from Amazon
Wednesday, October 08, 2008
The future of your team
winClass n next3 regr3 over500 reg500
0.378 26 0.458 0.451 19% 13%
0.429 157 0.478 0.472 35% 29%
0.476 265 0.489 0.490 45% 43%
0.522 311 0.507 0.509 57% 57%
0.571 144 0.530 0.528 70% 71%
0.617 39 0.572 0.547 95% 85%
From 1969-1971, the Expos had a .404 record. I put them in the “.400-.450” win class. (That’s an average of .429 for all the teams in that class, which is the second line in the chart, under the “winClass” column.) From 1972-1974, they had a .476 win%. That’s a close match to similar teams (actually, .478, under the “next3’ column). Teams in their win class (the .400-.450 win class) ended up with a 3-yr win% of over .500 35% of the time (that’s the “over500” column).
If we regress their win percentage 60% toward the mean, we end up with the reg3 column (that’s win% times .4 plus .3). As we can see, a pretty strong match, except for the really, really good teams. If we take their win%, times 3, minus 1, you get the reg500 column. Again, a decent match.
So, the 2006-08 Pirates, with a win% of .416, is expected from 2009-11 to have a .466 record, with a 25% chance of having at least a .500 record over those three years.
That is a nice toy. Again, it is one of those things that varies so much from team to team (even though it works nicely in the aggregate) that unless you knew absolutely nothing about the team, it is not very useful for anything.
For example, if we took the last 3 years’ record for TB and applied this toy, we would come up with a pretty conservative expectation for the Rays, because they are quite a different team now then they were last year and the year before that.
The things we know or could fairly easily find out about a team trump a model like that pretty soundly. Were they lucky, unlucky, or right on the money in terms of their record over the last 3 years? What is the projected payroll over the next 3 years? Did they change ownership or GM’s? Did the owner or GM indicate that they were going to change their philosophy in terms of spending money? Are they in a rebuilding phase or not? Did they change their personnel substantially over the last 3 years? Are they losing or gaining significant talent after this year?
Again, unless someone asks you what you would predict for some unnamed team that had X record over the last 3 years, I don’t find much, if any use, for this toy.
Then again, that is what toys are for - to play around with.