Tuesday, November 25

Which Big 12 South team...

...is tops in the division? Here in college football's toughest division, we've got exactly the end-of-season mess we knew we were going to end up with. Texas beat Oklahoma on a neutral field way back in early October. Texas lost on the road at Texas Tech on a last-second bomb to the end zone a few weeks ago. Then Oklahoma blew out Texas Tech at home just last weekend. It's the bizarre combination of timing, margin of victory, and game location that make this debate the most interesting.

For the purpose of argument, let's assume each team ends up 11-1. This means Oklahoma beats Oklahoma State on the road, Texas beats Texas A&M at home, and Texas Tech beats Baylor at home. Obviously this does more for Oklahoma than it does for either other team, whose victories would be about equally (un-)impressive. Who deserves to play in the Big 12 Champioship game, with the opportunity to go on to play in the BCS Championship?

First looking at the head-to-head games.

Texas beat Oklahoma 45-35 on a neutral field in Dallas, about equidistant from both campuses, on Oct 11. Texas outgained OU by a paltry 3 yards, but held a 161-48 advantage in rushing, a +2 turnover margin (practically speaking +1, as Bradford's meaningless final heave into the end zone was intercepted), had a 96 yard KO return, and a +12 yard average punt length. The powerful offenses played to a standstill while Texas won the turnovers and special teams contests.

Texas Tech beat Texas 39-33 in a home game on Nov 1. In a tale of two halves, the Red Raiders led 22-6 at halftime but were outscored 27-17 in the second half. Texas Tech's two second half touchdowns came off a pick-six and of course the game's final drive. The Red Raiders outgained Texas by 205 yards and were +1 in turnovers, but the Longhorns partially negated this with a red zone defense that forced three short field goals. The Horns blocked another FG attempt in the third quarter and Jordan Shipley returned a punt for a TD, as the Horns again won the special teams battle.

Oklahoma beat Texas Tech 65-21 in a home game on Nov 22. The Sooners' 35 point outburst in the second quarter put the game almost beyond doubt at halftime. The Sooners outgained the Red raiders by 219 yards and Sam Bradford threw the ball the manly way, averaging 16 yards an attempt. OU was also +2 in turnovers.

Head-to-head games pretty much narrow it down to Texas and Oklahoma. The Texas victory was by double digits on a neutral field, and the Oklahoma victory was by such a ridiculously lopsided margin that it likely didn't matter where the game was played. On the other hand, the Texas Tech victory came down to the last drive at home, and included a dropped interception which could have ended it with about 12 seconds left. On the basis of this, Texas has the most forgivable loss while Oklahoma has the most impressive victory.

Now looking at other games.

Oklahoma has the strongest nonconference wins, both home wins over Cincinnati (9-2, this week's Big East leader) and TCU (10-2, Mountain West #2). On the down side, they have scheduled a FCS team, Chattanooga, and the only winless team from a BCS conference, Washington. Oklahoma drew Kansas, Kansas State, and Nebraska from the North. Their statement road win would be a victory over Oklahoma State.

Texas has the strongest conference wins, as they are the only team of the three to face Missouri - a 56-35 home victory that was 35-3 at halftime. Texas also drew Kansas and Colorado from the North, both on the road. The Longhorns' nonconference schedule is unimpressive as they have faced Florida Atlantic, UTEP, Arkansas, and Rice. However, all four opponents are FBS teams and anywhere from 1 to 3 will be bowl eligible.

Texas Tech has faced a complete joke of a nonconference schedule - Eastern Washington, Nevada, SMU, and Massachussetts. UMass and EWU are FCS teams and SMU is sitting at 1-10. Within the conference, they did not face Missouri from the North, drawing Kansas State, Nebraska, and Kansas instead.

Again these games seem to simply eliminate the Red Raiders from discussion. Texas's victory over Missouri is the most impressive of the bunch, but OU's win over Cincy will look pretty good if (when) the Bearcats win out against Syracuse and Hawaii. TCU is also a reasonably impressive victory, and Oklahoma State will be a road game for the Sooners. On the other hand, scheduling FCS teams is a major detractor, and Washington is just awful. (the latter not being the Sooners' fault; then again, the results that Arkansas, TCU, and Cincy have posted weren't imminently obvious when these games were scheduled either)

Style Points?

Oklahoma has the #1 offense in the country, averaging a shade over 52 ppg. On the other hand, Texas is the only team in the bunch allowing under 20 ppg on defense.

Bottom Line.

Every argument involving all three teams seems to simply put Texas Tech as the 3rd-best team in the bunch. Their victory over Texas may have indeed been a fluke, as Texas basically didn't show up until halftime and even then TTU needed a dropped interception to win the game. But does the "fluke" loss make Texas #1? I don't know that you necessarily reward a team for pissing away an entire half of football in a known marquee matchup. At least when Oklahoma lost, they played well enough so that the only reason Texas was able to win was by elevating their own level of play in the second half. (the injury to OU's MLB also helped; then again, he's lost for the season so that can't really be written off)

However, if Texas Tech is clearly third in the group for multiple reasons, then the head-to-head matchup matters that much more. I think the tiebreaker goes to Texas, and the win over Missouri (more impressive than a win over Cincy) ices the deal.

What about Bob Stoops' argument?

Bob Stoops made the argument that ranking Texas ahead of Oklahoma on the basis of head-to-head also implies that Texas Tech should be ahead of Texas on the basis of head-to-head. Obviously, it is impossible to rank all 3 teams on that basis, as Oklahoma also beat Texas Tech.

Stoops' argument is one of retrodictive accuracy in polling, and it ignores one thing. Of the six possible ways to order these three teams, three such sets produce one ranking violation while three sets produce two ranking violations. The Occam's Razor solution is to try to have a set of rankings which relies on as few outcomes being "flukes" as is possible (and rational). Those three sets are:

1. Oklahoma, Texas Tech, Texas
2. Texas, Oklahoma, Texas Tech
3. Texas Tech, Texas, Oklahoma

A better version of Mack Brown's argument is that, once we have decided that Texas Tech is the #3 team, which it seems like just about everyone agrees on, it makes more sense to use the order
Texas, Oklahoma, Texas Tech
than it does to use the order
Oklahoma, Texas, Texas Tech
because the former only produces one ranking violation while the latter produces two. Or said another way, either way we are assuming that Texas is better than Texas Tech despite their head-to-head. Adding on top of that Oklahoma being better than Texas despite their head-to-head means having to ignore even more results of important games.

The great irony here is that the counterargument relies on margin of victory - only the OU-TTU result was a blowout where one team was clearly better than the other; therefore, its result is the most likely to be indicative of the teams' relative strengths, while the other two games could be considered coin flips that just happened to come out the way they did. A power rankings which include MoV could say that OU > UT > TTU, say at each time by a small amount (enough that upsets aren't too unlikely), but the two little gaps combined imply that OU >> TTU. Such a sophisticated ranking system would use both homefield advantage and margin of victory as well as how recently the games were played. HFA gives Texas an advantage, while MoV and timing give OU an advantage. The computer rankings are explicitly not allowed to include MoV in part because that helped Texas get a BCS berth over Cal in 2004. Texas may now benefit from a rule that was designed to "correct" the system for how it helped them (and thwarted the Rose Bowl/Jim Delany) several seasons ago. The non-explicit implication of this was that the BCS valued retrodictive accuracy over predictive power (also implied by the widespread outrage over the selection of 2000 FSU over Miami and 2001 Nebraska over Colorado - nevermind that choosing Miami ignores Washington, and choosing Colorado would ignore overall record). So in addition to throwing out margin of victory, the computer polls used in the BCS also stopped using HFA and most (if not all) stopped using the timing of the games.