Sunday, December 14, 2008

For every shot that hit his cheek he drew on a stitch mark (and I think he played with Brad Park)

Earlier this season when the Leafs were in a string of shoot-out games (losing most of them) I e-mailed Chemmy over at uber Leaf blog Pension Plan Puppets to see if we (ok, he) could find anything interesting from looking at three years of shoot-out data.

Was save percentage the key? Was there a shooting percentage at which teams found more success?

Our first cut at the data didn't produce much (who am I kidding, it was Chemmy's first cut, I just emailed him the idea and he crunched all the numbers).

Then I thought of that dusty old aphorism that you'll often hear tossed around by a colour guy during a hockey game, usually after a string of effective special teams play: if you add a team's powerplay and penalty killing rates together, successful teams have combined rates of 100% or better.

Or as the Toronto Star put it:

Adding penalty kill percentage and power play percentage gives you a number around 100. Teams whose total is greater than 100 are teams that are good on special teams, those under 100 are hurting in that category.
It was with that metric in mind that we took a second look at the data and found an interesting pattern.

Teams whose combined shooting and save percentage is around the 100% mark are hovering at .500 in the shoot-out. The higer a team's combined totals the more shoot-out wins they have; the lower their combined percentages, the fewer shoot-out wins.

Let's go to the blurry chart:

For the Leafs to start hitting the win column one of two things has to happen:

  1. If Toskala continues to put up a save percentage in the .500s, the Leaf shooters are going to have to shoot at 60% or better;
  2. If Toskala can get his save percentage up into the high .600s, the Leafs will only need to shoot at about 50%

The good news for the Leafs is they haven't been in many shoot out games lately...and as for any potential lost points, a top five pick at the 2009 draft would be a most welcome addition.


  1. I was trying to think of something clever to write about Toskala's...uh...seive-ness, but really I just wanted to give props to the Chixdiggit reference. Shweet.

  2. Since we're getting all mathy...

    Chi-square tests are a mathematical calculation you can run to compare observed and expected frequencies. Basically they tell you whether a discepancy between expected and observed is statistically significant, or whether it's within the expected range of outcomes based on random chance.

    If you run a chi-square calculation on the Leafs' poor shootout record using .500 as an expected outcome, they're just on the edge of statistical significance. If you use their actual winning percentage in non-shootout games as the expected outcome, then they are in the expected range (although just barely). I wouldn't be surprised if you'd get the same result for every NHL team (Ottawa might be an exception).

    In english: for all the hair-pulling about why some teams do well in shootouts and others do poorly, the math says that it's possible that shootouts are pure luck and the differences in won/loss record can just be explained by random chance.

  3. Clever Chixdiggit reference...sometimes I [feel](like) Gerry [Cheevers] to...

    seriously though, great work. fascinating stuff. what would happen if MSM types start putting this much thought and effort into their writing...?