“Analytics” and the Final Four 

If you’re involved with college football handicapping, or “analytics” in general, you were probably very frustrated this past weekend by the mainstream media discussions of the selection process for the national championship tournament.

It wasn’t anywhere near a complete discussion. The networks generally had pundits who each voiced their opinion…and they kept repeating variations of that same opinion over and over again whether it was a Thursday night halftime show…and Friday night pregame show…a Saturday highlight show…or a Sunday morning chat show. Same guys saying the same things.

Yet, coverage somehow managed to leave out the fact that multiple fairly well known “analytics” sites had Florida State nowhere near the Final Four. Now, I’m not saying Florida State didn’t deserve an invitation just because computers weren’t fond of them. The system has preferences in place, and it’s very hard to argue against a “Power Five” team that won all of its games if “winning games” is a priority (which it obviously is). Create parameters, then follow the parameters.

But, how could the discussions use the term “analytics” without noting how badly Florida State was punished by “analytics” methodologies for their season-long mediocre play? FSU did go undefeated, barely beating a lot of “okay” but not great teams.

Guys in suits were saying “analytics may say this, but I think that.” Or, “the committee has access to all the analytics,” and so on. At no point was a composite of systematic analytic approaches summarized, at least in the coverage I personally witnessed.

So, let’s do that here with a cross-section of identities and websites that have been prominently featured in the media whenever “computer” analysis is the topic. 

Jeff Sagarin of USA Today

Football Outsiders and their F+ rankings

TeamRankings “Predictive” rankings

PredictMachine

The PowerRank from Ed Feng

Those have all been quoted at one time or another in the past online by ESPN, Grantland, or obviously USA Today. Here are the end-of-the-season rankings from those five entities that were posted this past weekend. I’ll present the rankings in the order listed above…the first number you see by each team is from Sagarin, and so on.

Alabama: 1-1-2-1-2 (average 1.4, median 1)

Oregon: 3-3-1-2-1 (average 2.0, median 2)

TCU: 2-5-3-3-3 (average 3.2, median 3)

Mississippi: 4-4-7-4-5 (average 4.8, median 4)

Ohio State: 5-2-5-5-9 (average 5.2, median 5)

Baylor: 7-9-4-6-4 (average 6.0, median 6)

Mississippi State: 6-6-11-8-6 (average 7.4, median 6)

Georgia: 8-7-6-9-7 (average 7.4 median 7)

That works out pretty well to give you a consensus “best eight” The medians are a stepladder until Baylor and Mississippi State tie at six. Can’t imagine a more fair representation of how “analytics” saw the top teams in the country. They don’t agree with each other exactly. The most extreme differential on the list is F+ having Ohio State at #2 while Feng has them at #9. But, if you let the medians do the work, the picture becomes clear.

Florida State isn’t in that picture! 

Florida State: 15-8-16-14-17 (average 14.0, median 15)

If “computers” relying on analytical algorithms were tasked to evaluate teams based on their true quality…to find “the best four teams” for a tournament…FSU would have been nowhere near the brackets. In fact, if “computers” relying on analytical algorithms were asked to build an eight-team tournament…Florida State wouldn’t have made that either! Only Football Outsiders has them in the top eight…a clear outlier from the others. The Seminoles do make a Sweet 16 tournament. Barely!

And, if you include the betting markets in the process, because pointspreads are heavily influenced by advanced analytics (particularly from the “sharps” who shape the openers into better positions), the same story would be told. The quickest way to show that:

Florida State was -3.5 on a neutral field over Georgia Tech in the ACC Championship

Mississippi State is -7 on a neutral field over Georgia Tech in the Orange Bowl

Mississippi State barely cracks the top eight nationally…but is priced about a field goal better than Florida State in Georgia Tech’s back-to-back neutral field comparison.

The computers should have been in the discussions if the word “analytics” was going to be thrown around the way it was. Florida State’s won-lost record trumped a computer consensus that they were way off the pace in terms of performance level. The Seminoles can certainly lift that level from this point forward. They’ll need to in the Rose Bowl as 9-point underdogs to Oregon.

Some misleading scores from last Saturday…

(Note, I’ll only mention games involving teams who will be playing again in the postseason.)

• If you went to bed when Boise State led Fresno State 21-0 at the half of the Mountain West Championship, or 28-0 in the third quarter, you may have a false read on that game. The final score was just 28-14. Boise’s second touchdown came on an interception return. So, the offense only scored 21 points in a game the big favorite was expected to win by 22! Fresno State won total yardage 402-304, and third downs 48% to 42%. If Fresno State hadn’t lost turnovers 3-0, missed a field goal, and failed on two fourth down attempts (virtual turnovers)…we may have had a nailbiter finish. Worth remembering when Boise State steps way up in class to face Arizona in the Fiesta Bowl. Fresno State will be in friendlier offensive conditions when they face Rice in the Hawaii Bowl.

• Marshall was a bit more impressive in their CUSA Championship win than the 26-23 final score made it sound. The Thundering Herd won total yardage 429-268. The defense in particular graded out well, holding Louisiana Tech quarterback Cody Sokol to a 7-20-1-70 passing line, and the Bulldogs as a team to 4 of 17 on third down tries. Tech’s touchdowns came on short fields after Marshall turnovers. A disappointing end to the regular season for Marshall…but they will still likely have an edge in weaponry against Northern Illinois in the Boca Raton Bowl.

• Cincinnati beat Houston 38-31, but was outgained 594-500, outrushed 234-152, and lost yards-per-play 7.4 to 6.2. Houston had a shot to win, but lost the turnover category 1-0, and was 1 of 3 on fourth down attempts. At least Cincinnati will be facing a poor offense from Virginia Tech in the Military Bowl. Houston will have home-state advantage in the Armed Forces Bowl in Forth Worth against Pittsburgh.

See you again next Wednesday as we transition to bowl coverage.

---

Jeff Fogle is a freelance writer living in Austin, Texas. He writes about college and pro football, college and pro basketball, and MLB on his blog StatIntelligence. You can follow Jeff on Twitter @JeffFogle.

---