clock menu more-arrow no yes mobile

Filed under:

The Stoops Paradox

I see a lot of people parroting Stoops' line about how you can't use Texas' head-to-head victory over OU in voting unless you also vote Tech ahead of Texas...but yet OU beat Tech. OMG! There is no solution! I am surprised by how many people seem to buy this argument. It's an example of specious reasoning called a paradox. Merriam Webster defines "paradox" as:

2 a: a statement that is seemingly contradictory or opposed to common sense and yet is perhaps true b: a self-contradictory statement that at first seems true c: an argument that apparently derives self-contradictory conclusions by valid deduction from acceptable premises

The argument that Stoops lays out is a great example. Each element is sound. Texas did beat OU. Tech did beat Texas. OU did beat Tech. However, if you buy the proposed premise, then all of these head-to-head results are useless in rating these teams. Captain Kirk could have really stumped that alien computer with this one!

Obviously, that's stupid. Head to head results are the best data we have for ranking teams. In a playoff, head to head results are all that you use. The unique round robin aspect (thought- would everybody be willing to reject the RRS result if a non-conference opponent were subbed for Tech in this menage a trois?) adds a layer of analysis, but you still need to understand that they mean something.

So, what to do? Stoops proposes throwing them out, and evaluating on the remaining body of work. That is an argument, and lacking another proposal, it is the only one on the table. I think there are better ways to analyze this, though.

I'm not interested in proposing tie-breaking systems. That horse is long out of the stable. I'm interested in ranking teams. I know that the system (polls and computers) does not define what is being ranked- season's body of work, or hottest team now. Still, the question is- Who has the better team from among these three 11 - 1 teams? Can you use the head to head results? Yes, you can.

Each team has one win and one loss. Ask yourself about each team- if the teams replayed on a neutral field, could you expect a similar result (i.e. did the result seem sound, or a fluke). Obviously, point differential comes into play. Next, ask yourself about each team's loss- if the teams replayed on a neutral field, would you expect a similar result?

Texas- beat OU by 10 on a neutral field, lost to Tech by 6 on the road

OU- beat Tech by 44 at home, lost at a neutral field to Texas by 10

Tech- beat Texas by 6 at home, lost to OU on the road by 44

The results on the field suggest that in a neutral site round robin, Tech is most likely to go 0 - 2. Who rates higher- Texas or OU? I'm not getting into that. I have my opinion, the Sooners have theirs (wrong), and the voters need to form their own. They just shouldn't be stupid and intellectually lazy enough to rely on the Stoops Paradox and discard the results.