DPPI and Seeding

Based on DPPI, Michigan St., Purdue, Kentucky and KU should have been 1 seeds.

Teams that got screwed
1. St. Mary's -- Should have been an 8 seed but did not make the tourney. The composite of 60+ computers also say an 8 seed. Ridiculous.
2. South Carolina -- DPPI says 9 seed, computers say 11 seed. They were not picked at all.
3. Wichita St. - They should have been a 3 seed but got a play in on the 11 seed line. That is a head scratcher. Even on the composite of 60 computers they should be a solid 7 seed. Totally screwed.
4. Valpo -- Both computers and DPPI says 11 seed. Not picked
5. Florida -- Both computers and DPPI says 11 seed. Not picked
6. Purdue Got a 5 seed and should have been a 1 seed. Based on the composite computers should have been a solid 3 seed.
7. Kentucky -- Got a 4 seed and should have been a 1 seed. Based on computer composite should have been a 3 seed.
8. Michigan St. -- Based on DPPI the strongest team in the country did not get a 1 seed. Based on the computer composite they are the No. 2 team in the country.
9. UConn -- Based on DPPI should have been a 4 seed but got a 9 seed (In KU's bracket). Based on computer composite should have been an 8 seed.
10. VCU -- Should have been a 7th seed but got a 10 seed. Computers say 9 seed.
11. Indiana -- Should have been a 2 seed but got a 5 seed. Computers say a 4 seed

Teams that benefitted
1. Tulsa -- Should not have come close to getting in. Made play in at 11 seed. Ridiculous!!
2. Texas -- DPPI says 12 seed but they got a 5 seed. Computers say 7 seed.
3. Dayton -- Should have been an 11 seed but got a 7 seed. Computers say 9 seed.
4. USC -- Should have been 12 seed or possibly not made the tourney. Made an 8 seed. Computers say 11 seed.
5. Texas Tech - Should have been a 12 seed or possibly not made the tourney. Made 8 seed. Computers say 11 seed.
6. Colorado -- Should have been 12 seed or realistically not made the tourney. Made 8 seed. Computers say 11 seed.
7. Providence - Should not be in the tourney/ Got a 9 seed. Can anyone say 1st round loss.
8. Oregon - A DPPI 4 seed made the top line. Computers say should be the last 2 seed.
9. Michigan -- No way should be in the tourney. No way.

Overall, it clearly looks like the committee rewarded the Big 12 and Pac 12 and punished the SEC.

Thanks. Definitely a lot of strange entries/or not. Also seedings. What about the travesty of Syracuse getting in? Also Cal a 4 seed and Arizona a 6, when they finished tied in Pac 12 regular season and both got to semis in their tourney.
Agree about Wichita State - crazy, but HCGM won't have to drum up any reason to fuel the Play Angry mantra.

A valid argument that I've heard many of the more respected commentators make (Bilas, several coaches, etc) is that it would be completely possible to pick a computer ranking (or combination thereof) and allow a 100% computer generated bracket. It's totally possible and frankly any of us could probably come up with a fair set of standards. But what ALL of these guys are agreeing on is that, in the end, the "eye test" needs to be part of the process.

Not knocking your analysis in any way, but just pointing out that DPPI is only one data point in a process that takes in a large number of varying (and some subjective) data points.

What exactly is the "eye test"? Would hazard a guess that no two sets of eyes would see things the same way.

The Big 12 was truly a gauntlet this year. Going in, it could have easily come down to a 3 way tie for the crown at 13 wins, with a bottleneck at 12 wins. Winning it with breathing room to spare was a big accomplishment. Going 2-0 vs. OU, 2-1 vs. WVU, and 2-0 vs. Texas and Baylor is a big deal.

Our record vs. the 6 other teams who made the tourney: 11-2.

This was supposed to be in response to KU's surge to the top of the defensive ratings post. Oh well. Technology.

The eye test is the totally subjective opinion of the people on the selection committee. So if Michigan State is arguably a borderline 1 or 2, the deciding decision might come down to one vote by a guy who likes green uniforms.

I'm just commenting on the fact that it would be very easy to come up with a weighted scoring system (x% RPI, x% KenPom, x% BPI or whatever) and come out with an "official" computer ranking. That's really what the "secret" RPI started as. With that, they could then combine the conference champions into the list, rank all 68 teams accordingly and then do a strict rankings-based geographic selection. So the #1 overall seed gets sent to the closest region...period. Then the second #1 get the closest of what's left. Same for the third and fourth #1s. Then repeat the process for the 2's, the 3's, etc. etc.

With that, we really wouldn't even need the committee and they could cancel the selection show. The reason they still have the committee, however, is to add some of the intangible elements back in: i.e. who's hot right now, is Michigan State WITH Valentine the same team as Michigan State WITHOUT him, does the geographic alignment jibe with trying to give the highest seeds the easiest road?

In the end, I can make an argument for or against allowing the subjectivity into the process. In the end, even a totally automated system isn't guaranteed to give a better result.

When you look at all the computer rankings out there you get a very good cross-section of analyses from a number of different angles. If I were king I would generate the bracket from the composite of the top 60 or so computer systems that have different algorithms for different emphases. Then I would let the committee make arguments why it might need to be changed. I place a lot of credibility on the computers not because I am baised toward them but because I understand the nature of the programs and what they entail. The Bayesian formulas are designed with a glint to the eye test of the designer. So in reality, each computer program has some sort of eye test built into the code. If you go to Massey and look you can make a legitimate argument for 5 teams to be a No. 1 seed. Oregon is not one of them. Of the 54 computers currently in the composite, 10 have Oregon as 4th or better with a mean rank of 8.81. Compare that to Kansas (53 / 1.33 ), Michigan St. (39/3.94), North Carolina (374.01), Villanova (32/4.28) and Virginia (29/4.76). You can design systems to incorporate many "eye test" baises. You can weight however you like.

If I were king I would generate the bracket from the composite of the top 60 or so computer systems that have different algorithms for different emphases. Then I would let the committee make arguments why it might need to be changed.

I'm guessing that this isn't too far from what they actually do. Rather than taking all 60 systems and combining them, however, I think they probably look at 4 or 5 and don't actually compile it into one list.

I'd also think that of the 60 different programs, there are going to e versions that are similar enough to others to be eliminated. So really the question becomes is the "average of all 60" better than just one really well designed program, or is the best combined poll a weighted average of 5 or 6 of them?

Like I said earlier, whether to leave a subjective element involved or not is a legitimate question (or to what extend subjective measures are weighed). In the end, I think that no matter what method you use, people are going to gripe, and no method is necessarily "better" than another...you'll just get different outcomes.

In the end, I think the system works fairly well, or at least "good enough". Every year you're going to have discussions about who should have made the field, or who should have gotten a different seed or geography. At the same time, the system you propose would be just as good, and there would still be arguments and disagreements.

Seems the missing link (which we'll never get under this system) is having a press conference with the committee members and being able ask them about their decision making process. At least that might give them something relevant to discuss in the now 2-hour selection show.

Hopefully the ridiculous 2 hour selection show will go back to 1 hour or less.

Agreed. The reviews have all been terrible. Obviously the NCAA banked on making 2x as much money by going to a 2 hour show. Nevertheless it's so contrived that it's stupid. Hell, they could do it in half an hour and it would be fine. I ended up tuning out the 2nd hour and had dinner.

If they insist on dragging it out to 2 hours, then I'd hope that they'd at least announce when each "announcement" was being made. So at 5:05 they'll reveal the #1 seeds. At 5:15, they'll reveal the Midwest. Banter for half an hour and then at 5:45 they could do the West. If they were smart, they'd couple it online with an app that "revealed" the bracket, linked up to the CBS bracket game so that you could start monkeying with your picks straight away. At least by announcing times, you'd know that if you didn't want to listen to half an hour of banter, you could flip to golf and come back for the next scheduled announcement.

In it's current form, I think a lot of people bailed on it. If they insist on 2 hours, then I'd recommend what I suggested. Still, the best thing would be to just cut it to an hour (or even 30 minutes) and get rid of all of the chatter, maybe running an hour long post-bracket analysis show (that I wouldn't watch anyway).

Whoever the producer of that show was should be fired. It was quite literally one of the worst produced shows in the history of sports broadcasting. It was boring. It lacked any useful or insightful information. It was boring. It was excruciating to watch. It was boring. It was insipid. It looked forced and ad hoc. It was boring. Did I mention it was boring? One could surmise that it was thrown together over a day or two. I feel completely disrespected as a fan that anyone could think that was worthy of my time or that I would sit and watch that pap and feel edified or satisfied. I have had to lay off 75 people in the last year and I guarantee most of them could have done a better job than the producer of that show.