Projecting Talent into Win Totals

Paul wrote a great piece this week on predicting the SEC finish solely based on talent alone. It’s a great article, and you should totally go read it if you haven’t. Of course, in the comments, a few people asked for a needlessly more complex version which would only have limited increased utility. Boys and girls, that’s what the Poseur Alarm sounds like.

The key thing is that I’ve already created just such a projection system that does account for the actual recruiting rankings and not just the average star rating. I have a database of class rankings for SEC teams going back this entire decade, and I built a projection system last system to determine LSU’s level of underachievement.

There’s a lot of math, as I told you it’s needlessly complex, but you can read all the details here. For those of you who don’t want to dig through the math, let’s just hit the high points here, and then we’ll all get to the projections together.

Firstly, I take all the 247sports composite rankings for each team. I don’t use the ordinal rank or count the stars, I use the actual point score. So, for last year’s class, LSU gets 285.85 points, not the #7 national ranking.

Then, I weight each class for their projected impact. So we take each class point score and give a multiplier: 0.5x for freshmen, 2x for sophomores, 3x for juniors, and 2.5 seniors. This accounts for expected attrition, but not for redshirts. This is based on what the talent on hand SHOULD be, not precisely what it is.

This gives each team a total talent score for the season. We determine the average talent level and compare a team’s talent score to the average to get a projected winning percentage. The average SEC team this year has a score of 1968.85.

Whew. Even the abbreviated version is long. Here’s the projection:

My system likes Georgia, just not as much as Paul. I have LSU with a more talented roster and the only team within a win over Bama, and just barely at that. Auburn is not in a dead heat with LSU, instead, they are about even with Georgia.

The real surprise is how well Tennessee does. At 5.2 wins, it far outpaces the media front-runner Florida and is less than a half win behind the team we tabbed as the most talented in the East.

Arkansas and Mississippi St. are the clear basement in the West, lagging a win and half behind Ole Miss. The West cellar is still higher above ground than the crowded cellar of the East, which has three teams ranked behind the West’s lowest ranked team.

But talent isn’t everything, right? Some teams do better with less talent and have been able to make chicken salad out of their raw materials. And some teams can never live up to their lofty recruiting ranks, right? Well, because we have the prior performances and old recruiting ranks, we can see how teams have over or under performed the past five seasons.

We can add or subtract the wins that a team has missed its projection by one average, over the past five years. Let’s see where that gets us:

Georgia moves up to second, in line with Paul’s picks. Tennessee drops off the map due to their track record of failing to live up to their recruiting projections. On the other hand, Mississippi St. rockets up the rankings for its sustained success in the face of lower recruiting ranks.

Yes, Bama has a negative adjustment. They are usually projected to win over seven games, and like clockwork, they go 7-1 nearly every season. It added up over time.

So there you go everybody. Bama wins again, and LSU wins 5 SEC games. Again. It’s science.