14 March 2016

Extended bubble thoughts

Lunardi's apparently throwing a fit about Tulsa and San Diego St, and it appears the AD at St Bonaventure is throwing a fit as well.

Looking at these specific cases:

San Diego State of the 8-9 bubble teams had the worst case by far for inclusion. They have one quality win (Cal). That's it. The only other top 100 wins they have are Fresno (lost to them twice) and Boise St (split in conference). They did play an impressive non-conference slate (Kansas, West Virginia, Utah, UALR, plus the aforementioned Cal game), but they lost those games. Most of them were not close. UALR was at home. They also lost two more non-conference games to poor quality opponents (San Diego and Grand Canyon). The idea that they had a case was premised on one win and some decent computer rankings that the committee never looks at (they are a top 50 team in those). That's not a very good case.

St Bonaventure is a poster child for "why we should abandon the RPI". They lost both meaningful non-conference games they played; Syracuse and Hofstra, plus another loss to Siena for good measure. They have two wins over St Joe's and a split with Dayton as their best wins, and one win over George Washington. Somehow or another in the magical formula that is RPI, despite none of their other wins being over teams ranked comfortably in the top 100 in any other computer ranking I can find, this means they have 7 top 70 wins in the RPI. I count 4 good wins, and zero are over teams in the top 40 (Dayton is in the top 25 in RPI, but they're not in any other system). To add to this, they have random losses to LaSalle, Duquesne, and Davidson. They would be banking a lot on that win over UD and requiring us to overlook a bunch of bad losses. This is not a good case for inclusion. The only case for inclusion was an absurdly high RPI ranking.

Tulsa - Has a win over Wichita St. Has splits with Connecticut, Cincinnati, SMU, Houston, and Temple (all in conference), a random win over Iona, and losses to UALR, South Carolina, Oregon State, two (blowout) losses to Memphis, and one very curious loss to Oral Roberts. This is a much more impressive resume than SDSU and St Bonaventure based on these merits. It includes several wins over actual teams and only one strange loss. While I did not think they would necessarily merit inclusion, over say, Florida State, South Carolina, Valparaiso, or Monmouth, they were certainly an easier case than these two alternatives. Who had very weak cases in support.

The others that inspired some puzzlement. Were less puzzling for me.

Syracuse: Wins over Texas A&M, Duke, Connecticut, Notre Dame. 3 of which were at neutral sites or on the road. Secondary wins over Georgia Tech, Florida St (split), Virginia Tech, and NC State. All solid top 100 teams that did not get in the field. Only one baffling loss to St John's without Boeheim as coach. Every other loss was to a team in the top 70. Swept by Pitt, 3 times actually, and North Carolina. Virginia, Louisville, Miami, Clemson, Wisconsin, and somehow, Georgetown. Two of those losses were in overtime. They also beat St Bonaventure head-to-head by 13. Those 4 top 40 wins, 2 in the top 25 were easily enough to get them in. I don't even think this was a close decision. A better argument here would be that Lunardi has been slipping in not including them in the first place. Most every bubble conversation I heard over the last week was puzzled by his decision not to put them in.

Vanderbilt: A more interesting decision. Here's the case for them: Zero bad losses and they scheduled hard. The worst loss was to Mississippi State by 1 point on the road, another loss to Mississippi by 7 on the road. I assume Tennessee looks much worse on the RPI than it does on other rankings, but they only lost to them in the conference tournament (beat them twice earlier in the season). Losses out of conference to Kansas, Purdue, Baylor (2 points on the road), Texas, and Dayton (5 teams in the top 30). Swept Florida (underrated team), split with Kentucky (underrated team), Texas AM (not underrated), and also lost to South Carolina and LSU. Random win over Georgia. Then they absolutely clobbered every non top 100 team they played. Most wins were by 20 points or more. The problem is that only Kentucky and Texas AM got into the field from the list of teams they defeated. But Kentucky and Texas AM are top 4 seeded teams. St Bonaventure defeated a 7 and an 8 seed, by comparison. They were slotted as a play-in team. So I think this was a fair assessment.

I think there's an excellent case to be made that the committee struggles to place teams in a sensible manner on seed lines. There were numerous strange decisions this season (Oregon as a 1 seed, Michigan St not as a 1 seed, Kansas and Virginia's regional placement being weird, Kentucky below Texas AM, Arizona/Cal/Oregon St being in a very weird order, SFA and some other teams downstream being seeded oddly, etc), and this has been a long-running problem that largely stems from the futility of RPI as a sensible computer ranking system (it has Oregon #2, and Michigan State #11 I believe, as an example, essentially reversed from other rankings). But they actually do a pretty decent job of picking teams despite the flaw of using the RPI so extensively. There are usually no more than one or two very strange picks (UCLA was one last year, and they ended up in the sweet 16 anyway), and these are being pulled from a list of teams that all have pretty weak cases for inclusion. This is not their problem. They have a much harder job snaking teams into the 68 spots correctly.

No comments: