Last post on Regression to the Mean. I swear.
by Hawerchuk on Oct 21, 2009 9:00 AM EDT
4 comments
There are lots of statistical concepts that I was introduced to in baseball that I realize haven't been sufficient explored or even explained as they relate to hockey. "Regression to the Mean" is one of those concepts, and I probably should have posted this chart before I spent a week talking about it.
Basically, regression to the mean is the idea that any quantity that's better or worse than average in one season is going to rebound to the average in the next season. It's sometimes known as the "Plexiglass Principle." If we look at every team and season since the 1967-68 expansion, NHL team winning percentage regresses 35% to the mean from season-to-season.
The dotted line on the chart is what would happen if winning percentage did not regress to the mean at all - that is, team winning percentage didn't change. Obviously, that's not the case. The solid line shows the best fit for what actually happened for all of those teams - if we know nothing else, we'd expect a .600 team one season to be a .565 team the next. That is, the team is .100 better than average, and it regresses 35% back towards the mean of .500, which is .565.
So in general, we know that good teams will tend to get worse and bad teams will tend to get better. And if we know absolutely nothing else about those teams, we slide them back 35% towards a .500 record.
0 recs |
4 comments
|
Email
Print
Comments
What I think is truly remarkable about this is that you don’t have enforced parity in year-to-year scheduling like you see in the NFL, where a 1st-place team gets a tougher schedule in the next season.
The draft system is also unlikely to play a major role here, as even a top draft pick usually doesn’t help a team very much in his rookie campaign.
More fun than a stick to the face!
On the Forecheck is SB Nation's blog covering the Nashville Predators.
by Dirk Hoag on Oct 21, 2009 10:57 AM EDT reply actions 0 recs
And yet it’s vastly better than it was during the Original Six era, where bad teams stayed bad…For 25 years…
by Hawerchuk on Oct 21, 2009 11:31 AM EDT up reply actions 0 recs
Well, that may be an explanation for enforced parity on small schedule/ small amount of team leagues?
Also:And I, for one, love posts about regression to the mean.
by Olivier on Oct 21, 2009 1:38 PM EDT up reply actions 0 recs
I think Regression could be very useful in SV% and SH% analysis.
by Moneypuck on Oct 21, 2009 7:02 PM EDT reply actions 0 recs
Comments For This Post Are Closed
Secondary Sidebar
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2010 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
