PDA

View Full Version : 3 Game Change Theorem


Winger
08-07-2009, 08:07 PM
For vegasinsider.com is a story that notes the significance of a 3 game change from season to season as a predictor for the following season.

Basically it shows that if there is a 3 game or more improvement from one season to the next, there is an 80% chance that that team will have a worse record for the upcoming season.

The contrary also holds (but stronger at 85%).

That said, looking at the Pac 10 team records from last season:

USC 12-1
Oregon 10-3
Oregon State 9-4
California 9-4
Arizona 8-5
Stanford 5-7
Arizona State 5-7
UCLA 4-8
Washington State 2-11
Washington 0-12

And from the season before:

USC 11-2
Arizona State 10-3
Oregon State 9-4
Oregon 9-4
UCLA 6-7
Arizona 5-7
California 7-6
Washington State 5-7
Stanford 4-8
Washington 4-9

You get a roughly 80% chance that:

ASU will better than 5-7
Washington will be better than 0-12
Washington State will be better than 2-11

and ....

Arizona will be worse than 8-5

scumdevils86
08-07-2009, 08:15 PM
Well dang.

Winger
08-07-2009, 08:30 PM
Applying the 3 Game Theorem to the Coach's Top 25:

There is an 80% chance that each team with a "W" after their name will have a worse record this season.

1. Florida 13-1 W
2. Texas 12-1
3. Oklahoma 12-2
4. Southern California 12-1
5. Alabama 12-2 W
6. Ohio State 10-3
7. Virginia Tech 10-4
8. Penn State 11-2
9. LSU 8-5 B
10. Mississippi 9-4 W
11. Oklahoma State 9-4
12. California 9-4
13. Georgia 10-3
14. Oregon 10-3
15. Georgia Tech 9-4
16. Boise State 12-1
17. TCU 11-2 W
18. Utah 13-0 W
19. Florida State 9-4 W
20. North Carolina 8-5 W
21. Iowa 9-4 W
22. Nebraska 9-4 W
23. Notre Dame 7-6 W
24. Brigham Young 10-3
25. Oregon State 9-4

While LSU is only team to fit the criteria to expect an 85% chance of a better record for this upcoming season.

Talk about hindisight bias, no wonder the preseason polls are so inaccurate.

popque
08-09-2009, 10:52 PM
A team that goes +3 from year 1 to year 2 in a 12-game season has, on average, much more room to fall in year 3 than to rise.

Lots of the +3 teams would be 8-4, 9-3, 10-2, 11-1 in year 2, for example.

So...that statistic is fairly meaningless.

TheCat
08-10-2009, 05:32 PM
right....look at utah.....they will be worse.

bcrisp
08-10-2009, 07:35 PM
Does that theory factor in bowl games? If you remove the bowl we only improved by two games.

Winger
08-10-2009, 08:34 PM
Does that theory factor in bowl games? If you remove the bowl we only improved by two games.

I wondered the exact same thing ... they didn't say.

As to the, "well this is obvious" side of things: it is, but it is still very helpful if one is looking for a new angle on wagering, particularly early in the season (like going against some of those teams with a red "W" above".