Mostly just for fun, but also to justify using recruiting rankings as a predictor, I took at look at just how good they are at predicting. This is definitely not a very original thing to do. BHGP just had an excellent player development post that compared recruiting rankings to NFL draftee. More in-line of what I'm looking for though, is this study that's from the Journal of Quantitative Analysis in Sports. The study concluded that "up to 45% of the variances in the end-of-season ratings and total wins" could be predicted by the recruiting rankings. That sounds pretty nice, but it's from a few years ago and I still wanted to run the numbers myself.
I took the average Points from Scout.com's team rankings from 2006-2010 (5 years for the obvious reasons) and compared it to the winning percentage for the 2010 season and to the final Sagarin rankings. The results were actually quite different than the JoQAiS study. I came up with a correlation coefficient of 0.409. That translates into just 16% of the variance in win percentage was predicted by the recruiting rankings. Against Sagarin the results were better. The correlation coefficient was 0.626 which is more like a 40% predictiveness.
So yes, depending on how you define a team's success (wins or ranking), the recruiting rankings do provide some predicting power. The win percentage one was not great, but considering how many factors go into whether a game is won or loss, I'd say it's good enough to take into consideration.
For 2011
Moving forward I wanted to look at the 5-year recruiting ranking average for the B1G for this year. Since the data was there I went ahead and did it for the entire FBS (minus Western Kentucky since they weren't FBS 5 years ago). I again used the Scout.com Points as my basis.Here's the B1G:
Team | 2011 | 2010 | 2009 | 2008 | 2007 | Average |
---|---|---|---|---|---|---|
Ohio State | 4403 | 3032 | 4799 | 4271 | 2852 | 3871.4 |
Michigan | 2637 | 3593 | 3379 | 4007 | 3146 | 3352.4 |
Penn State | 2108 | 3762 | 3520 | 2126 | 2703 | 2843.8 |
Nebraska | 2788 | 2505 | 2360 | 3012 | 2622 | 2657.4 |
Illinois | 2043 | 1423 | 2263 | 3216 | 2638 | 2316.6 |
Wisconsin | 1978 | 2334 | 1853 | 2629 | 1581 | 2075 |
Michigan State | 2484 | 2424 | 2234 | 1594 | 1321 | 2011.4 |
Iowa | 2663 | 1857 | 1052 | 2047 | 1869 | 1897.6 |
Minnesota | 1239 | 1122 | 1946 | 2617 | 1001 | 1585 |
Purdue | 997 | 1737 | 1314 | 1931 | 871 | 1370 |
Northwestern | 1452 | 1364 | 1152 | 1282 | 1069 | 1263.8 |
Indiana | 1483 | 1343 | 1237 | 1418 | 828 | 1261.8 |
I used the line of best fit from the 2010 analysis and predicted the record for each B1G team based solely on their recruiting rankings.
- Illinois: 7 - 5
- Indiana: 6 - 6
- Iowa: 7 - 5
- Michigan: 8 - 4
- Michigan State: 7 - 5
- Minnesota: 6 - 6
- Nebraska: 8 - 4
- Northwestern: 6 - 6
- Ohio State: 9 - 3
- Penn State: 8 - 4
- Purdue: 6 - 6
- Wisconsin: 7 - 5
Most of those win totals are within a game or two of what I would expect. They seem to be a little bit condensed though with the best team at 9-3 and the worst at 6-6. I'd expect that there will be more seperation between the top and the bottom of the the B1G.