Discussion in '2012 Football Season Capsule' started by BB73, Jan 2, 2012.
November means an afternoon game in Madison this year.
Wonder if (or when) they'll ever get rid of the November night game rule.
Stave broke his collarbone against MSU so we will play against the transfer.
Awful blow for Stave and Wisconsin. The kid was coming along nicely. We'll see what O'Brien's got, I guess. So far, it hasn't been much.
Urban Meyer's Record in Games Following a Bye Week
W_____10-10-2009_____13____LSU__________3_____Baton Rouge, LA
* = Florida's only win in October in 2007
W_____10-16-2004_____46_____UNC________16_____Salt Lake City, UT
W_____11-15-2003_____47_____Wyo________17_____Salt Lake City, UT
W_____09-27-2003_____28_____CSU________21_____Fort Collins, CO
W_____10-05-2002_____72____Ohio________21_____Bowling Green, OH
W_____09-14-2002_____51____Mizzou______28_____Bowling Green, OH
L_____11-03-2001_____21____Miami U_____24_____Bowling Green, OH
W_____09-22-2001_____42____Temple______23_____Bowling Green, OH
Long story short: Urban has only lost one game after a bye week, and that was 11 years ago, in his second such game ever.
Plus the Miami team had Big Ben, who threw for over 300 yards. Dude turned out to be pretty good.
Behind the Numbers
I realize this is a week early, and the numbers might change a little after Wiscy's game with the Hoosiers; but I couldn't resist the temptation to see what DSA has to say about this game. If you know what DSA is, feel free to skip the explanation.
Differential Statistical Analysis (DSA) - explained
For starters Differential Stats have nothing to do with differential equations. There is no calculus involved; only simple arithmetic.
If a team usually gains 400 yards per game but Ohio State holds them to 200 yards, then the Buckeye's Differential Total Defense (DTD) for that game is 0.5 (their opponent gained 0.5 times as many yards as they usually do). If a team were to similarly hold all of their opponents to half their usual output, then their DTD for the season would be 0.5. It's that simple. From here it should be intuitively obvious that a differential stat of 1.0 is average for both offense and defense, and that above 1.0 is above average for offense while below 1.0 is above average for defense.
One of the benefits of DSA is that you can use it to make predictions. Take one team's average offense and the other's differential defense, multiply them and there's your expectation for that stat for the game. Alternatively you can take the first team's differential offense and multiply by the other team's average defense.
Before we look behind the numbers we first have to see what the numbers are. For the purposes of this discussion Wisconsin's game against UNI (FCS) has been discarded. What follows are the DSA expectations for the output of each team in a few categories.
Wisconsin Expectation: 122.3 to 132.7 yards
Ohio State Expectation: 179.1 to 185.8 yards
Wisconsin Expectation: 187.4 to 220.9 yards and Rating of 114.0 to 131.7
Ohio State Expectation: 156.8 to 168.5 yards and Rating of 120.9 to 130.4
(Total yards was found via DSA, by multiplying average offense by differential defense and vice versa. For this reason it arrives at a different number than adding the expected rushing and passing numbers.)
Wisconsin Expectation: 325.4 to 370.8 yards
Ohio State Expectation: 345.2 to 365.5 yards
Among the many, many things that have bothered me about the Badgers over the years is that they are bullies. Under Alvarez and now Bielema, they amass their stats by running the numbers up on lesser opponents and (seemingly) under-performing against good football teams.
It occurred to me as I perused the numbers that there is a way to quantify this. Using the statistical definition of correlation, you can determine how a team compiled its differential statistics.
Here's how it works: In statistics, one series of numbers can be compared to another to see how related they are. Correlation, as it's called, varies from -1 to 1 depending on how the numbers relate to one another. Take Differential Rushing Offense, for example. Ohio State's Differential Rushing offense is 1.595, meaning that they rush for 59.5% more yards than their opponents typically surrender. They compile those numbers however by having their best games against the best opponents. The result is that the correlation of their Differential Rushing offense to their opponents Rushing Defense is -0.629. The fact that the number is negative means that the Buckeyes' DRO tends to go up as the opponents' rushing defense goes down. The absolute value of 0.629 means that this is a fairly strong correlation.
The other side of that coin is the Wisconsin defense. They are 2nd to the Buckeyes in the conference in rushing defense, giving up only 116.5 yards per game, and their Differential Rushing Defense is 0.699, again right behind the Buckeyes' number (0.672). It might appear that this is a good rushing defense set to take on a good rushing offense... until you look at the correlation of differential numbers to opponents' traditional numbers.
As was previously stated, the Buckeyes' offense has a STRONG tendency to have their best statistical games against their best opponents. This by itself would lead us to tweak the expectation in the Buckeyes' favor. The really good news then, is that the correlation between the Badgers' Differential Rushing Defense and their Opponents' Rushing Offense is 0.505. Note that this is a positive number. In other words, the stinkin' Badgers have a STRONG tendency to have their WORST games against their best opponents.
While most of the other correlations (which I'll post later) are in the Buckeyes' favor, this one is dramatic. The combination of Ohio State's -0.629 correlation to Wisconsin's 0.505 means that Ohio State's rushing yardage expectation should be adjusted significantly upward.
Television analysts will probably tout this as close match-up, but don't buy it. The Buckeyes are going to run rough-shod over the Badgers.
Love that stuff, DBB!
DSA - Correlation
In my previous post I went into a lot of detail on correlation for the benefit of those who had not been exposed to the concept. That post showed that Ohio State's Differential Rushing Offense (1.595) had a significant negative correlation (-0.629) to their opponents' average rushing defense. Further, it was explained that this comes from the fact that Ohio State's rushing offense performed best against the best defenses. To put it another way, Ohio State's rushing offensive output is less susceptible to opposing defenses: the better the defense, the more that Ohio State will exceed what that defense usually gives up. In this way, OSU's rushing offense statistics can be thought of as "rigid". Conversely, Wisconsin's rushing defense was shown to have their worst games against the best offenses, having a positive correlation (0.505) between their Differential Rushing Defense (0.699) and their opponents' average rushing offense. The combination of the rigidity of OSU's rushing offense output and the corresponding lack of the same in Wisconsin's rushing defense indicates that the prediction of 179.1 to 185.8 yards should be corrected significantly upward. It would be unreasonable to expect that Wisconsin's defense can hold the Buckeyes to less than 200 yards rushing.
Instead of continuing to use the term "correlation", it is more convenient to coin a new term called "rigidity" and define it thus:
rigidity = -100 * correlation
So where correlation goes from -1 to 1 and negative numbers are best, rigidity goes from -100 to 100 and positive numbers are best.
The Numbers - Adjusted
OSU Rushing Defense____-12.7
Wisc Rushing Offense___-1.3
Wisconsin rushing offense prediction should be revised slightly upward by this analysis, but this does not take into consideration the style of play. Wisconsin's offense is running right into Ohio State's wheelhouse. We know from experience we can revise the DSA prediction downward.
OSU Passing Defense____-16.6
Wisc Passing Offense____58.0
Wisconsin passing offense prediction should be revised significantly upward. This makes sense, as the rest of this analysis suggests they'll be playing from behind for much of the game.
OSU Pass eff. Defense____56.5
Wisc Pass eff. Offense___15.7
Wisconsin passing efficiency prediction should be revised moderately downward. Even if they had Stave, which they don't.
OSU Total Defense____-40.4
Wisc Total Offense___-42.3
Wisconsin total offense prediction should be unaltered or perhaps adjusted very slightly downward
OSU Scoring Defense_____24.3
Wisc Scoring Offense___-20.5
Wisconsin scoring offense prediction should be adjusted moderately downward. While the OSU rushing offense prediction is the biggest adjustment based on rigidity, this adjustment is just as significant.
Wisc Rushing Defense___-50.5
OSU Rushing Offense_____62.9
As previously belabored: Ohio State's rushing offense should be adjusted DRAMATICALLY upward. The Buckeyes bludgeon the Badgers. This affects everything else too.
Wisc Passing Defense___29.9
OSU Passing Offense____24.6
Ohio State's passing offense prediction should be revised very slightly downward
Wisc Pass eff. Defense___53.7
OSU Pass eff. Offense____59.2
Ohio State's passing efficiency prediction should be revised very slightly upward
Wisc Total Defense___23.2
OSU Total Offense____62.1
Ohio State's total offense prediction should be adjusted moderately upward
Wisc Scoring Defense___52.2
OSU Scoring Offense____46.1
Ohio State's scoring offense prediction should be adjusted very slightly downward according to this analysis, but the huge adjustment to the rushing numbers overwhelms this result. Wisconsin will not be able to get the Buckeye offense off the field.
Without the addition of "statistical rigidity" to this analysis, DSA would have made the same mistake as all of the other analysts: that this game looks to be close (or would be if Stave were healthy). The deeper look behind the numbers afforded by this concept allows us to say with confidence that the Badger's are utterly out of their depth and Stave would not have saved them.
Having already gone way past the point where most people are interested, I won't post the numbers for the next level of abstraction. Instead, I'll simply provide the implications.
The numbers indicate that the Badgers are much worse at Passing when facing good Rushing defenses, both in terms of yardage and efficiency. Their susceptibility to good rushing defenses also has a significant effect on their total offense and scoring offense.
Long story short: The Badgers are very vulnerable to stout rushing defenses in a way that affects their entire offense; and the Buckeyes' rushing defense leads the league in spite of all the big plays they gave up earlier in the season.
Take ALL of the DSA predictions for Wiscy's offense and reduce them by 25%. This is at least a two touchdown game.
Love the mathematical models! Not to ask for more work, but any correlation to home/away? Only reason I'm worried is there is always a human factor(s). Two human factors that are in Wisconsin's favor:
1. Games at Camp Randall.
2. They have the revenge factor.
Doubt there's a way to work in the revenge factor, but home and away differentials in the numbers would be interesting.
As for home/away, the issue is that this exacerbates the problem with all college football statistics: sample size.
As we close in on the end of the season, the sample size (# of games) for most stats is ten or less. That's bad enough; but looking at OSU's away games, the sample size is 3. One bad game (in Bloomington) can (and does) dominate those statistics.
Let's just look at Differential Scoring Defense (DSD).
So even though the Silver Bullets played decent defense for most of those 3 games, the horrific 4th quarter against the Hoosiers leaves the overall DSD for those 3 games at 1.104 (anything over 1.0 is below average).
The real question then is "will the Buckeyes play with the same lack of passion that allowed Indiana to run wild?" The real answer is: That is highly doubtful.
This begs the larger question of the under-performances that skew the statistics in general. The best example this year is not the Indiana game, but rather the home game played against the other Big Ten team from that miserable state, Purdue. OSU-Wisconsin have 4 common opponents. If you look at the game each played against Purdue, Wisconsin looks better. If you look at the other 3 common opponent games, Ohio State looks better.
So the real-real-question is: "will the same Buckeyes who played Purdue show up". Considering that Meyer is 14-1 in games after bye weeks, that too seems doubtful.
The statistical analysis is great, but the biggest single factor IMO is the fact that the Badgers are forced to play O'Brien at QB. Had we been playing Stave, I'd be much less optimistic; as it is, I think we'll kick their butts.
Stubborn Bielema sticks with running game..tOSU wins easily..Wisky gets creative. tOsu wins close game. Meyer has schematic advantage LOL.
Curt Phillips will actually be starting this weekend in Bloomington. How he does/if he can stay healthy will go a long way towards being able to predict next Saturday's game.
Separate names with a comma.