Sunday, November 13, 2005

Poll Smokin'

Not to beat a dead horse, but HRB has come up with a way to solve (or at least improve) the gross deficiencies in the BCS computer rankings, while keeping the suits happy. We all know why Margin of Victory was eliminated from computer polling. Like it or not, there is at least a case to be made for not rewarding teams for running up the score. So instead of MOV, why don't we add Margin of Defeat to the poll? Hear me out now. Here's the upside, and a case study to go along with it:

Let's look at Ohio State and Colorado this year. Prior to this week's games, both teams had identical records (7-2). Let's look at their resumes coming into this week.

Record
7-2
7-2
Losses
Texas (9-0)
@Penn State (9-1)
@Texas (9-0)
@Miami(7-1)
Opponents' Losses
Texas: None
Penn State: @Michigan (6-3)
Texas: None
Miami: @FSU (7-2)
Human Poll Average
10
21.5
Computer Poll Average
7
11


To anyone who follows college football, it was very clear that Ohio State was the superior team. However, to a computer, the two resumes were completely indistinguishable. Both included two losses to Top 5 teams, and victories over the rest of their in-conference foes. In fact, CU's resume seems a bit more appealing from opponents' opponents records as well as the fact that they dropped both of their games on the road, whereas OSU's loss to Texas came even with the benefit of home field.

So what differentiates these two teams on a quantifiable level? Margin of Defeat.

First Loss Margin
3 Points
20 Points
Second Loss Margin
7 Points
25 Points
Total Margin
10 Points
45 Points


You see, more important than just asking "how impressive are the teams that beat you?", the real question that needs to be asked when ranking teams is, "how well do you play against top-level competition?" In Colorado's case, they've shown the ability to handle the lesser teams of the Big 12 quite impressively, but then be uncompetitive against top-level teams. No shame in that, but it means they're a mediocre team, and deserve a mediocre ranking (i.e., not #11). Ohio State, on the other hand, has made a game of every matchup they've had, and as such belong in the upper tier of the rankings, despite their two losses.
Note that this does not encourage running up the score, as there is no benefit to the team that does so. The only pressure it creates in the game is for teams to avoid being blown out, which, I would have to imagine, is part of every team's gameplan anyway. In the BCS picture, Colorado's (very predictable due to their proven mediocrity) loss to Iowa State shook everything up, since they were practically a top-10 team in the computer rankings. This shakeup was completely unnecessary and could have been avoided if Colorado had been ranked more realistically beforehand. And, lo and behold, it can be done without adding that dreaded Margin of Victory to the equation.

For those of you who haven't fallen asleep yet, thank you for reading this tirade. I for one refuse to read blog entries that don't have pictures, so all of you who got this far can take pride in having a much much longer attention span than I. Your patience shall not go unrewarded. Here's a picture of Otto the Orange for you, my favorite mascot in sports.



1 Comments:

At 2:57 PM, Blogger mayday said...

This is a completely wise and brilliant insight. However, I suspect it will not be adopted, because in my opinion, the BCS took MOV out not to pursue greater statistical accuracy, but to keep a one-loss team from getting ranked #2 ahead of an undefeated team that was #2 in the human polls. In other words, just to cover their own asses and dampen criticism.

Margin of defeat is more fair, but it doesn't serve the BCS any better than does margin of victory.

 

Post a Comment

Links to this post:

Create a Link

<< Home