Using Two-Tiered Power Ratings in College Sports


So far we’ve looked at two-tiered power ratings in the NFL, the CFL, the WNBA and the NBA. All of these are professional leagues, where the numbers will work without any adjustments, since teams play fairly comparable schedules. One team may play a little bit tougher or softer of a schedule than another team, but it’s nothing like the disparity we see in college sports.

In football, a team in the Southeastern Conference is going to play tougher opponents than a team in the Mid-American Conference, for example, just as a college basketball team in the Big East is going to have a tougher schedule than a team in the Big Sky Conference, at least once we get into conference play. Some college basketball teams are notorious for scheduling a collection of patsies at the beginning of the season, while some small-conference college teams take the payouts for bringing their team on the road to be thumped by bigger schools, which also happens in football.

Since the dual-tiered power rating numbers are determined by points scored and points allowed, you can see the obvious problem, as a team in a smaller conference can have the same points for and points against as a team in a bigger conference.

We’ll look at a college football game between UAB and Boise State as an example and the same premise holds true for college basketball.

Based on numbers alone, we’d have the following ratings for each team:

UAB 30 – (-12)
Boise 37 – (-7)

To get a predicted final score, add one team’s offense to the opposing team’s defense, so UAB’s 30 plus Boise’s (-7) is 23 and Boise’s 37 plus UAB’s (-12) is 25. The main problem here is that we all know Boise State is two points better than UAB. The one thing we aren’t taking into consideration is Average Opponent Power Rating, or AOPR, which is often referred to as Strength of Schedule.

UAB accumulated its numbers playing teams who had an average power rating of 57, while Boise’s numbers were posted playing teams with a power rating of 67. So the next step is to divide the higher rating by the lower rating and 67/57 = 1.175. We need to increase Boise’s predicted points by 1.175 or 17.5%, while also decreasing UAB’s predicted points by the same 17.5%. Boise’s 25 *1.175 equals 29.375 and UAB’s 23 minus 1.175 equals 18.975. Now, we have Boise State favored by 10.4 points with a total of 48.35.

Now, we’ll look at one more, this time using Alabama playing UAB and we get:
UAB 30 – (-12)
Alabama 49 – (-15)

Using just numbers, we’d have Alabama winning 37-15, which is obviously too low. But since Alabama’s AOPR is 71, we need to multiply 1.245 to Alabama’s predicted score and subtract the same from UAB and our new projection is 46.065 to 11.325, or essentially Alabama by 35 points, which may seem a bit low, but most sets of power ratings have Alabama 37 to 39 points better and defense tends to translate a little better than offense.

Yes, it’s an extra step, but many times you can won’t have to go through the extra step when you have teams in the same conference, as Oklahoma and Texas have nearly identical strength of schedule ratings, so based on numbers alone, the projection is Oklahoma 46-36, which isn’t too far off the line or the total.

The two-tiered ratings may be most advantageous in college basketball, where teams have extremely pronounced AOPR figures, particularly in the early going of the season due to scheduling.

Those who are a little proficient with Excel can quickly set up a spread sheet where all you do is enter the scoring averages and the AOPR and the program does the rest, otherwise it’s a bit of a chore, particularly in college hoops.


Please enter your comment!
Please enter your name here