Wednesday, February 20, 2008

Subjective Factors in Goaltending

How much weight should we give to subjective factors? Can scouts really evaluate the difference between two goalies? Are public perceptions of goalies often right? Do we properly value non-save skills in goalies?

These are important questions, especially for someone like Martin Brodeur who receives more glowing hagiography about his intangibles and his goaltending artistry than anyone since Roy. A lot of the feedback I get is along the lines of, "Haven't you ever seen him play?" The problem with putting a lot of weight into these reports, however, is that these effects are almost never observed in the data, and in a number of cases seems to suggest the complete opposite.

For example, here are a few common debate points, along with the numerical reality:

Brodeur never gives the puck away (According to NHL RTSS stats, he gives the puck away more than the opposing goalie at the other end of the ice in 2005-06, 2006-07, as well as so far in 2007-08).

Brodeur prevents shots (his backup goalies have faced essentially the same shot levels as he has).

Brodeur makes shooters miss more frequently through good positioning (He has been well below average in terms of making shooters miss this year (60th among goalies) and last (45th), source: Behind the Net)

Brodeur makes the big saves at the key times (Little evidence of that, and the playoff overtime performance sample suggests, if anything, the complete opposite).

Brodeur controls his rebounds very well...this one is probably true, but it is difficult to measure and is impacted by shot quality against and defensive play in front of the goaltender.

So there are several oft-repeated refrains in support of Brodeur. Other goalies have received similar treatment - e.g. Grant Fuhr and the "big save at the key time", or Patrick Roy's ability to "carry a team on his back". Both of these claims are just as doubtful as some of the ones listed above.

I think we have to clearly separate what is aesthetically pleasing, skillful and enjoyable to watch from what actually wins hockey games. There is very little margin in goaltending to begin with. The difference between good goalies is often something like one save in 50, a difference so small it is practically unobservable by the human eye. It can then become tempting to latch onto some other factor, like perhaps puckhandling ability, past playoff performance, or some other factor to break the tie, but this results in overweighting a single skill and ignoring the larger picture since by far the most important part of a goalie's job description is of course stopping the puck.

Two of the worst puckhandling goalies I have ever seen are Patrick Roy and Dominik Hasek. Both of them were responsible for many terrible giveaways, sometimes at the worst possible moments. Yet they both stopped a much greater percentage of shots than Brodeur did, which is why at the end of the day they were both significantly better goalies. Good puckhandling goalies are fun to watch, but the evidence is that their skills don't really make much of a difference in terms of winning or losing the game.

Another issue is clutch play, and more specifically whether it exists. Studies in other sports have found evidence of clutch play to be very slight or non-existent, and that generally players perform up to their level of ability in the clutch. Good players or players that have a lot of chances to succeed with the spotlight shining on them gain reputations as being clutch, when in reality their performance at the most important times is no better than the rest of the time.

So it is difficult to reconcile the subjective viewpoints with the objective record. I probably tend to weight the measurables more highly than some, which creates a difference of opinion. I believe that a skill, to be valuable, has to have an impact on winning hockey games, and if it does have an impact on winning games then that will sooner or later show up in some kind of observable way. But the existence of discrepancies between what is observed and what is measured ensures that debates will continue to rage, especially regarding Mr. Brodeur.


Jeff J said...

Great blog. I think you're spot on with pretty much every post and agree completely. Just to play devil's advocate...

"Can scouts really evaluate the difference between two goalies?"

Awhile ago there was a comment on the Sabermetric Research blog (mostly baseball, but lots of great ideas) that caught my attention. The proprietor postulated that if you took:
- a group of scouts who could watch players perform but had no access to stats
- a group of statisticians who had all the numbers but couldn't watch players,
then compared their evaluations, the numbers guys would be more valuable. I tend to agree.

That does not mean that subjective evaluations are always worse than stats, especially in a game like hockey where an individual's stats are affected so much by environment and luck. It is at least feasible that some individuals could produce subjective evaluations/predictions that outperform statistical evaluations/predictions. Given enough practice, the human mind can become very adept at things like pattern recognition. Top chess masters are competitive with the biggest supercomputers.

Again, I'm not saying I disagree with you. When you speak of public perceptions or blog commenters who say 'have you seen him play?' you're not talking about top chess masters.

"...he gives the puck away more than the opposing goalie at the other end of the ice..."

The best skaters have more giveaways too, partly because of icetime and partly because they're the ones with the puck. I wonder if Brodeur's puckhandling reduces the number of giveaways committed by his defense?

"Brodeur prevents shots (his backup goalies have faced essentially the same shot levels as he has)"

You might have addressed this in a previous post: Has his backup played against the weaker teams?

a witness said...

Another thing with perceptions and the common fan that i find annoying, especially in the case of Brodeur...

The context of comparison.

For most Brodeur fans, they have primarily only witnessed other goalies playing against New Jersey.

So you are comparing Brodeur's performance behind NJ to the performance of the other goalie playing AGAINST NJ.

As one who has watched MANY NJ Devils games, i know that New Jersey thrives on creating odd-man rushes. So often NJ makes it look like they have an EASIER time scoring than their opponent. So when Brodeur makes the ONE or TWO BIG saves...he STEALS the game, in the mind of a common fan.

More often than not, this hasn't been the case and Brodeur has not had to make as many BIG SAVES as his opponents have (or would have had to do in order to win).

The Contrarian Goaltender said...

Jeff J:

I remember reading the same thing at Sabermetric Research and agreeing with the premise. And if that is true for baseball, where individual contributions are more distinct, then it is probably much more true for hockey where teammates and opponents skew everything.

The best skaters tend to have more giveaways, true, but they often give the puck away attempting to create a scoring chance in the offensive zone. Giveaways by goalies are much more costly events, and since goalie puckhandling plays are usually very high percentage one giveaway probably outweighs many positive plays. Brodeur is a good puckhandler, but I agree with the school of thought that there "isn't enough margin in it", as Tyler Dellow once put it.

A Witness:

I very much agree with your comment about the perceptions of Devils fans. Another thing New Jersey does is play low-scoring games with few scoring chances, so each save is magnified, even the simple ones. New Jersey has also been successful, which means that they often have a one or two goal lead, and that makes the goalie look even better because his (mostly routine) saves are keeping his team in front. Of course goalies who play well in that situation can help their team win, but nearly every goalie in the league will stop around 90% or more of the shots they face so any goalie on a team that plays with the lead and gives up few shots will quite often be the hero just by statistical chance alone.