A New Methodology
A couple weeks ago, I took a look at a new recruiting ranking system called Rank By Offers (RBO). The ratings had been making the rounds because MSU's 2016 class has been insane, yada yada yada, I wrote a post critiquing the system.
The amazing thing is that RBO read my critique and agreed with (at least some) of what I wrote. This caused them to change their team weightings. RBO's new methodology, per their site:
The relative value of each program's offers are determined using players' commitments and the offers they rejected. Using this approach, it is assumed that the offers players accepted are more valuable than those they passed up. In essence, program offers are ranked by the players themselves.
This is awesome. Using ELO ratings, RBO has used their database of scholarship offers and commitments and used these to rank how valuable offers are from various schools. This methodology is brilliant, because it accounts for all of the factors a school may or may not have in its favor. Some of those factors, like the use of bag men, would be totally impossible to account for otherwise. This, by itself, is an accomplishment.
Further, it passes the sniff test. The top ten include Alabama, Clemson, LSU, Texas A&M, Texas, Florida State, Auburn, Georgia, Florida, and Ole Miss. Essentially, teams with both historical and recent success, geographic advantages, and deep-pocketed boosters.
Michigan State ranks sixth in the Big Ten and 42nd nationally in terms of offer value. I think that will change once RBO updates their system on June 21st, considering the impressive offer lists for Auston Robertson, Josh King, and Naquan Jones. The Big Ten rankings:
The big surprises to me are Maryland and Northwestern. My theories here are that geography and academic prestige matter with these offers to some extent, which would explain the performance of those schools relative to the rest of the Big Ten. I also believe the value of a Minnesota offer is rising, given Minnesota's 2016 class is ranked 23rd nationally per 24/7 composite at the moment.
Aggregation Still an Issue
This new ratings system also begins to deal with the offer aggregation issue I mentioned previously. From the raw values above, you can see that a player with offers from Indiana (offer value = 39.53) and Purdue (offer value = 30.49) gets less value (total = 70.02) than a player with an offer just from Ohio State (offer value = 83.91). Still, though, the value of a recruit is still too closely aligned with the number of offers they've received:
Offers Versus Offer Points in RBO
Previously, the correlation coefficient was 0.99, so this is a step forward. The overall player rankings (and therefore, the team rankings) are too dependent on the aggregate number of offers a recruit receives. A recruit getting offers from Illinois, Iowa, Indiana and Purdue (total value = 149.40) shouldn't be more valuable than a recruit getting offers from Michigan and Michigan State (total value = 137.51).
I'll be using RBO's team weightings to help contextualize offer values. The value of a Tennessee offer, for instance, was greater than I'd expected. But the aggregation issue is still too great for me to adopt RBO as my ranking system of choice; despite the flaws of the various recruiting outlets, the meta-analysis of the 24/7 composite is still probably more accurate overall. If RBO could do something like an average value of the top 5 offers, I think that might have real value.