Pages

The Bracket - Initial Thoughts

3/16/2008
Here is my final, data-driven bracket projections taking into consideration all of the conference championship games (my 'Final bracket' that I submitted was an average of two different outcomes and assumed Arkansas beat Georgia.

1: North Carolina, UCLA, Memphis, Tennessee
2: Kansas, Duke, Wisconsin, Texas
3: Georgetown, Stanford, Drake, Washington St
4: Butler, Pittsburgh, Louisville, Xavier
5: Southern California, Clemson, Connecticut, Marquette
6: Michigan St, Notre Dame, Indiana, Kent St
7: Vanderbilt, Purdue, West Virginia, Oklahoma
8: Texas A&M, Arizona, Gonzaga, Miami (FL)
9: BYU, Davidson, Arizona St, Baylor
10: Oregon, Illinois St, St. Mary's, Villanova
11: Arkansas, Kansas St, Florida St, UNLV
12: Ohio State, Temple, Western Kentucky, Oral Roberts
13: Cornell, George Mason, Siena, Georgia
14: CS Fullerton, Boise St, San Diego, Belmont
15: UMBC, Winthrop, Portland St, American
16: Austin Peay, UT Arlington, Mt St. Mary's, Coppin St, Miss Valley St

Last 4 In: Arkansas, Kansas St, Florida St, Ohio State
First 4 Out: Syracuse, Mississippi St, Saint Joseph's, Massachusetts
Next 4 Out: Dayton, Creighton, Virginia Tech, Mississippi

This one probably performed worst than my previous version since it drops Mississippi St. I had 50 teams within 1 seedline of actual, which I'm sure will be the worst of all bracketologists. Most of my differences were not unexpected, I think there's a good chance that a lot got all of the teams and nearly all of the seeds right. My hats off to you for predicting what the Committee would do.

Of course, that's not what I'm about. I'm all about exposing the weaknesses in the RPI and highlighting the teams that deviated most from the 'body-of-work' approach that the Selection Committee continues to say they judge teams on. Overall, I think the Selection Committee did a good job with the information they had in front of them, especially given some of the unique situations that went down this week. However, I think that misleading information may have skewed their decision making, and ultimately caused less deserving teams in the tournament at the expense of more deserving teams. Obviously, this wasn't the intent of the Selection Committe, they are hard working, well-intentioned people that take this task seriously. I only blame the NCAA for continuing to provide the Selection Committee with bad information that makes already a tough job that much tougher, especially when these problems would be easily fixed.

I will likely dive deeper into some of these issues later this week, but here's the list of winners and losers.

Losers - Teams Lower Than They Should've Been

ProjAct RANKSOS NC SOS
SeedSeedTEAM WLJCIRPI JCIRPIJCIRPI QWQL
3 5 Drake 26 4 11 10 82 68 114 126 15 15
4 7 Butler 29 3 12 17 127 128 78 109 21 7
6 8 Indiana 25 7 23 22 66 57 150 71 22 20
6 9 Kent St 28 6 25 21 128 111 96 56 27 30
8 10 Arizona 18 14 33 37 1 2 19 5 23 48
9 11 Baylor 20 10 36 43 47 48 213 130 38 40
10 12 Villanova 20 12 42 51 37 47 228 157 34 59
9 Arizona St 19 12 37 83 17 77 305 293 39 42
10 Illinois St 23 9 39 33 84 72 184 115 42 38
12 Ohio State 19 13 43 49 24 17 25 12 51 43
11 Florida St 19 14 47 59 23 14 197 128 36 65

The average Colton Index for these teams is 32 versus an average RPI of 39. But the RPI's aren't deflated across the board. There seems to be a bias against the mid-majors here...an assumption that teams such as Drake, Butler and Kent State (and Illinois St by association) aren't as good as their records or RPI's indicate, or put in other words, they didn't play anybody therefore have inflated records. Well, each of those mid-majors have profiles of wins and losses factoring in weaker schedules that still put them as deserving higher seeds, so penalizing them further would be like giving them losses for games they either won or didn't play. Drake had a 26-4 record with an RPI SOS of 68. Notre Dame had a weaker schedule (according to the RPI) and went 24-7 with a RPI SOS of 81, and had the same seed. But Kansas has a RPI SOS of 60 and got a 1 seed with a 29-3 record. There seems to be a disconnect.

I want to dive into the issue deeper later this week, but Arizona St simply flat out got screwed of the tournament this year. Of course, they didn't do themselves any favors with their non-conference scheduling. I can't argue that their non-conference wasn't weak, I had it even easier than the RPI (305 vs 293). However, the difference comes with how much this weighs on the final decision. I consider it as part of the body of work and Arizona St still rans 37th overall even with that very weak schedule (the cupcake wins don't tell us much about how good Arizona St really is, therefore don't help their profile much at all). In the RPI, these games drag down the team's overall RPI and SOS to the point like they barely look like a bubble team at all. (Plus, if the Committee is going to list that as a reason you didn't get picked, then they are already overweighting that factor more than it should be. Of course, Stanford and Notre Dame had equally weak non-conference SOS's and it didn't hurt them at all. Notre Dame even got seeded a little higher than expected.) Even if you win those games against the cupcakes, just showing up drags your RPI down. The seemingly harmless decision of playing Florida Gulf Coast, Idaho and St Francis (PA) is what kept them out of the tournament, not what they did or didn't do during the other 29 games of the season. If they had played Vermont, Wichita St and Navy instead (games they still would've won), they'd be playing in the NCAA instead of the NIT right now.

I could (and may) write a whole post about the plight of Florida St. This is the third year in a row that I felt they deserved to get in but didn't. They are the quinticessential bubble team. But every year they have a deflated RPI that's right in the 40-60 sweet spot, an undesirable position to be in. Check out their numbers the last three years.

2007-08: JCI 47, RPI 60
2006-07 JCI 33, RPI 40
2005-06 JCI 37, RPI 63

You gotta feel for these guys. Every year there are going to be winners and losers in this process, but imagine if it was your team that was on the outside looking in for three years in a row. And based on e-mails I've received from Seminole fans, it seems like they've resigned themselves to the fact that it will keep on happening. Florida St gets so little respect that they weren't even listed among the 8-9 teams more 'most surprising team left out' fan poll.

Winners - Teams Higher Than They Should've Been

ProjAct RANKSOS NC SOS
SeedSeedTEAM WLJCIRPI JCIRPIJCIRPI QWQL
7 4 Vanderbilt 26 7 24 12 91 40 277 93 30 25
11 8 UNLV 25 7 40 24 152 63 221 60 62 34
8 Mississippi St 22 10 45 40 90 58 217 124 50 51
10 South Alabama 24 6 53 38 214 127 127 69 83 35
11 Saint Joseph's 21 12 54 45 73 51 135 116 29 99
11 Kentucky 18 12 60 58 58 21 129 98 55 63
15 13 Winthrop 20 11 128 107 211 178 56 38 112 123

Here's where you can really see a pattern. The average team listed here has a RPI of 46 and average Colton Index of 58, and average RPI SOS of 77 vs 127, and an average RPI NC SOS of 85 vs 166. Remember those differences are based purely on the nuance (ie flaws) of the RPI formula. It's clear that the RPI has a role in the process. It's almost unfathomable that the NCAA would rely on SOS numbers that may be off by 90 or more (look at Vandy's NC SOS of 93 vs 277...yikes!) slots for any given team. How does the Committee make any informed decisions using inputs with margins of error that large? Certainly, it skews the profiles of Vandy, UNLV, Mississippi St and South Alabama in particular.

The SEC was the main beneficiary of the seeding process, as the Committee must've felt that Mississippi St and Kentucky were better than their records and profiles indicated (or maybe it was just the Silve factor). The only way to determine strength of a conference is to view how the perform against non-conference foes. Unfortunately, nearly all of this occurs in the first half of the season. The rest of the season is just teams beating up on each other in conference. The SEC doesn't fare too well against other conferences, mainly because Kentucky, Florida and Missippi St performed poorly in non-conference play. It drags down the conference looks overall. Maybe it's true that those teams improved as the season went on, but is it 'body-of-work' or not? I even factored in last 10 games and Kentucky was still way down the list. By discounting their non-conference season, you might as well be giving Kentucky wins for games that they lost.

THIRTEEN TEAMS MORE DESERVING THAN SOUTH ALABAMA

South Alabama is the anti-Arizona St. The danger of being a bubble team with a deflated RPI is that there's going to be less deserving bubble teams with inflated RPI's of the same amount or more. South Alabama's profile and tourney-worthiness falls apart if their RPI were 53 instead of 38 and SOS was 214 instead of 127. I have them 83rd in quality wins (a summation of a team's wins scaled by difficulty), the next closest at-large is BYU at 56th. And considering they got a 10 seed, it looks like they got in the tournament by a pretty safe margin.

ProjAct RANKSOS NC SOS
SeedSeedTEAM WLJCIRPI JCIRPIJCIRPI QWQL
9 Arizona St 19 12 37 83 17 77 305 293 39 42
10 Illinois St 23 9 39 33 84 72 184 115 42 38
11 Florida St 19 14 47 59 23 14 197 128 36 65
12 Ohio State 19 13 43 49 24 17 25 12 51 43
Syracuse 19 13 51 55 36 9 195 20 35 72
Massachusetts 21 10 50 42 93 71 138 88 43 61
Dayton 21 10 49 32 92 33 147 26 41 64
Creighton 20 10 46 46 76 65 177 91 63 45
Virginia Tech 19 13 52 52 38 38 176 135 53 55
Mississippi 21 10 55 48 105 67 269 122 47 67
VCU 24 7 57 54 191 161 82 85 60 56
Nebraska 19 12 56 97 57 94 281 310 69 52
11 Kentucky 18 12 60 58 58 21 129 98 55 63
California 16 15 58 92 5 26 291 175 52 62
10 South Alabama 24 6 53 38 214 127 127 69 83 35

0 comments:

 
Wegoblogger #31 © 2011 | Designed by Bingo Cash, in collaboration with Modern Warfare 3, VPS Hosting and Compare Web Hosting