from Amazon
Sunday, July 26, 2009
The uncertainty of save percentages
I thought it’d be interesting to explore the contrast between Martin Biron and Nikolai Khabibulin a bit more. In a previous post, I noted that they’ve both faced a lot of shots since the lockout and that there was a substantial gap in their performance, with Biron having faced 5605 shots with a .912 save percentage, while Khabibulin faced 5628 shots and stopped them at a .904 clip. A 47 goal difference or so over five years, which is a lot - you’re basically talking about two wins a year.
...
Anyway, after I identified my pairs that met the criteria, I proceeded to score the thing. Basically, if the “after” save percentages were within +/- .004 of each other, I scored it a tie. If the goalie with the better save percentage coming in posted a save percentage of .005 or more above that of the fellow with a lesser save percentage, I called it a win. Otherwise, it was a loss. The players identified as having the better save percentage posted a record of 104 wins, 65 losses and 78 ties.
Facing 5600 shots means that one standard deviation (SD) from the mean is .004. So, if you have two goalies who are separated by 8 points, that puts them 2 SD apart. Tyler’s experiment gives the better goalie a win% of .580, which looks consistent with being 2 SD apart: that is, we are pretty sure there is a non-zero difference between the two, but one is not the clear-cut winner over the other.
On Monday, I’ll work up an experiment that tells us what true save percentages would be consistent with a .580 record that Tyler found. My guess, just a guess for now, is that their true save percentages gap is less than half what we’ve seen from them.
Recent comments
Older comments
Page 1 of 344 pages 1 2 3 > Last »Complete Archive – By Category
Complete Archive – By Date