Big 12 Games Preview - 10/8/11

by Huck L Berry on Oct 3, 2011 10:51 AM CDT in Texas Longhorns and Other Football

Time for the weekly look at the upcoming Big 12 games based on ratings and adjusted stats to date. Keep in mind that personnel matchups, coaching decisions, injuries, etc. will not play into the numerical predictions and are all important information that will still need to be considered.

Some of you may be familiar with my matchup analyzer, which predicts a final score based on the power ratings of each team as well as specific stats for the game. What it doesn't yet have, but may be added soon, is a second final score prediction based on the adjusted stats it shows. As discussed in the Adjusted Stats Year in Review, regression analysis can be used to estimate points scored against rushing, passing, and turnovers. I have chosen to use 2011 stats to update that regression weekly based on the season so far. For 2011 through the games of 10/1/11 that number is currently 2.71*TPPA + 2.76*TRPC - 113.91*TVPP, which is used for the below analyses.

Also note that my power rating predicted scores below will differ form the analyzer because I'm using a standard 3 point homefield advantage while the analyzer uses the season results to date. Additionally, the official picks published later in the week may differ from the Other Games of Interest section because standard deviations aren't considered here.

Oklahoma (-9) vs. Texas - (@ the Cotton Bowl) 11:00 ABC
Power Rating Predicted Score - Oklahoma 34, Texas 15
Stat Regression Predicted Score - Texas 25 (25.4), Oklahoma 25 (24.8)
Matchup Analyzer Link

Huge discrepancy between the two models but the point spread basically pegs the middle of the two. I'd like to take solace in the fact that the stat regression seemed to have a better week last week (what a brutal beat both models took in the Georgia Tech game), but honestly it makes sense that Texas would do better in a model that doesn't include our woeful special teams performance to date. At least that's what I thought. Then I checked both Oklahoma and Texas against a 2011 Average team and was a little surprised. The result was that Texas would be expected to beat an average team 31-17 in the power model and 31-19 in the stat regression model. Oklahoma, meanwhile, is predicted to win 45-12 and 33-21, respectively. So it seems the Sooners are the ones with a major discrepancy in the two models. So where does that leave me? Honestly, with no idea on how to predict this game, which shouldn't be a surprise to fans of either team. This game frequently hinges on big plays, ridiculous levels of emotion, and who handles the tension best. Not just the players, but the coaches calling the plays. The line seems reasonable and I'll just hope that Texas gets at least serviceable performances from special teams units and wins the turnover battle. That being said, I will officially predict a Texas blowout because OU sucks, that's why.

Missouri (-3) @ Kansas St. 2:30 ABC
Power Rating Predicted Score - Kansas St. 34, Missouri 22
Stat Regression Predicted Score - Missouri 38, Kansas St. 35
Matchup Analyzer Link

This makes two weeks in a row that the ABC 2:30 broadcast will be LIVE FROM THE LITTLE APPLE! Last week the Wildcats under Wizard Snyder made a huge play when it mattered to knock off Baylor. The current line on their matchup with the Tigers is right on the nose with the stat model but it seems the power rating model believes in Snyder Magic. And the numbers show that Kansas State is indeed the squad with the power/stat discrepancy. Probably some of that Manhattan Magic that I refuse to pick against. All Longhorn fans know that we are owned by the purple cats, so I am unwilling to offend our masters by picking against them. Mark it down as an outright Kansas State victory.

Kansas (+33) @ Oklahoma St. 2:30 No TV
Power Rating Predicted Score - Oklahoma St. 59, Kansas 17
Stat Regression Predicted Score - Oklahoma St. 48, Kansas 20
Matchup Analyzer Link

This should just be a fantastic game. Turner Gill is about to turn the corner up there in Lawrence and this could be their breakout moment. Jayhawks cover and lose 52-20.

Iowa St. (+16.5) @ Baylor 6:00 FSN
Power Rating Predicted Score - Baylor 45, Iowa St. 36
Stat Regression Predicted Score - Baylor 34, Iowa St. 22
Matchup Analyzer Link

Fairly consistent margin from the two models even if the totals vary wildly. The Bears suffered a late game defeat on Saturday but still have the best player in the conference, while the best name in the conference comes into town off a big loss to the Longhorns. The two models agree that Baylor wins but the Cyclones cover, so that's what I'll go with on this one.

Texas A&M (-6) @ Texas Tech 6:00 FX
Power Rating Predicted Score - Texas A&M 48, Texas Tech 33
Stat Regression Predicted Score - Texas A&M, 41, Texas Tech 28
Matchup Analyzer Link

Along with Texas/OU this is the second Participating Schools' Biggest Rivalry Game of the week. As the last conference game edition of this soap opera, we can only hope for as entertaining a game as these two teams have put on for us in the past. Both models agree that the Aggies should cover but it's important to note that I have not yet coded a Mike Sherman Second Half Adjustments Factor into the equations. There's no good reason to expect the Red Raiders to win on paper, but then again there weren't any good reasons on paper to expect the Aggies to blow 17-point and 18-point halftime leads the last two weeks. Bad news for the Aggies is that the second half of this game will be at night in Lubbock. Worse news is that Gus Johnson will be calling it. SETH DOEGE STANDS AND FIRES!!!!!... Better hope for an un-Shermanable halftime lead, like maybe 45 points or so. Still, I'll go with the Aggies to cover just because it's a good jinx situation.

OTHER GAMES OF INTEREST

Michigan (-5.5) @ Northwestern
Power Rating Predicted Score - Michigan 45, Northwestern 12
Stat Regression Predicted Score - Michigan 43, Northwestern 21
Matchup Analyzer Link

Georgia Tech topped this list last week and was up by 24 points with less than a minute remaining. Shortly thereafter they completed their 10-point win and brutal beat for the computer models who had them as the top pick as 11-1/2-point favorites. There isn't any line as out of whack with the models this week, but the Michigan game is close. The Wolverines are seen as an undefeated pretender by lots of college football fans that remember the Rich Rod years, but this year's edition is also playing decent defense. After ranking 97th in adjusted yards per play allowed last year the 2011 edition is 21st through five games. This line actually opened at only 3-1/2 points but has quickly moved past the important number at 4 and is closing in on another one at 6 points.

Troy (-4) @ Louisiana-Lafayette
Power Rating Predicted Score - Louisiana-Lafayette 44, Troy 33
Stat Regression Predicted Score - Louisiana-Lafayette 33, Troy 23
Matchup Analyzer Link

Night game home dog (does it matter in the Sun Belt?) that both models say should win. I'm not going to pretend like I've seen either team play. Line opened at 4.5 points and has moved a hair to the Ragin' Cajuns.

TCU (-4.5) @ San Diego St.
Power Rating Predicted Score - San Diego St. 38, TCU 28
Stat Regression Predicted Score - San Diego St. 43, TCU 25
Matchup Analyzer Link

Another night game home dog that the models agree should win by double digits. TCU's defense has been simply awful so far and this line almost seems too good to be true as a long term EV play. On offense the Horned Frogs have run the ball well but their passing game is mediocre. I haven't seen the Aztecs play aside from a few minutes against Michigan, but I don't have much confidence in TCU right now.

Kent St. (+16.5) @ Northern Illinois
Power Rating Predicted Score - Northern Illinois 30, Kent St. 25
Stat Regression Predicted Score - Northern Illinois 18, Kent St. 16
Matchup Analyzer Link

Haven't seen either play but this is the final game of the week where both models disagree with the line by double digits in the same direction. Can't say I'm a big fan of betting on the Golden Flashes in a road game, but there it is.

That's it for this week.

Tweet Comment 20 comments | 0 recs |

Do you like this story?

Comments

Good stuff, as usual, Huck… but there ain’t but one game I care about. Already starting to hyperventilate.

by Tex Long on Oct 3, 2011 10:57 AM CDT reply actions

I was wondering if you charted success with these games. Could you put that in the top of next weeks preview if it isn’t too much trouble? I love reading these, but when Saturday comes around I never know what happened on Monday. A charted success of each (correct if 10 points different from vegas line, correct against vegas line straight up, correct predicted win, etc. for each rating would be very interesting to me as I read).

Thanks so much for all you do.

hook’em

by uttuck on Oct 3, 2011 11:43 AM CDT reply actions

Looks like the Other Games of Interest was 2-2 last week

by Last Week on Oct 3, 2011 11:57 AM CDT reply actions

I’m so ready for Saturday!

http://www.youtube.com/watch?v=QyEfbRkKcmI

http://www.youtube.com/watch?v=FWFbGw-jZvc

Oh, and OU still sucks!

by One flag. One star. One state. One school. on Oct 3, 2011 11:59 AM CDT reply actions

I have not yet coded a Mike Sherman Second Half Adjustments Factor into the equations

Commonly called a “nap.”

by parlin on Oct 3, 2011 12:17 PM CDT reply actions

It’s hard to track it in terms of difference from the line simply because the models and the lines start to squeeze closer and closer as the season progresses. This weekly post doesn’t really use the full system including standard deviations as I mentioned above and I don’t have any history yet with the stat regression model. I will probably go back and add history during the offseason.

Yes, they went 2-2 last week with the brutal beat on the Yellowjackets. The Air Force/Navy game also missed but that is at least partially explained by Air Force only having one game in the data set prior to last week. Still a miss is a miss, a beat is a beat, and the other games went 2-2 last week.

Keep in mind that a 0.524 winning percentage into -110 odds treads water over the long haul. 55% is good stuff and 57.5% is outstanding. 60% is unheard of. Around 0.576 nets you a unit every 10 bets made.

by Huckleberry on Oct 3, 2011 12:22 PM CDT reply actions

Huck,

Where do you find adjusted yards per play statistics?

by bigdukesix on Oct 3, 2011 12:23 PM CDT reply actions

I don’t “find” them so much as I calculate and publish them. Ha.

Adjusted Stats home page
Links page (if the dropdown menus on the main pages aren’t working for you)

by Huckleberry on Oct 3, 2011 12:47 PM CDT reply actions

Oklahoma should win because they are a seasoned team and Horns aren’t. But we have a true freshman qb and a red shirt qb that are playing really good -at least as far as their college experience is concerned, which is zilch. I’m not expecting Texas to win, but if they do, it’ll be more to gloat over. Its a rebuilding year for Texas so our expectations can’t exceed more than that. So far we’ve had a great year-undefeated. We have moved on from GG and that’s very satisfying.

by staylucky on Oct 3, 2011 1:13 PM CDT reply actions

Fantastic. Thank you.

by bigdukesix on Oct 3, 2011 1:24 PM CDT reply actions

Thanks, Huck. I’m personally glad that the AF-Navy game missed.

I am curious how you implement the regression-based predictions. In the year-in-review document you linked to, I saw that you estimated models for both points scored and points allowed. Are your predicted points reported here based on the points scored model, the other team’s points allowed model, or some sort of average of the two? The latter approach seems like it would make the most sense to me.

If you have the points data at the game level, another approach that may yield good results would be to regress a team’s points scored in a given game on its average adjusted passing, rushing, and turnover stats, as well as the corresponding defensive stats of the opponent. If you estimate the regressions separately for home and visiting teams, it would also incorporate any home field advantage into the predictions.

Another minor question: do you have an intercept in your regression model? If not what’s the reason for suppressing it?

by AFHorn on Oct 3, 2011 2:24 PM CDT reply actions

I’m using a regression based on actual games and the stats accumulated in the specific games. Your approach of regressing actual game scores against incoming ratings is a valid approach but I have never run that analysis for previous seasons. It may be worth following moving forward.

The intercept is forced to zero. The reason is I’m lazy and it makes me happy without having to think too much.

by Huckleberry on Oct 3, 2011 2:34 PM CDT reply actions

Huck -

This is great stuff. As the data points grow, your predictions really gain in accuracy.

I remember last year when we were all giddy post-Nebraska, your model predicted a losing season. Everyone chuckled, since we were 4-1 at the time and little did we know….

by Scipio Tex on Oct 3, 2011 3:13 PM CDT reply actions

Well right now the power model has us with 4.23 expected wins for the remainder of the schedule and the stat model has 5.56 wins.

Splitting the difference roughly at 5-3 in the final eight games seems like a reasonable expectation/hope for Texas fans.

by Huckleberry on Oct 3, 2011 3:39 PM CDT reply actions

My reverse analysis regression global positioner reflux gambit predicts mudholes a plenty for OU’s ass.

by Thujone on Oct 3, 2011 9:56 PM CDT reply actions

My double trade secret ubereconomic statistical t distribution predicts that this year Oklahoma graduates maintain an average of less than 74% of teeth retained by the age of 30. Fully 6.3% have at one time played naked Jack Daniels slip-n-slide with their fellow AA meeting attendants. This down 2% from last year due to the more wide sprea distribution of less expensive sour mash vatieties such as Evan Williams. Finally, of those who were able to retain their teeth, 56% were actually able to eat a corn on the cobb through a tennis racket. These data are accompanied by a statistical uncertainty od + or – 2%.
Hookem!

by Jkabuldog on Oct 4, 2011 8:02 AM CDT reply actions

“The intercept is forced to zero. "

Were it only this easy on the football field. Sigh.

by roach on Oct 4, 2011 9:50 AM CDT reply actions

Player 1: [sighs] Christ. OK, uhh well what we’ll do, I’ll run in
first, uh gather up all the eggs, we can kinda just, ya know blast
them all down with AOE. Um, I will use Intimidating Shout, to kinda
scatter’em, so we don’t have to fight a whole bunch of them at once.
Uhh, when my Shouts are done, uhh, I’ll need Anfrony to come in and
drop his Shout too, uh so we can keep them scattered and not have to
fight too many. Um, when his is done, Bass of course will need to
run in and do the same thing. Uhh, we’re gonna need Divine
Intervention on our mages, uhh so they can, uhh, AE, uh so we can of
course get them down fast, cause we’re bringing all these guys, I
mean, we’ll be in trouble if we don’t take them down quick. Uhh I
think this is a pretty good plan, we should be able to pull it off
this time. Uhh, what do you think Abduhl? Can you give me a number
crunch real quick?

Abduhl: Uhhh.. yeah gimme a sec… I’m coming up with thirty-two
point three three, repeating of course, percentage, of survival.

by Bravo on Oct 4, 2011 2:10 PM CDT reply actions

LEEEEEEEEEROY JENKINS!!!!!!

Also, I put together AFHorn’s suggested model that runs the regression between actual points scored and the ratings of the two teams. Predicted scores for all the above games (will obviously be closer to the basic stat regression model than the power model):

Texas 25.0, Oklahoma 24.9
Missouri 38, Kansas State 33
Oklahoma State 52, Kansas 18
Baylor 35, Iowa State 21
Texas A&M 45, Texas Tech 21

Michigan 46, Northwestern 18
Louisiana-Lafayette 31, Troy 22
San Diego State 41, TCU 28
Kent St. 15, Northern Illinois 14

One thing to note is that I have incorporated net punting and net kickoff numbers into this model.

by Huckleberry on Oct 4, 2011 2:29 PM CDT reply actions

Thank you for the auspicious writeup. It in truth was once a amusement account it. Look complicated to far added agreeable from you! However, how can we communicate?

by iphone 5 iphone 4s ipad mini ipad 3 apple mac news steve jobs ipod free jailbreak ios 5.01 redsn0w youtube on Nov 13, 2011 12:09 AM CST reply actions