Wednesday, September 23, 2009

Doug Lesmeresis likes to be different

I often look at the AP poll and wonder how Team X is rated above Team Y, and when I was younger, I assumed that it was just a difference of opinion -- when you have 65 (or whatever) voters, there are obviously going to be significant discrepancies. But I've come to the realization that the actual voting philosophy differs from person to person, and I can't figure out why.

Isn't the purpose of the poll to rank the teams in order of how good you think they are? That was always my assumption. But for people like Doug Lesmeresis of the Cleveland Plain Dealer, that doesn't hold true, and his reasoning comes from a very literal interpretation of the ballot guidelines put out by The Associated Press:
Base your vote on performance, not reputation or preseason speculation.
Fair enough. But his ballot this week includes such oddities as Houston at No. 3, Cincinnati at No. 4, Florida at No. 5, Michigan at No. 10 and Ohio State at No. 22. Ummm ... yeah. I will give him some credit for thoroughly explaining his ballot each week, but this is where it falls apart:

First, my criteria.

1. Results
2. Results
3. Results
4. My ballot is a reflection of the past, not a prediction of the future.

As for the most common question from emailers questioning my sanity: Do I really think Team X would beat Team Y? Let's say, for example, do I really think Houston would beat Florida? Houston is No. 3 on my ballot, Florida is No. 5.

My answer: What I "think" is not what matters most. What I know is that Houston has more impressive on-field results this season than Florida.

Really? How does he know Houston has had more impressive on-field results this season? Beating Oklahoma State was nice, but how does he know that the Cowboys are that good? What makes that win better than, for example, Florida's win over Tennessee? The only way he can justify ranking Houston third is by arguing that Oklahoma State is a top-10 team, and the only way he can do that is by relying on preseason assumptions.

What about Michigan? Has UM performed better than Ohio State, whose only loss is the three-point heartbreaker to USC? Even as a Michigan fan, that'd be a difficult argument to make.

And what about Cincinnati? Big wins over Rutgers and Oregon State are nice, but neither of those teams is ranked. Florida is also 3-0, so what makes their two big wins and one relatively close win over Tennessee less impressive than Cincinnati's 3-0 start? His answer:
The Bearcats have two road wins against competitive BCS schools with 2-1 records - Rutgers and Oregon State. Road wins are always going to carry more weight than home wins. In between, Cincinnati came home and blew out Southeast Missouri State 70-3. I moved the Bearcats between No. 4 and No. 6, sliding behind Florida and Boise State, several times. But the fact that the best wins for Florida and Boise State were at home caused me to leave Cincinnati at No. 4.
So because Florida has beaten two cupcakes and one above-average team at home while Cincinnati has beaten two teams with winning records on the road, that means the Bearcats have performed better? I don't think that's necessarily a true statement. And if you apply that logic to a team like Boise State, who everyone agrees is very good but probably won't play a team all year that finishes in the top 25, you can see why this method is epically flawed. On Lesmeresis' ballot, each team can only be ranked as high as its schedule allows.

I guess this is where we differ. It's not difficult to tell which teams have performed the best -- just give me each team's record and strength of schedule, and the calculation is simple. That's basically what Lesmeresis is trying to replicate here, and I don't understand the purpose of having a poll to tell me this.

The purpose, in my opinion, is to tell me which teams you think are the best. I have a general idea going into the season (so does every AP voter -- that's why they have a preseason top 25), and each result -- more specifically, each performance -- influences that opinion. But losing a game or having a stretch against weak opponents doesn't necessarily make a team worse. I'm not going to assume that Miami is better than Florida just because Miami has beaten better teams through two or three games.

The same goes for head-to-head matchups. A close loss -- for example, Oklahoma's loss to BYU with Sam Bradford out for the second half -- wouldn't necessarily drop Oklahoma behind BYU on my ballot. I'm still confident that Oklahoma is a better team (especially after BYU got waxed by Florida State in Provo on Saturday), so I'm going to continue to rank the Sooners higher. I haven't heard many people arguing that Washington should be ranked ahead of USC, and for good reason -- no one can tell me that Washington is a better team. Think about this: If the Huskies' kicker had shanked that field goal, USC would still be in the top 10 and in the running for a national title, despite having nothing to do with the result of the kick. This is why you can't judge teams purely on record; you have to look at the in-game performance and combine it with your cumulative judgments from throughout the season.

And on a related note, why does a loss always cause a team drop? If LSU (ranked seventh) loses to Florida in two weeks on a last-second field goal, does that mean I should now assume that they're worse than seventh? NO!!! If they're that close in quality (depending on how each team plays) to the top-ranked team in the country, they're probably better than No. 7.

As you can see, my disputes are with the philosophy at large that leads to a lot of knee-jerk reactions, and Lesmeresis -- while he represents a different methodology -- is actually the prime example. USC beats Cal next week? Move 'em back up to No. 7, because they now have two excellent wins. Michigan crushes Indiana? Sorry, you're headed down 10 spots because Indiana is terrible and your strength of schedule now sucks.

I know it shouldn't matter since the AP poll doesn't actually count for anything, but if you're just gonna tell me who's beaten the best teams rather than giving me an opinion about who is the best team, don't bother voting. That's what the computers are for.

No comments: