Tennessee is going to beat up the Wolverines. First off, let’s just appreciate the world of sports that makes that sentence coherent. But even in the context of college basketball and March Madness, it still doesn’t make total sense.
Regardless of who moves on to the next round, the Tennessee Volunteers are going to win the physical battle against Michigan when they meet in the Sweet 16. Jarnell Stokes is a 6’8”, 260-pound monster inside for the Vols, and he is wreaking havoc on opponents to the tune of 20 points and 15 rebounds per game so far in the NCAA Tournament. He’s just one of the bullies in a strong-as-hell Tennessee frontcourt, and they’ll pound the ball inside at every opportunity against the diminished corps of Michigan bigs.
Given that description, you wouldn’t think Michigan is the 2 seed and Tennessee an 11. But the regular-season Big 10 champs, while athletic in their own right, have tied their fate to three-point shooting and ball movement, whereas Tennessee, which finished 20-12 and plays in the middling SEC, had to win a play-in game (no matter what, I won’t call that the first round) just to secure a spot in the round of 64. Not that Tennessee is a plucky mid-major school just making its name on a national scale, but it’s rare that a team seeded nine spots lower will enter a second-weekend game looking to physically overwhelm the favorite.
There’s a simple explanation to all this: though they overcame their seeding, the Vols got shafted by the tournament selection committee. The 12 losses aren’t pretty, but that raw figure underrates Tennessee’s actual ability. Per ESPN’s Basketball Power Index (BPI), which “accounts for the final score, pace of play, site, strength of opponent and absence of key players in every Division I men’s game,” the Vols are the 25th-best team in college basketball; that they were one of the last teams into the tourney isn’t a sign of their weakness, but rather a blatant oversight of their strength.
So what happened? Well, let’s tweak the acronym a bit. BPI, designed to provide a predictive rating of a team’s ability to win, is just a three-year-old metric. The Ratings Percentage Index (RPI), on the other hand, is more interested in traditional win-loss record and overall strength and schedule rather than the minutiae factored into BPI; Tennessee was ranked 41st in RPI. I’d write some snark about how much better BPI is than RPI, but I’m pretty sure my B+ in Elementary Statistics was inflated, and I’m two years removed from that class. So let’s kick it over to someone much more qualified than I, Nate Silver of FiveThirtyEight.com fame, to burn RPI.
“The system, developed in 1981 in the era of the DOS prompt and the Commodore 64, tends to give computer rankings and other objective attempts at analysis a bad name,” Silver wrote in The New York Times. “Over the long run, RPI has predicted the outcome of NCAA games more poorly than almost any other system.”
RPI and the selection committee underestimated Tennessee, and it beat the tougher-than-deserved odds to make it this far. As the sports world grows more comfortable with more sophisticated analytics, future seedings should more accurately represent each competitor’s relative ability.
That would be good news for misunderstood teams like Tennessee, but I can’t shake the feeling it wouldn’t necessarily be so for fans.
My stance is rife with internal conflicts: I’m not advocating the preservation of RPI or a Neanderthal opposition of more finely-tuned advanced statistics, but I also fear that the more precise the bracket gets, the fewer upsets there will be. Of course, the flaws in the current system mean that an 11 seed like Tennessee was likely straight-up better than the 6th-seeded Massachusetts squad, which the Vols drubbed 86-67. That kind of mis-seeding defeats the ideal of the true underdog winning, at which point a statistical conversation enters the realm of philosophy and we lose all hope of divining a comprehensible thesis.
I’m just going to chalk that all up to the spirit of March Madness; we crave the moments when the action strays into the irrational, so let’s just count this as one of them.
Let’s get back to UMass vs. Vols for a second. The Minutemen, ranked 21st in RPI and 36th in BPI, were picked over Tennessee by 60 percent of participants who filled out brackets in Yahoo! Fantasy Tourney Pick’em. While that still qualifies Tennessee as a popular upset pick, those who believed the Vols were empirically better were still in the minority by a significant margin.
For the sake of argument, let’s assume BPI is right and Tennessee was the superior team heading into the matchup. My question is, if that more advanced system is, in fact, more accurate, then what would it have said about past upsets? How many teams that we perceived to be Davids that downed Goliaths would we have seen otherwise with more information at our disposal?
As we try to understand sports even more fully, March Madness resists. As much as we want to be absolutely right and get some of that sweet, sweet Warren Buffet money, the tournament is only so exciting because we’re always so wrong. I don’t totally believe that something ineffable about tournament play leads to significantly more outcomes we’d consider upsets. When it comes to evaluating teams, I think our ignorance brings about the bliss.
Cohen is a member of the class of 2014.