Trust the Computers

A couple weeks ago I posted a piece hoping to alleviate fans’ BCS fears, urging people to trust that the voters would make a good choice. In the wake of the Oklahoma beatdown of Texas Tech and the snarled tangles that followed, this post is directed to those voters.

I know what you’re thinking. In the three-way tie that is the top of the Big12 South, you’ve written off Texas Tech because of the nature of their loss – getting pasted 65-21 (even on the road) is terrible. Oklahoma only lost by 10, and Texas only lost by 6, so the Red Raiders are basically out of it. (Assuming of course that all three win this week. If they don’t then the whole point is moot because someone will automatically wrap up the South title.)

So if it’s between Texas and Oklahoma, you voters have two ways you’re going. First, in the head-to-head matchup, Texas beat Oklahoma, therefore they should be ranked higher. Second, in the margin of victory matchup (which you used to eliminate Texas Tech), Oklahoma beat Texas Tech much more handily than Texas beat Oklahoma, therefore Oklahoma should be ranked higher. Both are logical and valid, and neither is right while the other is wrong. That’s just the way it is.

But here’s the thing, voters – these aren’t your only two options.

The reason these are overwhelmingly the two ways people are seeing the situation is because a lot of the other variables that humans can and usually use aren’t available. Both teams are 10-1, both went 1-1 against the others in their little trio, both played a lot of the same opponents since they’re in the same conference, both lost their one game on the road or neutral site… So humans are left with few options for comparison.

But what about the computers? There haven’t been any huge, non-statistical, game-changing issues on the field like the Oklahoma-Oregon debacle of 2006, so you should feel comfortable that the computers aren’t going to “miss” anything major. And although the BCS computers can’t take margin of victory into account, a lot of other computer models can, along with a host of other data that will help you separate Texas and Oklahoma. They can compare stats and teams in ways that humans can’t, and we’d be foolish not to use them to do some of the things that we’re unable to.

Now I’m not saying that voters should let the computers decide this one – I’m saying that in order for voters to make a better, more informed choice, they should consult the computers and take their data into account. The computers were included in the BCS rankings to help us figure out situations exactly like this one – so use them!

Let’s look at some of the numbers.

Texas & Oklahoma schedule by Opponent Strength
Texas	Now	Then	Opponent	Record, Points	Now	Then	Oklahoma
			Texas	10-1, 401.27	#5	#4	L, 35-45 (neutral)
W, 45-35 (neutral)	#2	#1	Oklahoma	10-1, 385.41
L, 33-39 (road)	#8	#7	Texas Tech	10-1, 344.05	#2	#8	W, 65-21 (home)
			TCU	10-2, 275.14	#24	#14	W, 35-10 (home)
			Cincinnati	9-2, 255.20		#16	W, 52-26 (home)
W, 28-24 (home)	#12	#6	Oklahoma St	9-2, 243.75	#12	#18	11/29 (road)
W, 56-31 (home)	#11	#11	Missouri	9-2, 232.17
W, 52-10 (home)			Rice	8-3, 127.90
			Nebraska	7-4, 120.05			W, 62-28 (home)
W, 35-7 (road)			Kansas	6-5, 33.18	#16		W, 45-31 (home)
W, 38-14 (road)			Colorado	5-6, -36.62
W, 52-10 (homa)			Florida Atl	5-6, -51.98
W, 42-13 (road)			UTEP	5-6, -63.87
W, 52-10 (home)			Arkansas	4-7, -75.83
W, 45-21 (home)			Baylor	4-7, -87.77			W, 49-17 (road)
			Kansas St	5-7, -92.45			W, 58-35 (road)
11/27 (home)			Texas A&M	4-7, -134-51			W, 66-28 (road)
			Washington	0-11, -386.82			W, 58-35 (road)
			I-AA Chattanooga	0-2, -407.96			W, 57-2 (home)

Here’s the breakdown of Texas’ & Oklahoma’s opponents. As you can see, Oklahoma played more teams that were great – Texas, Texas Tech, TCU, and Cincinnati are all ranked and were harder than Texas’ hardest four. But Oklahoma also played Kansas State, Washington, and Chattanooga, which were easier than Texas’s easiest.

(Personally, I’m a much bigger believer in teams’ rankings at the end of the season rather than at the beginning or middle. Seems like common sense, yeah? Why should Alabama get credit for beating a top ten team (#9 Clemson) when the Tigers turned out to be barely above average this year, finishing at maybe 7-5 at best? Why should Oklahoma get credit for beating #16 Kansas when the Jayhawks aren’t in the Top 25 anymore? They should get more credit for beating the current #16, Cincinnati, if you ask me.)

Texas’ best road win was over 6-5 Kansas, while Oklahoma’s best was over 4-7 Baylor, so the Longhorns get the nod there. (But after this weekend, if the Sooners win they’ll get the nod for beating 9-3 Oklahoma State on the road.)

Overall, according to my SoS, Texas’ opponents have been more accomplished and tougher than Oklahoma’s – 915.88 points earned to 563.14. And that includes their opponents this weekend.

Let’s look at things a little differently, checking out how their offenses and defenses did against their opponents’ averages in terms of points and yards. In these next two tables, the first four columns of numbers are Texas’ and Oklahoma’s opponents’ averages against the other teams not on their schedule. For instance, Texas Tech is on both of their schedules, but the Red Raiders’ average defensive points per game (“AvgDpts”) is slightly different – 22.2 on Oklahoma’s slate and 25.4 on Texas’. That’s because the Red Raiders gave up 65 to Oklahoma and only 33 to Texas. So the way to read the table is, “Against their non-Oklahoma opponents, Texas Tech gave up an average of 22.2 points per game – Oklahoma scored 65 on them, a difference of 42.8 points (column 5) which was nearly 200% above the Red Raiders' average”. Make sense? The same thing goes with the defensive points and yards, but in this case negative is better. So looking at Oklahoma vs Texas we can read, “Against their non-Oklahoma opponents, Texas averaged 43.3 points per game – against Oklahoma they scored 45, which was 1.7 (or 3.9%) above their average”.

Points or yards shaded red were times that Oklahoma or Texas didn't score as many points or gain as many yards as their opponents' usually allowed (or if it's in a defensive column, times when their D's gave up more points or yards than their opponent usually scored or gained).

Oklahoma Offense & Defense against Opponents' Averages
Opponent	AvgDpts	AvgDyds	AvgOpts	AvgOyds	Opts	%	Oyds	%	Dpts	%	Dyds	%
Texas	17.9	339.9	43.3	474.3	17.1	95.5%	95.1	28.0%	1.7	3.9%	-36.3	-7.7%
Texas Tech	22.2	351.4	47.9	566.3	42.8	192.8%	273.6	77.9%	-26.9	-56.2%	-160.3	-28.3%
TCU	8.7	194.7	37.3	426.5	26.3	302.3%	241.3	123.9%	-27.3	-73.2%	-112.5	-26.4%
Cincinnati	19.6	298.6	27.0	381.9	32.4	165.3%	293.4	98.3%	-1.0	-3.7%	-55.9	-14.6%
Nebraska	25.7	353.9	36.6	467.4	36.3	141.2%	154.1	43.5%	-8.6	-23.5%	-49.4	-10.6%
Kansas	27.2	367.4	32.1	424.6	17.8	65.4%	306.6	83.5%	-1.1	-3.4%	66.4	15.6%
Baylor	26.8	375.9	29.1	392.8	22.2	82.8%	218.1	58.0%	-12.1	-41.6%	-123.8	-31.5%
Kansas St	33.8	474.6	34.9	388.6	24.2	71.6%	53.4	11.3%	0.1	0.3%	161.4	41.5%
Texas A&M	33.4	435.4	26.3	356.8	32.6	97.6%	217.6	50.0%	1.7	6.5%	-78.8	-22.1%
Washington	36.0	428.1	13.8	262.2	19.0	52.8%	162.9	38.1%	0.2	1.4%	73.8	28.1%
Chattanooga	46.0	494.0	7.0	164.0	11.0	0.0%	-7.0	0.0%	-5.0	0.0%	-128.0	0.0%
Oklahoma St	23.8	377.3	41.6	492.6	?	?	?	?	?	?	?	?
Totals	26.8	374.3	31.4	399.8	25.6	126.7%	182.6	61.2%	-7.1	-18.9%	-40.3	-5.6%

Texas Offense & Defense against Opponents' Averages
Opponent	AvgDpts	AvgDyds	AvgOpts	AvgOyds	Opts	%	Oyds	%	Dpts	%	Dyds	%
Oklahoma	21.2	342.4	54.4	568.7	23.8	112.3%	95.6	27.9%	-19.4	-35.7%	-133.7	-23.5%
Texas Tech	25.4	376.5	46.1	549.0	7.6	29.9%	-2.5	-0.7%	-7.1	-15.4%	30.0	5.5%
Oklahoma St	23.4	364.6	43.4	500.3	4.6	19.7%	139.4	38.2%	-19.4	-44.7%	-84.3	-16.8%
Missouri	20.0	372.9	47.2	528.7	36.0	180.0%	218.1	58.5%	-16.2	-34.3%	-180.7	-34.2%
Rice	32.5	436.7	43.3	475.8	19.5	60.0%	163.3	37.4%	-33.3	-76.9%	-157.8	-33.2%
Kansas	28.2	392.7	34.5	442.9	6.8	24.1%	28.3	7.2%	-27.5	-79.7%	-137.9	-31.1%
Colorado	27.0	374.1	19.7	326.5	11.0	40.7%	56.9	15.2%	-5.7	-28.9%	-60.5	-18.5%
Florida Atl	25.0	381.0	23.5	388.2	27.0	108.0%	122.0	32.0%	0.1	-57.4%	-96.2	-24.8%
UTEP	34.9	475.8	36.1	409.6	7.1	20.3%	-71.8	-15.1%	-23.1	-64.0%	2.4	0.6%
Arkansas	29.2	377.7	22.2	388.8	22.8	78.1%	43.3	11.5%	0.2	-55.0%	-197.8	-50.9%
Baylor	27.2	385.9	28.7	392.5	17.8	65.4%	108.1	28.0%	-7.7	-26.8%	-120.5	-30.7%
Texas A&M	36.4	455.2	26.5	349.6	?	?	?	?	?	?	?	?
Totals	27.5	394.6	35.5	443.4	16.7	67.1%	81.9	21.8%	-16.8	-47.2%	-103.4	-23.4%

Overall, Oklahoma’s offense was phenomenal, scoring an average of 126% more points and gaining 61% more yards than their opponents usually allowed. Texas’ offense, while great, was under that at 62% and 22%, respectively. But the Longhorn’s D was better overall, holding their opponents to 47% and 23% of their point and yardage averages. The Sooner D only held their opponents to 7% under their average points and 6% under their average yardage.

Even though these tables are number heavy, they don’t even scratch the surface of ways that computers can compare teams and statistics. I fully encourage everyone to check out some different computer rankings to see what they say and figure out why they rank the teams the way they do – it’ll probably help you understand more not only about what computer rankings do (and can do) but also about what you value in a team and season. (Provided you can put your subjectivity aside, of course.)

Kenneth Massey has a huge list of rankings, computer and other, at his site here - just click on each of the names to be taken to that rankings’ website. Browse around, see what other rankings are doing, and use them to figure out your own rankings. You might not be able to get this choice right, since there is no “right” answer between Texas and Oklahoma, but you certainly can feel better and more informed about the choice you do make.