Rivals Total Points Rankings Losing Credibility

[GuardianFrog]

If you ask me ( I know no one did ) TCU, SMU, and Baylor's class are all very underrated by rivals this year. I don't think they give enough credit or "points" to quality Texas kids like Derrius Bell or Jonathan Jones. All three schools definitely recruited better than programs like Arkansas St, Ball St. and some of those other jokers that were put ahead of both of our schools.

[OC Mustang]

(edited b/c the site grabbed the post before I could check it for spelling/content)

I have a theory that I do not intend to check (somebody else can check it), but I derive it from something Lee wrote in his comments, and my own analysis experience with credit scoring in the 1990s. Rivals’ recruiting stars are inconsistent with Rivals’ ranking of recruits in their own position vis-à-vis other recruits. Either somebody fat-fingered information about recruits or their layered models (I see two models being used) conflict with one another in how they assess the same/similar attribute/s.

It could come from poor data quality. It could come from bias. It could come from the seemingly irregular nature of coaches to be consistent with their own assessments of players, much less with Rivals scoring attributes. That would indeed create outliers, which would, when the models are layered, accentuate the original inconsistency.

My money is that the star system is measuring something that is prone to bias, fat-fingering, and subjectivity, and any statistical aberration is exposed in the model that ranks classes among the universities. Lee pretty much admits this when he says that they are modeling discrete data on indiscrete attributes. He just says it in a way that rationalizes the results, rather than digs deeper into their star system and attempts to tweak it.

I understand why he would loathe un-wrapping and re-wrapping that box. He has no statistical stake in the ground to evaluate it since they have not been using the model "historically" and haven’t established a proved-up baseline to their rating of players, much less teams.

But then again, their relevance comes to a grinding halt when they discriminate past the obvious for precisely the reason that Lee suggests is at the root of the weird TCU results.
The system is useful for introducing recruits to a wider audience who otherwise would not really know how to view the recruit vis-à-vis other recruits in that position. If Rivals is compelled to take the next step and compare aggregated results vis-à-vis teams, then that is their call. But their audience should not expect bulletproof results. You will not get them when the root information is subjective, whether intentionally or unintentionally biased.

[GuardianFrog]

If you ask me ( I know no one did ) TCU, SMU, and Baylor's class are all very underrated by rivals this year. I don't think they give enough credit or "points" to quality Texas kids like Derrius Bell or Jonathan Jones. All three schools definitely recruited better than programs like Arkansas St, Ball St. and some of those other jokers that were put ahead of both of our schools.

[GuardianFrog]

If you ask me ( I know no one did ) TCU, SMU, and Baylor's class are all very underrated by rivals this year. I don't think they give enough credit or "points" to quality Texas kids like Derrius Bell or Jonathan Jones. All three schools definitely recruited better than programs like Arkansas St, Ball St. and some of those other jokers that were put ahead of both of our schools.