The Case for Resume Ranking: A Diatribe Against BlogPoll Favoritism

The initial 2007 BlogPoll ballots are being submitted even as we speak. The BlogPoll was invented, and is administered, by MGoBlog's Brian Cook, who has turned it into one of the most respected institutions in the intercollegiate athletics blogosphere.

Consequently, we who have the privilege of voting in the BlogPoll take that responsibility seriously---not because we have any illusions about its impact, but because we believe what is worth doing is worth doing well---and those voters who cast their ballots carelessly (such as, for instance, someone who drops a team in the standings because he believes that team lost a game it actually won) are subject to being ridiculed by their peers.

I therefore found it heartening when I read this:

That sentiment comes from Mike of Black Shoe Diaries, whose life has included recent positive developments that no doubt help him to keep college football in its proper perspective.

I'm just getting started, but there will be an Edmund Burke reference before I'm done. Also, there will be Calvin Coolidge and Jennifer Love Hewitt references, as well, so you'll want to stick around for those, too.

Writes Mike:

Because of this, Mike concludes sensibly that he will be using the resume ranking approach this season, to which I also subscribe. His point about teams being favored, however, is one that warrants closer consideration and ought to give us pause.

When a college football fan uses the term "favored," of course, he is not referring to the expected outcome of an obvious mismatch; Florida is not "favored" to beat Western Kentucky because there is no realistic scenario under which the Gators conceivably could lose their season opener against the Hilltoppers.

However, the over/under on the number of times the poor sap wearing this suit in Gainesville, Fla., in an open-air stadium for three and a half hours of daylight on Labor Day weekend passes slap out from the heat is four.

Rather, "favored" is a gambling term. It refers to the betting line set by the Las Vegas oddsmakers regarding the outcome and margin of victory of an upcoming contest. I note this neither as an endorsement nor a condemnation of wagering, but only to call attention to the source and the underlying motivation behind the selection of favorites and underdogs.

Oddsmakers design betting lines in order to generate significant wagering on both sides of the line, so as to ensure that the bookies clean up, irrespective of which team wins or by how much. Betting lines move over the course of the week leading up to a game because the wagering is too one-sided and the bookmakers want the betting to be more evenly distributed.

These facts matter because they disabuse us of the misconception that a betting line has predictive value, or even that it represents a prediction. If the most knowledgeable oddsmaker in Las Vegas announces that Team A is favored by three points over Team B, he is not predicting that Team A will beat Team B by a field goal on Saturday. (This is more clearly illustrated by the frequency with which teams are favored by, say, "two and a half points." Such a line is designed to avoid a "push," but, obviously, it could not represent the actual margin of victory of the contest, unless you're using the metric system or something.)

Darn you, N.F.L. Europe, with your metric four-and-a-half-point field goals!

When a bookie says Team A is favored by three points, he isn't saying Team A will win by three points; he's saying that he thinks roughly half of his gamblers will think Team A will win by more than three points and the other half will think Team A will win by fewer than three points or will lose outright.

If the oddsmaker has estimated the inclinations of his bettors incorrectly, he will adjust the line accordingly. (As USA Today's rather simplistic odds primer puts it, with emphasis added: "Player injuries and wagers are among the top influences on line movement.") In other words, a bookie's goal is to have 50 per cent of the folks who put their money where their mouths are disagree with him.

Mike is right that picking favorites is no way to rank teams because upsets occur, but there is a deeper problem related to the very concept of the "favorite." In any contest expected to be competitive---in other words, in any game likely to have a meaningful effect on the rankings---a narrow favorite is, by definition, the team the oddsmaker expected would be deemed less likely to win by that much (or even win at all) by half of the folks who put money on the game. That is hardly a ringing endorsement of a favorite's expectation of victory, especially when we consider the high rate of inaccuracy even among the best bookies and shrewdest gamblers.

Point spreads neither have nor are intended to have predictive value. Much like Stewart Mandel's assessment of the national prominence of particular teams, betting lines are based upon oddsmakers' perceptions of other people's perceptions. Even though bookies, unlike Stewart Mandel, are not fools, neither are they savants intent on influencing pollsters' evaluations of particular teams. They just want there to be enough bettors both ways for the house to be assured of victory.

The house always wins. Unless you get George Clooney involved in the heist, of course.

Whether you put money on college football games is your business. (For the record, I don't, both because I am lousy at predicting the outcomes of college football games and because I am familiar with Paragraph 163(G) of the Social Principles contained in Part IV of The Book of Discipline of the United Methodist Church.)

When casting a college football poll ballot, though, a voter (be he a coach, a journalist, or a weblogger) should give consideration to several criteria . . . but the point spread ain't one of 'em. The favorite in a competitive contest is the team that the guy setting the line hopes half the bettors won't pick.

As an astute commenter at Mike's site noted, pollsters are forced to "use the theoretical head to head matchup"---as Brian puts it, "[a]t all times it should be an approximate ranking [of] who would beat who[m] on a neutral field this year"---until enough games have been played to give anything like enough data to draw reasonable conclusions. Mike's commenter, PSU Mudder, continues:

That is a succinct way of stating precisely the correct position. (Not having been blessed with concision, I regard it as a strength in others.)

Suffice it to say that no one ever lost by betting that I could be persuaded to utter at least three words.

At this point, it's all guesswork, so prospective assessments of the my-dad-could-beat-up-your-dad-and-the-Partridge-Family-would-win-a-fight-with-the-Brady-Bunch playground variety are all we have upon which to rely. This leaves folks like me in the position of sending e-mails to in-laws, old pals, and Sunday Morning Quarterback asking if I am completely crazy, thereby eliciting responses such as this one, which arrived in my in-box from a longtime friend whom I have known since he was nine at 9:45 on Wednesday evening accompanied by a mild adult language advisory:

That analysis could prove thoroughly prescient or absolutely baseless, but no one yet knows which for sure. Once we get into the meat of the season, though, facile evaluations of underdogs and favorites will have seen their utility dissolve into futility. When we arrive at that juncture, Mike is right that we should cast our ballots based not upon what we think might happen next Saturday, but on what we know happened on the preceding Saturdays.

Call it the Jennifer Love Hewitt theory of poll voting: "I Know What You Did This Autumn."

Until enough facts are known for reasonable conclusions to be drawn, though, those denizens of Bulldog Nation and visitors from other schools who form the Dawg Sports reader community may feel free to treat me as your BlogPoll representative. Some of you already have shared your views upon my ballot, which I appreciate and encourage, with the understanding that I subscribe to the view of representation enunciated by Edmund Burke 194 years to the day before I was born (and 233 years to the day before this year's Georgia-Troy game):

Such are the weighty considerations pressing upon the minds of those BlogPoll voters sufficiently refined in their thinking to recognize the linkage between college football and self-government.

Is it football season yet?

Go 'Dawgs!