SB Nation Arguing about UH in the CFP

“Compare UH’s potential 2016 resume with 2014 Ohio State’s. The Buckeyes made it in at No. 4 after playing 10 bowl teams, only three of them in the final rankings (with one of those barely in, at No. 25), and losing to a team that’d gone 5-6 otherwise. Houston could play two or more final-ranked teams, eight or so bowl teams, and not lose to anybody. That’s very comparable. If UH whoops mid-majors like an Oklahoma-beating team should, there’d be no argument for letting in some one-loss team instead. This won’t be 8-0 Marshall remaining unranked after beating nobody. Keeping out this UH team would have people howling.”

Bill Connelly is a moron, and I question the accreditation of whatever school that taught him math, computer science, or whatever it is that makes him think he’s qualified to model college football. He currently has us at #48 and LSU at #3 in his computer rankings. No, really.

He has also chalked up our defense as “luck” for the past 3 years, and quadrupled-down on it this year projecting us for 7 or 8 wins.

1 Like

Connelly’s actually been pretty complimentary this year and has constantly mentioned that the Coogs are making his “stats” look bad.

Problem with his stats (or any stats) is that they rely on too many things that have too many variables. Teams don’t play similar schedules, constant churning of rosters, etc. Right now, his S&P+ is still using 90% of his preseason data…which is based on last year. Except, at least 1/4 of each team has had turnover. In the end, that doesn’t make any sense.

Kind of like ESPN’s FPI which relies too much on subjective recruiting rankings without taking into account transfers, drop-outs, or the fact that 2-stars can play like 5-stars and 5-stars can play like 2-stars.

Back to Connelly, his “turnover” issue with us is a normal statistical model relying on regression to mean. Except, that’s not how college football works. Take for example Virginia Tech back in the 90’s and 00’s under Beamer. For years, they would block punts and field goals and everyone would be amazed that they kept doing. Thing was, they put extra emphasis on it, practicing it more, and becoming proficient at it. That’ the thing, college football relies on coaching and recruiting, not just statistical regression models. Until someone can come up with models that can quantify coaching, these stats really don’t matter. Nice to look at, but useless in the long run.

2 Likes

The problem is no stats should be published until at least 10 or more games are played and pre-season rankings should have absolutely no value whatsoever.

The idea that recruiting rankings come into play in any system is appalling. NFL teams with millions of dollars in coaches and scouting with virtually unlimited time, and a very small target set, have trouble figuring out who will be successful. The idea that a couple guys in their basement can accurately rank HS recruits when there are literally hundreds of thousands of players to take into account is absurd on its face. What they do is rank guys that have offers from top teams high, medium teams medium and low teams low. That’s it… then they sell it for $10/month to suckers with more money than sense. If FPI is using recruiting rankings that explains a lot. They had UH in the 30’s after beating FSU.

2 Likes

Well said.

If he would just admit that particular variable in his model is flawed, I wouldn’t have a problem with him, but the doubling and tripling and now, quadrupling down is just sad.

He may be acting nice now, but I wouldn’t be surprised if he is just tired of hearing it from UH fans on social media.


I[quote=“TheMandell, post:4, topic:2875”]
What they do is rank guys that have offers from top teams high, medium teams medium and low teams low. That’s it… then they sell it for $10/month to suckers with more money than sense. If FPI is using recruiting rankings that explains a lot. They had UH in the 30’s after beating FSU.
[/quote]

Interesting- On ESPN they ranked the teams after we beat OU in FPI which has to do with Vegas- The announcer had OU and LSU ranked 5 and 6 in FPI and Houston 18- that after we beat OU down. The announcer went on to say that if we Wisconsin and Houston played each of their respective teams again that OU and LSU would have the same Vegas odds as the day they both lost.

©Copyright 2017 Coogfans.com