Friday, August 3, 2007

Recruiting Rankings: Do They Matter?

It's almost a timeless question: Do recruiting rankings really matter? Does highly-rated recruiting classes mean that you will win a lot of football games in the future? In more statistical terms, do recruiting rankings correlate with future team performance?

As said earlier, it's essentially a timeless questions, and just about every college football fan you'll ever run across will have an opinion on the subject one way or the other.

So I want to see how it stacks up empirically, and to that end I've done some research on the subject.

Basically this is what I've done: I've taken the team recruiting rankings from 2002-2004 from both Scout.com and Rivals, Inc, the two major recruiting services. Once compiling the data as to where each SEC team ranked in those three years (on a national basis, out of the one-hundred plus Division 1-A teams), I simply divided by three and came up with a three-year average recruiting finish for each SEC school. I took that three-year average recruiting finish, and ranked all twelve SEC schools on it, from one to twelve. From there, I stepped forward in time a bit, compiling the overall win-loss records (note overall, not conference) for each SEC team in the three-year stretch of 2004-2006. After that, I ranked the SEC teams by overall winning percentages, and then I started to compare all of the information that I compiled.

So here we go. The following is the three-year rankings for Scout.com, including the three-year average recruiting finish, and the conference rank:



And now here is the exact same data, except this time using Rivals, Inc. as the source:



Now that those two are out of the way, next let's move on to the three-year overall records in terms of winning percentage:



So that's basically all of the data in a spreadsheet format, for those who wanted it. But we can trim all of that down, simplify it greatly, and make things just much more easy to comprehend by collapsing a lot of that into one spreadsheet image, so here goes:



At bottom, Scout Rank is where that particular team finished among SEC teams in average recruiting rankings, and Rivals Rank is the exact same thing except it, of course, uses the Rivals, Inc. data. Win Rank is simply how that particular team finished among SEC teams in overall winning percentage the past three years.

So, all of that out of the way, what do we see?

It's quickly obvious that there is a very high correlation between good recruiting classes and future wins. If you run the numbers, there is a .8 correlation between Scout.com average recruiting finish rankings and future winning percentage rankings. And there's actually a little higher correlation for Rivals, Inc. as it crunches out at a robust .818.

What can you say? That's a pretty solid case. After all, a correlation in the neighborhood of .8 is very high, no two ways about it.

Moreover, just eyeball it, things are generally dead on. Take a look at the Scout.com average finish rankings. All told, eight of the twelve SEC teams finished in terms of winning percentage rank either exactly where they finished in Scout.com average recruit finish rankings, or within +/- one spot. The same goes for Rivals, Inc. as well. Again with Rivals, Inc. that same occurrence happened with eight teams. Beyond that, it's just elementary. The teams that did well in recruiting (LSU, Florida, Georgia, Auburn, and Tennessee) are the ones that generally did well on the field, and the ones who struggled in recruiting (Ole Miss, MSU, Kentucky, Vanderbilt, and Alabama) generally struggled on the field.

Again, what can you really say? Despite some people vehemently arguing to the contrary, the harsh statistical truth is that high recruiting rankings do correlate very highly with future wins, no two ways about it. Aside from a couple of exceptions (mostly Auburn), you just aren't going to finish low in terms of recruiting rankings and do particularly well in the SEC in the coming years.

I think you have to give the recruiting services a good deal of credit here. I personally believe they have improved by leaps and bounds over the over-inflated rankings of the late 1990's, and today they generally present very accurate overall projections. Sure, they miss on individual players (of course, nothing is perfect, and granted individual prospect ratings are very speculative), but as a whole, average team rankings over the course of several years seem to make a major impact later on in terms of how many games you win.

So, at the end of the day, to answer the question, "Do recruiting rankings matter?" Yes, they matter very much.

1 comment:

JeffAtlanta said...

When calculating the number of 4 & 5 star recruits, did you count all the ones that signed or a the ones that actually qualified? Some schools like UF and USC have been able to be picky about the blue chip athletes they go after while others (Dubose as an example) had to take what he could get.

Also, while more work, using only SEC games in the winning percentage calculations would probably be more accurate as well. This would eliminate inequalities in OOC scheduling.

Great work. Just stumbled onto your blog and liking what I see so far.