Don’t trust the computers on Ohio State in the NCAA tournament: WFNY Stats & Info

I can’t ever recall a dichotomy quite like this. The base support for this year’s Ohio State men’s basketball team has been lukewarm at best. A team some thought to be a top-20 contender languished for most of the season, stumbled into the NCAA tournament, and landed a No. 10 spot opposite 7-seed Virginia Commonwealth. If it weren’t for D’Angelo Russell highlights, hardly anyone would care about this squad. Despite this rare fan apathy, the March Madness computers actually seem to LOVE the Buckeyes.

A quick roundup of some rankings and probabilities: Ken Pomeroy has the Buckeyes as the No. 21 team in the nation. He has them beating the Rams 56 percent of the time. The projections at FiveThirtyEight are even more optimistic at 64 percent. Sports analyst Christopher Long takes it up a notch higher at 71.2 percent. The ESPN Basketball Power Index also has OSU at No. 17 overall. NumberFire says they’re No. 15. The Washington Post’s Fancy Stats blog pegged them as a tournament sleeper two weeks ago. So what gives? Why do the bracketology stats love this mediocre team so gosh darn much?

Ohio State had practically no bad losses …

Yes, it was a relatively down year for the Big Ten. It would be wild to find the last time the conference placed only two 1-through-6-seed teams in the tournament. Ohio State’s 23-10 record and 11-7 conference mark doesn’t seem to be too inspiring at first, given a somewhat middling conference landscape. But dig just slightly deeper, and you can see there isn’t an outlandish defeat on their schedule. This matters when analyzing teams numerically.

The worst three losses by far on paper: at Indiana (10-seed, No. 53 in Pomeroy, No. 46 in BPI) on Jan. 10; at Purdue (9-seed, No. 49 in Pomeroy, No. 39 in BPI) on Feb. 4; and at Michigan on Feb. 22 (no postseason, No. 76 in Pomeroy, No. 74 in BPI). All three were on the road and Michigan was OSU’s only loss to a non-NCAA tournament team. Not many major conference teams pulled that off¹. The only two non-conference losses were to national names Louisville and North Carolina.

… and more importantly, all the losses were by super close final margins.

This main point can lead back to a sticking point with some of the advanced metric rankings out there. Largely, they’ll use some type of margin of victory input. This is imperfect as blowouts can eventually turn into misleading final results, one way or another. The rankings are then stuck using this final result to measure offensive efficiency, defensive efficiency and a net rating. Sure, margin of victory is often a better indicator of future results than win-loss record … but there can be oddities, especially within a sample of only 33 games.

In the 10 Ohio State losses, the average margin was only -8.0. That’s outrageously small already. Yet the only double-digit loss was the 24-point thumping by 1-seed Wisconsin in the regular season finale. The other nine losses averaged out to only -6.2 points. And again, all but one of these losses were to tournament teams. That’s a stunningly close average loss and this one stat shows why a team could jump up way higher in advanced rankings compared to the RPI (where OSU finished No. 40). Thus, TeamRankings had the Buckeyes as the “unluckiest” team in the tournament. Sorry, but I call B.S.

Want a few pivotal examples of slightly misleading final margins? They trailed by 19 to Louisville on Dec. 2 (lost by nine), trailed by 18 to UNC on Dec. 20 (lost by eight), trailed by 17 to Iowa on Jan. 17 (lost by nine) and trailed by 17 to Michigan State in the Big Ten Tournament (lost by nine). All four of these games were clearly out of hand in the second half before a totally meaningless comeback. The final scores made the Bucks look just a tad bit better.

Does this team have some magical ability to come back in blowouts? Not a chance. So this is why, as a stats-loving Ohio State fan, I’m actually not that high on these Buckeyes. Yes, I see that Vegas now has them as the slight favorites and over 57 percent of ESPN bracket-pickers are choosing them to pull off the upset. But I won’t be picking the Bucks myself². And this is a classic case of the paradox of how “great” losses can carry a great deal of weight.

1. Then, note the paradox: What was Ohio State’s best win this season? A 24-point home win over overrated Maryland? Home wins over Indiana/Purdue? Road/neutral wins over Minnesota? Yikes, that’s pretty poor for a “top-20” computer team. [↩]
2. Another reason for pessimism: VCU played the No. 13 toughest non-conference schedule, with wins over Oregon, Northern Iowa and Cincinnati. OSU played the No. 329 (out of 351) toughest non-conference schedule, with a best win over a terrible Marquette team. [↩]