clock menu more-arrow no yes mobile

Filed under:

On Polls and Such

Earlier this week I put out my preseason blogpoll ballot for the world to see. It got some criticism and I respect that and thank you all for it. I never really thought about what goes into making a poll until last year when I participated in the blogpoll for the first time.

Polls are complicated. Try putting one together sometime. Coming up with a top five is pretty easy. After that it's not hard coming up with a list of ten schools to plug into the 6-15 slots, but ranking them in order can be mind bending. Positions 16-20 usually aren't hard. You can find good teams that have a few losses and are just a few players away from being great to plug in there. Slots 21-25 are agonizing to rank. By this time you're swamped with a sea of 20 or so teams that aren't bad but just don't impress you that much. Picking the top five out of that group is a crap shoot. Luckily nobody usually cares about the teams ranked that low to notice who you put in there.

Brian is pretty selective in who he allows in the blogpoll, so I take my vote pretty seriously. I probably put more thought into it than the average blogpoll voter, and I'm sure I put more thought into it that the coaches and AP writers put in their ballots.

Although it will never happen I guess I'm striving for the perfect ballot. I'm an engineer, so I tend to put everything into equations when I pick them apart.

SEC > ACC. Therefore #2 in SEC > #2 in ACC

Loss to unranked team = -8 in rankings

Win over unranked team = 0

Win over top five team = +5 in rankings

Loss to top five team = -4 in rankings

If you struggle against an unranked team and you are a top 10 team then -2 spots

If you struggle against an unranked team and you are not a top 10 team then 0 spots

This seems to be the collective thought process of the polls, but does not necessarily reflect the though process of each individual voter.

There are a few different ways you can look at the teams when putting together a ballot. You can judge teams based on the theoretical head-to-head matchup method. Basically, who would win if Team A played Team B? The problem is you can't always answer that question because Team A does not always play Team B on the field. All you're asking yourself is who would be favored to win? We all know that upsets happen every week so this is far from a perfect way to judge the college football landscape. Every year we see the lineages where Team A beat Team B who beat Team C who beat Florida and that's why the St. Marys School for the Blind is the #1 team in the country. This makes a mockery of the polling system, and rightfully so.

While the theoretical head-to-head matchup method is vastly popular, you are making assumptions that the favorite team will win every time disregarding the potential for upsets. You are ranking each team based on your perception of their performance and stature. But upsets change our perception of each team. If Purdue beats Michigan we suddenly realize Purdue must be better than we thought and/or Michigan must not be as good as we thought. So another way to rank teams is based on the resume method.

The resume method takes each team's real performance into account and nothing more. It bases a teams position in the poll on who they have beaten, not who they could beat. Of course the resume method has some flaws. It leads to major swings in the ballot in the early going of the season. Last year after Tennessee knocked off Cal in week one they rightfully would have been bumped to the top of the poll since they had the most impressive win in the young season. They had the best resume. Cal would have been banished from the top 25 altogether. This sounds crazy based on the currect AP and Coaches' Poll practice, but it's the other extreme to the theoretical head-to-head method. It discounts preconceived notions and perceptions to judge a team solely on its performance on the field. By mid to late season one would expect the wild swings to slow down as the top 25 picture comes into focus.

The resume method is the one I intend to utilize this year in putting together my blogpoll ballots. So you're going to see some wild swings. I may get a few wack ballot and Mr. Manic Depressive awards, but that's ok. I'm prepared for it and the ridicule that comes with it.

Please give me some feedback in the comments. What method would you use to rank teams? What do you like/dislike about the polls? I'm willing to take suggestions and I would like my ballot to be a collective effort of the BSD community. Each week I plan to post my ballot for discussion before I submit it. So please pick them apart.