Tuesday, August 19, 2008

Goalie Fatigue

I have argued for a long time that games played for goaltenders is an overrated stat. A goalie does not become better than another goalie simply because he played more games. He could potentially be described as more valuable, but there are many factors that determine games played and nearly all of them are out of the goalie's control. That is not to say that the games played stat is useless - it helps us determine whether a performance was likely to be a fluke or not. For example, if we are evaluating whether Ty Conklin is likely to repeat his 2007-08 performance where he posted a save percentage 20 points above his career average, we know the fact that he did it in only 33 games implies that it is much more likely it was a lucky season rather than a true measure of his skill.

The justification for the importance placed on games played seems to be that it is more difficult to play more games. I remain unconvinced that playing additional games is significantly more difficult. I think that virtually every goalie in the NHL is capable of handling the workload. If they weren't, then they would have failed on some lower level (junior, college, minor-league) when they were asked to play 70-75 games a season.

But what do the numbers say about it? Yahoo Sports provides breakdowns of month-by-month stats, so I thought it would be interesting to see if goalies tired as the season went on, which is what would be expected if fatigue is an issue. To ensure that I had a full selection of statistics and that era issues wouldn't come into play, I chose a group of goalies that have played a lot of minutes in recent years and that began their careers relatively recently (e.g. late '90s or later).

Here is the list of goalies in the sample: Roberto Luongo, J.S. Giguere, Marty Turco, Tomas Vokoun, Evgeni Nabokov, Miikka Kiprusoff, Jose Theodore, Nikolai Khabibulin, Rick DiPietro, Marc Denis, Ryan Miller, and Henrik Lundqvist.

I broke down their statistics by month to see what the results were (minutes, win %, GAA, save %, shutouts):

October: 543 GP, 2.71, .905, .497 win %, 28 SO
November: 730 GP, 2.53, .912, .531 win %, 53 SO
December: 774 GP, 2.48, .914, .547 win %, 54 SO
January: 768 GP, 2.58, .910, .521 win %, 53 SO
February: 653 GP, 2.56, .912, .535 win %, 53 SO
March: 817 GP, 2.53, .913, .539 win %, 53 SO
April: 242 GP, 2.52, .913, .566 win %, 14 SO

What do the data show us? That the goalies tended to start slowly in October, and after that kept up a similar level of play the rest of the season. Their best month was December while their worst month was January, but there was only a .004 difference in save percentage and a 0.10 difference in GAA between the two. Even though these goalies played big minutes all throughout the season, they did not get worse as the season went on.

Not only that, but the goalies played the most games and minutes in March, late in the season, and their performance did not deteriorate. In fact, they were more likely to win games and stop the puck in March and April than in any month except for December. The correlation between save percentage and games played was essentially zero.

However, there could be some goalies that deviate from this trend. Perhaps some iron man goalies improve as the season goes on and get better when playing big minutes. If anybody fits those criteria it would be Martin Brodeur, who year after year plays more games than anyone else. However, Brodeur's splits are are virtually identical to the splits of his goaltending peers. October is easily his worst month (2.45/.904). He peaks in January (2.09/.920), and also does well in April (2.08/.919). The rest of the months are pretty similar, ranging from 2.12-2.22 in GAA and .911 to .916 in save percentage (gaps of .10 and .005 respectively, again virtually exactly the same variances as the larger group).

So goalies apparently do not get worse as the season goes along, and Martin Brodeur is no different than anybody else in terms of monthly splits. This suggests that goalie fatigue is not really a major issue.

To look at this a bit further, I looked at the goalies who took on large workloads (75+ games in a season) to see if they tired late in the year.

I compared how well they did in games 71 or later compared to games 1-70. Here is the breakdown of the average results:

Games 1-70: 2.54 GAA, .910 save %, .567 win %
Games 70+ 2.59 GAA, .903 save %, .571 win %

So the goalies did slightly better in the earlier games, however it was not by much. And if there was a fatigue effect, it was again apparently something that affected Martin Brodeur just as much as everyone else - after game 70, Brodeur slipped from 2.18/.918/.595 to 2.20/.913/.588.

So it appears that fatigue may have an effect on goalies, but if so it appears to be a very small one. Furthermore, goalies who routinely play a lot of minutes like Martin Brodeur have not been better at maintaining their level of play late in the season than other goalies facing heavy workloads. The fact that goalies did far better late in the season than they did in the month of October implies that playing a lot of games may actually be beneficial to goalies, in that it is harder to overcome a summer off than it is to overcome playing 70+ games in a season.

This means that playing a lot of games is probably more accurately characterized as a choice, rather than a skill, at least for the large majority of NHL goalies. If this is correct, that means that Vezina Trophy voters have placed an inappropriate weighting on games played in recent years, as they have repeatedly voted for goalies that played more games over others that played fewer games at a high level.

Wednesday, August 6, 2008

Reviewing Tony Esposito's Playoff Record

I have to issue a mea culpa here - in light of further evidence presented by Bruce in the comment thread to my last post on Tony Esposito, I think it is clear that Esposito did play poorly in the playoffs in the early 1970s and most likely contributed to his team's underachieving playoff record. The key here is the contrast between Chicago's record against Boston and Montreal in the regular season and in the playoffs:

Chicago vs. Boston and Montreal in the regular season (1970-1974)

25-24-11, 175 GF, 177 GA

2.92 goals per game
2.95 goals against average

Chicago vs. Boston and Montreal in the playoffs (1970-1974)

7-16, 71 GF, 101 GA

3.09 goals per game
4.39 goals against average

Boston and Montreal were both outstanding offensive teams, but something must have gone very wrong with Chicago for them to allow that many extra goals compared to their seasonal rates. The Hawks were good enough to hold their own during the regular season, but were shellacked in the playoffs 4 times in 5 years. I doubt this was entirely Esposito's fault, but given that unusually high GAA number it is hard to see how he couldn't have contributed to it.

We have one season of save percentage data from the early 1970s (1971), where the numbers indicate that Esposito faced an average of 28 shots per game. Let's say that Chicago allowed an average of 30 shots per game in their regular season games against Boston and Montreal, since they were strong opponents. That would put Esposito at a .902 save percentage, which according to the Hockey Analysis Group was the average save percentage from 1970-1974. Even if Chicago was badly outplayed in the playoffs and allowed an extra 10 shots per game against the same opponents, Esposito's save percentage would still drop to a mediocre .890. If they allowed an extra 5 shots per game, he would be at .875, and if he was still facing the same 30, the stat would be a horrific .854. Of course Esposito didn't play all of those games and there may be some empty netters involved, but the numbers indicate that the goaltending was probably subpar.

Later in his career, Esposito's teams were weak and didn't provide much goal support, which is why he suffered through a string of playoff sweeps and early exits. But the early 1970s Blackhawks were Cup contenders, and could have won a Cup or two with strong goaltending. Esposito had some great games (2nd in playoff shutouts during the period) and series (Chicago twice upset strong Rangers teams), but never played well for an entire series against Boston or Montreal, which is where the Blackhawks repeatedly saw their Cup dreams dashed.