Why Wilson is to Blame (Not Gustavsson)

Jonas Gustavsson Needs To Be More Mostrous

Editor's Note: This is an interesting FanPost by Van Ryn's Neurologist. By the time you are done reading it your brain might have melted. Basically, Ron Wilson might need to start updating his resume.

Mired in yet another mini-funk, there's been a lot of negativity floating around the Barilkosphere in the last 24 hours. Much of this has been directed towards Jonas Gustavsson. Specifically, there's been some suggestion that goaltending has been a bigger issue than other aspects of the Leaf's performance (defense, scoring, coaching,etc). 

Now, while I admit that The Monster's performance has been more diminutive of late, as a former goalie, it upsets me to see him slandered when, to my eyes, there are much bigger problems to deal with. 

So, I decided to look at some numbers. Now some of this gets a little complicated, so for those of you who aren't numerically inclined I apologize. Editor's Note: This is an understatement. Even if you are kind of numerically inclined this is going to get complicated. I think. I couldn't really figure it out.

Essentially, I wanted to look at the relationship between a number of measures of performance (Goals For/Goals Against; PP %; PK%; Shots/Game; Shots Against/Game; SV%), and success (wins). So, to start with, I calculated the correlation between these measures and regular season wins obtained from the 2009-2010 season. I've also provided the % of variance explained by each of these stats. Variance explained is the most useful for interpreting a correlation, and also for remembering that correlations are not causal. That is, if a given variable is correlated with another, you can predict the variance in one domain simply by looking at the values in the other. In this case, I want to know what factors can predict success (wins), and how much variance in wins can be predicted by each of these variables.

The results are interesting, though not entirely surprising. Overall, the strongest correlation is found in the 5-on-5  Goals For/Goals Against, explaining a whopping 69% of the variance in wins. I believe this is similar to the Pythagorean expectation that others have mentioned before. It's also painfully simple. If we score a lot more than we get scored on, we're going to win lots of games. End of story.

The more important question is, what affects our ability to score goals, as well as have goals scored against us, that ultimately results in wins? 

Somewhat surprisingly, PP% and PK% aren't correlated that strongly with the number of wins a team has. Essentially, the old adage that "special teams wins games" is wrong. Each of these things accounts for less than 20% of the variance in wins. Possibly, these weak correlations indicate that one of these things alone can't make or break your team. Chances are, if you are good at either PP or PK, you can get by, so long as you score more goals 5-on-5 than get scored on. If you are horrible at both, chances are you're going to have a difficult time getting wins (essentially, this is what the Leafs have been for the last 3 years: horrible at both). Interesting, PK may have greater importance than the PP, since if you account for both the 5-on-5 goals for/against, AND the PK, you can predict wins even better than with just the 5-on-5 for/against. 

Interestingly, shots for and shots against are both pretty good predictors, accounting for about 27-31% of the variance in wins. Thus, the Jason Blake method might have some advantages, so long as you can limit the chances in your own end. Actually, I think this amounts to the Babcock Effect: if you always have control of the puck, you'll get a lot of chances to score, and you'll limit the chances against on your less-than-stellar goalie. 

What about save %? If your goalie is horrible, shouldn't this lead to more losses? On the contrary, good sir! Of all the stats I looked at, SV% had the lowest correlation with wins, explaining about 16% of variance. 

Now I hear you saying "If your goalie is great, your 5-on-5 goals for/against should be higher. So shouldn't this affect wins indirectly?" In fact, even accounting for SV%, 5-on-5 goals for/against still accounts for about 64% of the variance in wins (.8 correlation). Thus, because a SV% can't translate into goals, SV% has a much smaller effect on wins than many of these other variables. 

So what does any of this have to do with Wilson? Well, in my mind, virtually all of the variables in that table, except perhaps SV%, can be affected by coaching. Skill certainly plays a part. But coaches need to find ways to get guys to score goals 5-on-5, generate chances, kill penalties, score PP goals, and keep the pressure off their goalie. This will be easier for some coaches than others, since some teams are more skilled than others.

However, to simply blame goaltending on our pitiful performance of late overlooks the fact that there are many other more important aspects to the Leafs play that are a much bigger concern than goaltending (like being last in PK and not scoring goals).  Because these aspects have a greater effect on whether we win a game or not, I'd say Gustavsson is not our biggest concern. 

So lay off The Monster.