? Is it just the sum total of all of the shots faced by NHL goaltenders this season (i.e. each shot as opposed to each goalie treated as equal, and data from just this season)?
Abney, Abney, oh why TF did we have to pick Abney?
by Scott Reynolds on Feb 25, 2011 2:42 AM EST up reply actions
The thing is, that’s going to overvalue the contribution of the good goalies since they play more often. If we look at just goalies who have faced 250 shots (there are 61 of them) this season, and weight each shot as “1” we get an average save percentage of .912. If we weight each goalie as “1” we get an average save percentage of .910. It’s not a huge difference, but I do think it matters, and I’m confident that the effect would be consistent.
Further, if we knock off the thirty goalies who have faced the most shots (i.e. a loose approximation for the starters) and weight either each shot or each goalie as “1” we get an average save percentage of .906 (the each shot method is very slightly higher). I’m not all that comfortable saying that the average backup goaltender is somewhat below replacement level, but maybe I just have a bad understanding of what replacement level is supposed to be (my understanding is that freely available talent should be expected to reliably provide replacement level performance).
Abney, Abney, oh why TF did we have to pick Abney?
by Scott Reynolds on Feb 25, 2011 3:21 PM EST up reply actions
You have starters who are freely-available talent too, though. So I don’t think you can knock off the top 30 goalies.
I’m not sure that I’d want to knock starters off the list entirely either, but calling the average NHL backup below replacement level seems wrong to me. Or does that just imply that most teams are very poor at judging goaltender quality?
The biggest fanana of the Havana Bananas
by Scott Reynolds on Feb 26, 2011 12:12 PM EST up reply actions
It could also be the result of the high level of luck involved in goaltending. For instance, if all backups were replacement level goalies, you’d expect that half of them would produce below replacement level performance each year.
While this is true, my assumption would be that most backups should be better than replacement level, some of them significantly so.
The biggest fanana of the Havana Bananas.
by Scott Reynolds on Feb 26, 2011 8:21 PM EST up reply actions
I can’t help but feel that consistency plays a major role. Would the distribution of save percentage over a full season look similar for both a very talented and a replacement level goalie (around a different mean of course)? Are the “best” goalies simply more consistent at being average or better, while others are more inconsistent?
Another consideration in a cap-world (and it ties in to your Reimer post) – how can you be sure that you will GET a replacement level goalie? On average, sure – pick up a guy from the Swiss league and you can expect a certain sv% – but any one ’tender could be much better, or much worse. As a GM, would you be willing to pay a small premium for a goaltender that has an observed save percentage (over a decent sample size) equal to replacement level over an unknown commodity?
Is there a website where I could quickly download game-by-game stats for goalies for the past few seasons? I would be interested in looking at the distribution of sv% for individuals to see what sort of differences exist. In particular, I think it would be interesting to measure the degree of kurtosis between, an consistently elite goaltender (Vokoun?) and one that ranges from average-to-elite in any given year (Bryzgalov?).
by Ricci Handsome on Feb 26, 2011 11:27 AM EST reply actions
Is there much consensus on which goaltenders currently in the league are definitely “elite”?
I did a quick search yesterday (going back to 2006-2007) that a) played at least 40 games (25 in 2010-2011), and b) ranked in the top 15 in save percentage. A completely subjective measure I know, but something to start with. Here were the ’tenders that met the criteria in…
Five Years: Vokoun
Four Years: Lundqvist, Thomas, Backstrom, Luongo
Three Years: Miller, Brodeur, Bryzgalov, Hiller*, Lehtonen
Two Years: Nabokov, Kiprusoff, Ward, Riine, C. Mason, Biron, Giguere, Huet, Price
I put an asterix by Hiller since he has hit the mark in three years straight since he became the starter in Anaheim (interesting to note that Giguere met the criteria in the two years when he was their starter).
Would most agree that it is fair to assume that Vokoun, Luongo, Thomas, Backstrom, Lundqvist and possibly Hiller are “elite”? At least as a starting point?
If someone else has already done the work – stop me now… (but show me the link please!).
by Ricci Handsome on Feb 26, 2011 11:40 AM EST reply actions
This is the subject of much debate. Vokoun and Lundqvist benefit from their home scorers; some see Backstrom as a creation of Minn’s defensive system. Luongo is probably unassailable. Brodeur should be up there – the NJ scorer undercounts. The bottom line is that there isn’t much of a discernable difference among the top/10
by Hawerchuk on Feb 26, 2011 11:59 AM EST via mobile up reply actions
Is there a spot I can download the game-by-game stats for an individual player without needing to try to create it myself? And could you link the deflators for shot count in each arena?
by Ricci Handsome on Feb 26, 2011 12:15 PM EST reply actions
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Related FanPosts on Arctic Ice Hockey
FanPosts
Community blog posts and discussion.
