Esteemed Blog Poll voters have been asked to submit an early ballot by today, the better for scrutiny, naysaying and revision in the week leading up to the vote for the official preseason poll, to be released next Wednesday. This doesn’t "count," in other words, and I’m still working on compiling numbers for my methodically-generated, binding poll, discussed further below. But in the spirit of the enterprise, this is my hasty, non-methodical, "eyeball test" ballot:
Waiting: South Carolina, Boston College, BYU, Oklahoma State, Clemson, Arkansas, Tennessee, Texas Tech, Texas A&M, Georgia Tech
- - -
Now, as I mentioned, the point of this initial poll is to generate discussion and feedback to take under consideration for next week’s vote. But I have no intention of looking at this order again, regardless of what anyone has to say about it, because it’s going to be replaced by a meticulous version that’s still in the works. Instead, I’ll solicit feedback and criticism re: that methodology.
I think the goal of a preseason poll, rather than serving as a "power poll," should try to predict the final poll in January as closely as possible. Since that poll in my case will be based on comparing year-end resumes, the purpose is to come up with the best guess of what those respective bodies of work will look like at the close of the season, and that means taking schedule seriously into account. Anyone who’s read my Big Ten and Pac Ten picks the last two weeks has encountered the first step of this process, wherein I assign a number between 0-9 to each game based on the following scale:
8: Comfortable win; chance of competitive game
7: Competitive win; minor upset threat
6: Close win; major upset threat
5: Toss-up win (likely, but not specifically, awarded to the home team)
4: Toss-up loss
3: Close loss; major upset threat
2: Competitive loss; minor upset threat
1: Big loss; chance of competitive game
0: Blowout loss; no conceivable victory
|5||at Arizona State|
Now, within the conference format, because the schedule strength is the same, the total is sufficient to generate an order of finish by itself. Once non-conference games are added, though, differences in out-of-conference scheduling and between conferences (a 6.5 total in the SEC is worth vastly more than a 6.5 in the MAC) necessitate some kind of control for strength of schedule. For that, I generated a fairly complicated "Win Value Rating" for each team based on about ten criteria ranging from average wins under the current coach to Phil Steele’s back-of-the-book experience ratings, filtered through a lot of arbitrary division to weight the numbers evenly and generate a manageable total at the end. Way way more inside baseball than it's worth to get really specific, but that number will be divided again into a small fraction – to make sure it doesn’t overwhelm actual winning and losing – and applied to each opponent for every team, then averaged and added to the game-by-game total.
I apologize to anyone who read all that. But the point of the Blog Poll is transparency: you deserve to know. Hopefully it will become clearer when I roll out whatever my low-fi widget spits out as a top 25 next week. All I know is that USC will still be number one.
Questions? Suggestions (besides "get a life")?