Blogging about college football by an Oklahoma Sooners fan.

Does Oklahoma Overlook Weak Opponents?

Tommy Tuberville and Bob Stoops
Twenty-eight points.

As soon as the final gun sounded a few weeks back in the Texas Tech Red Raiders' 41-38 shocker over Oklahoma, putting a serious dent in the Sooners' national title aspirations, the four-touchdown spread was trotted out as evidence of the enormity of the upset. Even Tommy Tuberville made note of it in his postgame comments. In the subsequent fallout, pundits have pointed to the mammoth line in support of the meme that Bob Stoops' teams have a tendency to suffer inopportune letdowns against inferior competition.

Intuitively, that sounds right. But how to test that hypothesis?


We typically associate point spreads with scores, because that's how bets are made. If a team is favored by 10 points, a 31-20 win represents a cover by the favorite.

If you break down the data, however, you'll find that point spreads also correspond to the likelihood that the underdog or favorite will win outright. As you'd expect, the higher the line, the higher the likelihood that the favorite wins straight up. In 2009, college football guru Phil Steele demonstrated this phenomenon.

Spread % Favorites
That Win
>31 98.7
24.5-31 96.1
17.5-24 93.0
14.5-17 86.5
10.5-14 79.9
7.5-10 73.6
3.5-7 65.9
<=3 51.1

Now, let's look at Stoops' performance relative to expectations and see what kind of conclusions can be drawn (if any).

Spread Games Wins Win %
>31 25 25 100
24.5-31 22 19 86.4
17.5-24 28 26 92.9
14.5-17 9 7 77.8
10.5-14 21 20 95.2
7.5-10 16 12 75.0
3.5-7 19 11 57.9
<=3 8 7 87.5

Also, OU has been an underdog a total of 19 times under Stoops, winning nine of those games (47.4 percent). Considering the projected win probability vis a vis the spread, the Sooners have outperformed as underdogs by 1.5 games. (Take my word for it.)

If we're using point spreads to project how many games the Sooners should have won under Stoops, it would be 131. In fact, OU has won 134. (Sounds like a pretty good predictive model to me.)

So, what should we take away from this little exercise?

*First things first, Steele's research used the results of nearly 17,000 games from a span of 12 years. In this exercise, we're using a total of approximately 170 games. The most we have for any cohort is 28 events. As such, the sample size is small enough that we should be skeptical of drawing any hard-and-fast conclusions based on these numbers alone.

*Looking at how OU has performed in the 24.5-31 range, it certainly suggests that the Sooners are more prone to being upset in those kinds of games than your average team. Does that mean OU overlooks opponents?

In one of the three losses, TCU in 2005, the oddsmakers were clearly overrating an OU team coming off back-to-back appears in the national title game. The other two, Oklahoma St. in 2001 and Texas Tech this year, definitely smack of looking ahead or just lacking focus.

*You can't knock OU for its letdowns as chalk and not give the Sooners props for their level of performance as a small favorite or underdog. In those situations, you could argue OU's success hints that the teams was especially dialed-in.

Of course, exceeding expectations in such cases makes the big upsets stand out all the more.

Overall, I'd say the data support the contention that the Sooners tend to blow off middling opponents. What Crimson and Cream may have to ask themselves, though, is if there's a trade-off? Is it possibly a matter of giving up some of your prep time leading up to chump in exchange for having just a little extra for a big game? Do you have to sacrifice some of your mental edge against perceived cream puffs to be sharp for your better opponents?