The Analytics Suggest Tamping Down the Hype on Houston

Article

This seems to be a recurring theme as all of the analytics say that we shouldn’t be good. Most of this is purely based on our recruiting rankings and our turnover luck, 1 of which is completely subjective (and doesn’t account for transfers like Catalon, Samples, and Allen) and the other has seen Houston have sustained success forcing turnovers.

Just not a strong believer in the analytics that our out there because too much of it is based on subjective data that comes from recruiting rankings. Ah well, just have to prove them all wrong.

I think trying to predict how 17 and 18 year olds will grow and develop in college is a fool’s errand.

Recruiting rankings are appealing to the most popular teams with the largest fan bases (P5). They are not concerned with scouting good football players. Recruiting websites are trying to obtain the most subscribers possible.

I genuinely think it is impossible to predict how a college football season is going to play out. You can grade the returning player’s performance the year before, but you have no way of knowing how a new starter is going to perform. We don’t know what Joshua Jones is going to do at LT. We don’t know how Garrett Davis will handle full time snaps at safety.

Just have to wait and see.

The fact that recruiting sites have an impact on preseason rankings is mind boggling. But that brings me to the real problem.

Why are we ranking anything? Why not just have the conference champions play a tournament at the end of the year? :rage:

3 Likes

Too much mention of Bill Connelly and his pseudo analysis. He’s been claiming our defensive success (particularly our ability to cause turnovers) is “luck” for 3 years now.

1 Like

The way I look at college football recruiting is like the MLB draft. Specially under the helm with coach Herman. UH may not get a lot of 4 and 5 stars like the handful of schools. But they do get players who have UPSIDE and not finish products. Eventually, the goal is to get the best high end 3 stars who the coaching staff can develop into a 4 or 5 star type player bout the time that player becomes a junior in college.

But getting a few 4 star and maybe a 5 star to a class is never bad.

They kept calling it luck, but EVERY single person responsible was returning every year. This year I expect to regress with Trevon and Adrian gone.
This is where analytics get it wrong in football. They see teams regress towards the mean constantly on a stat like this but they fail to take into account the individuals making it happen and, in particular, the the synergies of a specific group.
Another thing that bugs me is that they use ratings that are proven to be bonkers.
S&P+ is a piece of crap.
Even after beating FSU it had us ranked 44. It had an 8-5 Louisville team we beat on the road at 28 and Navy team we took to the wood shed at 20 and FSU at 7. We BEAT all those teams.
If you look at FEI it had us at 10.

Ed Feng is just as bad.

Over the last 4 years, he’s had two G5 teams finish in the Top 30. He had Houston 33 last year.

I called him out on his subjective SOS adjustment, and his response was “the brutal truth is that P5 plays the best football.”

These “bottoms up” “per play” numbers guys like to portray themselves as neutral and unbiased, but their biases are inherent in their systems (or they rely on subjective data like recruiting rankings in their initial rankings which is not only skewed but incomplete because it does not account for transfers).

Fremeau (FEI) seems the most reasonable of the bunch.

1 Like

Why don’t Connelly and his ilk just drop the pretenses and stop rating G5’s all together?

Lets list the highest rated G5 teams from the four ranking sited in the referenced article:

BCF toys/FEI Rankings - #25 Boise, #26 UH, #27 BYU
ESPN FPI - #36 Boise, #44 BYU, #45 USF, #47 UH
PowerRank - #27 BYU, #41 UH, #45 Memphis
SBNation S&P+ - #35 BYU, #36 Boise, #41 USF, #45 Western Kentucky, #52 Georgia Southern, #53 UH

Only one of these rating systems has a single team in its top 25. Seems far fetched.

2 Likes

I agree with this. There are just far too many variables–so many of which are completely ignored in these “analytics”–in college football to accurately predict anything. Last year’s 13-1 team is the perfect example. On defense this year, it’s fair to guess we will regress in the secondary with the loss of WJIII, Trevon and McDonald, but it’s just a guess. Because of injuries, it’s not like we have no experience returning. Plus, JJ Dallas and Terrell Williams have college experience (although not at this level). No where is it accounted for that we had one of the best D lines in the country last year that only gets better this year with the addition of Ed Oliver. Even more pressure on QBs mean not as much pressure on the secondary. Quite frankly, I think the only position group that even has a chance to be worse this year than last is the secondary, and I don’t actually think there will be much of a drop off there. That’s just a guess, but at least I’m willing to admit it.

In my old neighborhood in the days of my yewt … we use to say the fella is trying to make a point …

THE HARD WAY …

Over a few weeks, I conducted a comparative analysis using four preseason college football publications – Athlon Sports, Lindy’s, Phil Steele, and The Sporting News – to develop a consensus top 25 among The Gurus. I then compared those composite rankings with four advanced metrics systems released by The Geeks (Fremeau’s FEI, ESPN’s FPI, Connelly’s S&P+, and Feng’s The Power Rank).

Uhhh … why doesn’t he just state that he doesn’t like us and thinks we were lucky …

It would have saved his reading FOUR magazines and attempting to quantify it against some advanced metric systems.

But hey in South Bend its called … “the luck of the irish” … metric systems … schmetric systems … :stuck_out_tongue_closed_eyes:

How idiotic is SBNations system? They rank UH the 6th highest G5 team in their S&P+ list but have UH the #1 G5 school. Even they don’t buy their own crap.

1 Like

Of course the one analytic that dominates all others…are you in a P5 conference !!

Stopped reading when I saw the highlight of Ward vs Memphis was from 2014

As I have noted on several occasions, last year’s team that finished No. 8 nationally was composed primarily of 2- and 3- star athletes which were definitely under the radar in recruiting. We had some ranked players but they were a minority. This has been our history and I personally have come to love it when so-called experts point to our recruiting and extrapolate our ranking from that information. Moreover that has been going on for over 50 years now. Seems a fool never learns.

Fortunately for 50 years our coaches have done an excellent job of evaluating young men at the high school level and then working to develop them once they are on campus. Let the others grab the low hanging fruit. Our way has worked. What will be interesting will be to see how it works now that UH is getting committments from ranked recruits. At this point we will just have to wait it out. Will Oliver be another Wilson Whitley or will Lark become another Greenberry? Too soon to tell.

It’s really important to start the G5 teams as far down in the rankings as you can get away with. After all, they are fighting among themselves for a NY6 bowl, so it really doesn’t matter. Or does it. Most teams start out ranked pretty near where they finished the previous year - unless there has been a huge turnover in personnel and/or coaches. And even then things may not change all that much in the rankings.

But it’s important to keep the G5 teams as far away from the Top 4 as possible. You start by being sure none of them begins the year ranked in the Top 10, regardless of where they finished last year’s season. It lessens the chance that they will rise and pose a threat to the Big 4 and (gasp) make the playoffs. THAT can never, never, never be tolerated. It reminds me of a quote from a non-BCS HC just after the advent of the BCS. He noted that the only way a non-BCS team could play for the national championship would be if it were “unbeaten, untied, and un-scored upon”. Pretty tall order.

2 Likes

I’m sure analytics also said that we would not be 13-1 last year too.

As a big believer in analytics, I feel the need to clarify a lot of issues on this topic.

First, Football analytics at large, and particularly College Football analytics, is relatively nascent. I don’t think anybody is going to make the claim that the current state of College Football analytics is anywhere near as good as, say, Baseball or Basketball. If anybody does make that claim, they’re wrong. Period. There’s a lot less data in football and the specialized nature of positions makes it hard to obtain reliable data for about half the positions on the field at any given time.

That said, human polls are far from perfect. Especially in the preseason, they anchor HARD to the previous year – effectively, the preseason poll is the postseason poll + 3 months to think it over. Likewise, the in-season polls lean heavily on a team’s record and their previous poll ranking, which may not be representative of the team’s actual quality. For example, Tennessee was almost certainly better than Navy last year, but because Navy had only two losses to Tennessee’s four, Navy was ranked higher. A lot of things go into a team’s record that aren’t reflective of the team’s talent – injuries, strength of schedule, and just plain ol’ luck. Analytics tries to correct for this.

And yes, for the record, UH is almost certainly being overrated by the human polls. They had absolutely tremendous luck last year, and if you disagree with that, you weren’t watching. The Coogs blocked, IIRC, exactly two field goals last year, both of which would have been game-winners had they gone in. On top of that, they lose an extremely large portion of their production on both sides of the ball, but particularly on offense. Three of UH’s top four rushers and their top reciever are all gone. Sure, blah blah blah injuries and backups and Ed Oliver etc. But the fact of the matter is, Tom Herman’s team has A LOT of holes to fill, and relying on Freshman and JuCos may not end well. Maybe there’s no drop-off, but that’s far from the likely outcome.

I will concede that there’s a lot of bad stats going on, particularly on the part of Bill Connelly. These analyses are often very simplistic, but they almost certainly get the job done better than human polls, especially in the preseason.

So the reasons we got lucky are valid, but the reasons we didn’t are, “blah blah.” You really proved your point!

1 Like

Appreciate your opinion on this topic. I will respond to some of the points you made. Sorry if this is too long. I have been looking forward to talk through this, because this topic is why CFB is so interesting. Feel free to respond with any objections. Italicized and bold is me and normal font is T-Moar.

I think analytics are a fascinating tool to use when trying to compare teams and/or players in any sport. That said, analytics in college football are inherently biased because we combine statistical analysis (passing yards, tackles for loss) with observational data (projecting players from high school to college). Why is Tennessee obviously better than Navy? Correct me if I am wrong, but I think you believe UT was better than Navy in 2015 because of things like conference affiliation, recruiting, and school popularity. But, if you look at each team’s schedule I think you will find that Navy and UT were very comparable. I don’t care about “talent” and “eye test”. UT beat NW (above average), UGA (average), and BGU (average) but lost to 2 average teams in UF and Arky, while Navy beat USF (average), Memphis (average), and Pittsburgh (average), but only lost to 2 of the best 2015 teams in CFB in UH and ND (although they were blown out in each). It is not fair to penalize Navy because they did not notch a win against an above average opponent. Navy was consistent while UT hit their stride at the end of the season.

As a big believer in analytics, I feel the need to clarify a lot of issues on this topic.
First, Football analytics at large, and particularly College Football analytics, is relatively nascent. I don’t think anybody is going to make the claim that the current state of College Football analytics is anywhere near as good as, say, Baseball or Basketball. If anybody does make that claim, they’re wrong. Period. There’s a lot less data in football and the specialized nature of positions makes it hard to obtain reliable data for about half the positions on the field at any given time. Agreed

That said, human polls are far from perfect. Especially in the preseason, they anchor HARD to the previous year – effectively, the preseason poll is the postseason poll + 3 months to think it over. Likewise, the in-season polls lean heavily on a team’s record and their previous poll ranking, which may not be representative of the team’s actual quality. Agreed

For example, Tennessee was almost certainly better than Navy last year, but because Navy had only two losses to Tennessee’s four, Navy was ranked higher. Disagree – see 1st paragraph

A lot of things go into a team’s record that aren’t reflective of the team’s talent – injuries, strength of schedule, and just plain ol’ luck. Analytics tries to correct for this. Agree, to a certain extent. Explanation in next paragraph.

And yes, for the record, UH is almost certainly being overrated by the human polls. They had absolutely tremendous luck last year, and if you disagree with that, you weren’t watching. The Coogs blocked, IIRC, exactly two field goals last year, both of which would have been game-winners had they gone in. When looking at the 2015 UH football team as “stand-alone”, not as a G5 team on a spreadsheet or a list of the past 4 years of recruiting rankings, you will notice that they FORCED their opponents into mistakes. I do not consider that lucky! Steven Taylor blocking a field goal is not luck. Adrian McDonald catching an interception versus dropping one is not luck, it is a skill that he possesses. The players executed, plain and simple. Saying they SHOULD have regressed to the mean is irrelevant to the 2015 and 2016 Cougars, and not a reason to downgrade them.

On top of that, they lose an extremely large portion of their production on both sides of the ball, but particularly on offense. Three of UH’s top four rushers and their top receiver are all gone. No disputing that.

Sure, blah blah blah injuries and backups and Ed Oliver etc. But the fact of the matter is, Tom Herman’s team has A LOT of holes to fill, and relying on Freshman and JuCos may not end well. Maybe there’s no drop-off, but that’s far from the likely outcome. New projected starters – LT-SO, RG-Jr, RT-JuCo, H-Transfer, RB-Transfer, DE-Jr, MLB2-Sr?, CB1-SO, CB2-SO, FS-Jr, SS-SO, Slot-Sr…….

I will concede that there’s a lot of bad stats going on, particularly on the part of Bill Connelly. These analyses are often very simplistic, but they almost certainly get the job done better than human polls, especially in the preseason. Agreed although not based on a comparison of how preseason polls fared versus analytical models, just a gut feel. (2015 Auburn!!!)

_To summarize: IMO predicting the 2016 CFB season should start and end with, how did the players that are returning fare last year, and what is the track record of the coach that is preparing the new starters. Everything else is guess work. I don’t think averaging or running regressions for certain factors to determine how the new players will perform is reasonable, accurate, attainable or relevant. It is a good business to be in though. I do concede that, on average, the most popular programs=most revenue=afford best coaches=getting the most of players. However, I do not agree with the notion that because Alabama averaged the 1st rated recruiting class for the past 4 years is why they won the championship. I have discussed multiple times to anyone who will read how I feel towards those rankings. I think Alabama has good players that are extremely well coached, which was evident in the 2016 title game when one could argue CLEM had the more “talented team”, but BAMA out-executed them. Who cares how highly rated a safety is if they let the TE run downfield wide open, or if a 4 LB does not run in the correct KO coverage lane and gives up a return TD._*

P.S. Please no “moving the goal-post”. Let the general focus center around why or why not UH was underappreciated or overrated in 2015, and why or why not they will succeed in similar ways in 2016.

Analytics lol. What has that gotten Daryl Morey and the Rockets or Billy Beane with the Athletics lol. Nothing. I don’t see them hoisting up championship trophies. Every team has holes to fill. It just depends who has more than others and what type of backups you had last year. New comers will just have to be ready for the challenge. But that is the job for the coaching staff and I don’t doubt UH’s at all.