Seeding the Top Lines (2013 Edition)

Building on the previous post, I ran the same analysis for the schedule of each team currently vying for one of the top seeds in the tournament. This time I used Wisconsin, currently #8 in KenPom, as the standard.

After doing that, I compared the actual result of the game to the expected result, and either credited or debited the team in question based on that. So if Wisconsin would be 62% to win at Illinois, they get +0.38 "Net W" if they win, and -0.62 if they lose. Do that for every game, add them all up, and you get a number for the strength of each's team's résumé. Here's how they stack up:

First Tier:

Obviously this can change in a hurry, but for now the Bulldogs and Blue Devils have created a bit of separation from the rest of the pack.

For Gonzaga, the reason is simple: they just keep winning. Hate on the WCC all you want, but sweeping your way through the regular season in the 10th best conference in the nation is a notable accomplishment; there were about 1.9 expected losses in there by my methodology, so those 16 games account for the vast majority of the Zags' total "Net W". They played pretty much to expectation in non-con, losing a couple games they should have won but also beating Oklahoma, Kansas St. and Oklahoma St. away from McCarthey Athletic Center.

Lunardi doesn't even have Duke on the top line, which makes sense if you aggressively discount non-conference games and ignore Kelly's absence, but not so much otherwise. Duke had more expected losses in November than most of their peers had in all of non-conference play, a fact that becomes even more impressive when considering the swept through the entire Minnesota/VCU/Louisville/tOSU slate.

And not that they really need it, but Coach K's squad can also get a boost if you weigh the games with Kelly more heavily, as their "Net W%" is a ridiculous +0.178 in those contests (they're 17-0).

Second Tier:

Georgetown has a pretty similar profile to Duke (four losses, nearly identical SOS); the Hoyas are lagging a bit only because the Blue Devils have played three extra games (and won them all, obviously). They Hoyas have a chance to make up that ground this week, with a visit to Villanova and then the regular season finale against Syracuse at the Verizon Center. Easier said than done, though.

Michigan was covered earlier; only five losses against that schedule is very impressive. They would jump into the top tier themselves with wins in their last two (at Purdue, vs. Indiana).

Lunardi has Kansas at second overall on the S-Curve, which seems like a bit much; Palm has them as a #2 seed. The Jayhawks have done quite a bit of winning lately, but this year's Big 12 is pretty uninspiring. That loss to TCU is killer, of course; -0.97 "Net W" for just one game. Here are the 10 worst losses for the 12 teams featured in this post:

Other than the top two, none of those are really that horrific. I think we've found a weakness on Miami's résumé, though.

As you can see above, KenPom lists the location for the Indiana-Butler game as "Neutral"; you could reasonably switch that to "Semi-Away", although that only ups Indiana's "Net W" from 1.17 to 1.22. I think the Hoosiers are getting a bit too much respect in Bracketology right now, but that will probably sort itself out one way or another with the game in Ann Arbor on Sunday.

I don't think New Mexico (#25 in KenPom) is all that good, but other than the SDSU mishap they do win quite a bit. With a decent non-con slate and the strength of this year's MWC, I have their schedule as comparable to Miami, Pitt, and Kansas. We'll see if they can keep it going with two road games this week and then the conference tournament in Vegas, but their seeding could become very interesting on a few levels.

Not much to say about Louisville; they probably need to at least get to the final at MSG to have a shot at a #1 seed, which seems reasonable.

Third Tier:

I already discussed my issues with Miami's schedule a couple weeks ago; the other three teams here are at about the same spot on Lunardi's S-Curve.