« Big Number Country - Big Ten Preview, Week 8 | Main | MRISports' AP Ballot - October 24 »

MRI Football 2010 - Rankings Week 6

October 27, 2010  |  By Benjamin Miraski

Is it time to bash the computers? They are the most hated component of the BCS after all.

Forget that the human voters often make more mistakes than their microchipped counterparts. Forget it because you won’t ever hear one of those talking heads on television criticize a voter.

Forget it because there is no way that the coaches have any idea what happens across the country every week, yet are expected to form an opinion that we have learned from years past is often passed on to someone else in the athletic office.

Forget it because the qualifications for voting in the Harris poll are that your mom’s cousin once went to school with someone who knows someone and mentioned you at a party.

But we all love to hate the computers, which sounds like sacrilege when it appears alongside the weekly MRI standings.

Admittedly, the MRI isn’t the most accurate tool out there. Usually it predicts about 70 percent of the games correctly.

But it seems to be correct more often than what we are seeing out of the BCS computers.

The New York Times took a preemptive shot at the computers prior to the initial BCS rankings. And they may have a point, frankly admitted to by one of the computer rankers themselves.

“You’re asked to rank teams that don’t play each other, that don’t play long seasons, and you can’t include margin of victory?” said Massey, who provides a “better version” on his Web site, masseyratings.com. “It’s a very challenging problem from a data-analysis standpoint. It does require sacrificing a bit of accuracy. It’s not the best way to do it.”

And lacking accuracy is something all too clear two weeks into the Blasted College System.

How else do you explain Oregon have almost 99 percent of the highest score in both human polls yet they are unable to sniff the No. 1 spot in the overall BCS?

How can two teams that don’t even get 10 percent of the first place votes in the human polls lead the BCS based on the computer rankings (Oklahoma last week and Auburn currently)?

It doesn’t make sense.

Maybe it is that missing component, the margin of victory. After the first few seasons of the BCS, the computers were asked to remove it from their formula.

The was college football after all. The players and coaches were supposed to be civilized. They aren’t supposed to run up the score to improve their rating on a computer.

Those computers that couldn’t make the cut were cut themselves, and the nation was left with just six boxes left standing.

But when one of their own admits that the rankings are flawed without some margin of victory component, shouldn’t someone take notice?

The MRI uses a weighted margin of victory to help rank the teams. You can blow out a bad team, but it won’t necessarily matter as much as if you win by 10 against a good team.

And there is a cap of 35 points on the raw margin before weighting. After a five touchdown lead, even the humans have stopped paying attention, so the computers should do the same.

With the margin of victory added in, perhaps the BCS rankings would resemble reality, where Auburn has had to squeak out almost all of its wins.

Or Boise State and TCU would have a little more love after they trample their admittedly softer schedules. A simple weighting would tone that down, but it would show how much better they are than their schedule alone suggests.

Or maybe Oregon would manage better than an average ranking of eight in the computers. They couldn’t process the Ducks’ 47 point victory over Pac-10 conference foe UCLA as anything more than just a simple check mark in the win column.

The MRI gave Oregon an additional 9.3 points for the dominating win, on top of any other gains from other components, almost 11 percent of its total MRI score.

So what is holding back Oregon (currently 4th) from the top spot in the MRI?

Part of Oregon’s issue may be their strength of schedule to this point. According to the MRI, the Ducks have played the 117th ranked schedule in the FBS, counting among its vanquished an FCS team, two-win Tennessee, winless New Mexico and one-win Washington State (whose one victory was over an FCS school).

Meanwhile Auburn has played the sixth hardest slate in the country. That will change in two weeks when the Tigers face FCS Chattanooga.

If Auburn can hold it together against Mississippi and are still undefeated at that time, it will be interesting to see how the computers react to that scheduling blip. Chances are the BCS microchips will punish them for it (as will the MRI to a degree).

For those out there wondering about the supposedly cream puff schedules of Boise State and TCU, they come in at 39 for the Broncos and 66 for the Horned Frogs.

Plus Boise will not play an FCS school this season and has two good games remaining on their schedule in Hawaii and Nevada. TCU (which did play an FCS opponent) will still get bumps from Utah and San Diego State, although they finish with a cake walk over New Mexico to bring them back to earth.

There is still quite a bit of football left, but the computers have shown one thing so far. Without all the data, you can make any assertion you want.

It just doesn’t make you correct.

Check out all of the (flawed?) rankings in week six of the football MRI.

Posted October 27, 2010 1:00 PM