Dork Brackets

[Wander]

The computers like them because beating teams by large margins tends to be a fairly good predictor of future performance. That said, even though I am a huge "dork stat" guy, I think the committee should rightfully ignore a team's Pomeroy/Sagarin/etc. ratings (although I would prefer they use them to assess the difficulty of a team's schedule vs. the RPI). Why? Because sports are more interesting when winning and losing actually matters. Sure, deep down Ennis hitting a half-court shot against Pitt really means very little in terms of a team's actual basketball skill, but do we really want to live in a world where that barely matters (as it doesn't in Pomeroy)? Awfully boring.

Absolutely. Tennessee and Louisville provide pretty good evidence that the committee doesn't strongly consider metrics that take into account margin of victory, and as much as I love kenpom, that's the way it should be. I don't get the complaints about the inconsistency of the committee - most everyone who did mock brackets knew Louisville didn't have a realistic chance to be a 1 seed.

[Kedsy]

Every year, one or two (or three) top ten Pomeroy teams are in the situation where their seed is disproportionately worse than their Pomeroy ranking would suggest. This year it's Louisville -- #2 in Pomeroy with a 4-seed. Here are some others from the past five years:

2013:
Florida, #1 Pomeroy; 3-seed; made Elite Eight;
Pitt, #7 Pomeroy; 8-seed; lost in round of 64;

2012:
Wisconsin, #6 Pomeroy; 4-seed; lost in Sweet 16;
Memphis, #9 Pomeroy; 8-seed; lost in round of 64;

2011:
Texas, #4 Pomeroy; 4-seed; lost in round of 32;

2010:
Wisconsin, #3 Pomeroy; 4-seed; lost in round of 32;
BYU, #7 Pomeroy; 7-seed; lost in round of 32;

2009:
Memphis, #1 Pomeroy; 2-seed; lost in Sweet 16;
Gonzaga, #5 Pomeroy; 4-seed; lost in Sweet 16;
West Virginia, #8 Pomeroy, 6-seed; lost in round of 64;

I don't know if it's because the teams were overrated in Pomeroy, or because seeding matters a lot more than Ken Pomeroy and others seem to believe, or just dumb luck in a small sample, but other than last year's Florida team (that beat a #14 seed, a #11 seed, and a #15, before losing to a #4 seed in the Elite Eight), these "underseeded" teams haven't done so well.

Obviously we'll have to wait and see about Louisville.

[rtnorthrup]

Every year, one or two (or three) top ten Pomeroy teams are in the situation where their seed is disproportionately worse than their Pomeroy ranking would suggest. This year it's Louisville -- #2 in Pomeroy with a 4-seed. Here are some others from the past five years:

2013:
Florida, #1 Pomeroy; 3-seed; made Elite Eight;
Pitt, #7 Pomeroy; 8-seed; lost in round of 64;

2012:
Wisconsin, #6 Pomeroy; 4-seed; lost in Sweet 16;
Memphis, #9 Pomeroy; 8-seed; lost in round of 64;

2011:
Texas, #4 Pomeroy; 4-seed; lost in round of 32;

2010:
Wisconsin, #3 Pomeroy; 4-seed; lost in round of 32;
BYU, #7 Pomeroy; 7-seed; lost in round of 32;

2009:
Memphis, #1 Pomeroy; 2-seed; lost in Sweet 16;
Gonzaga, #5 Pomeroy; 4-seed; lost in Sweet 16;
West Virginia, #8 Pomeroy, 6-seed; lost in round of 64;

I don't know if it's because the teams were overrated in Pomeroy, or because seeding matters a lot more than Ken Pomeroy and others seem to believe, or just dumb luck in a small sample, but other than last year's Florida team (that beat a #14 seed, a #11 seed, and a #15, before losing to a #4 seed in the Elite Eight), these "underseeded" teams haven't done so well.

Obviously we'll have to wait and see about Louisville.

That is interesting. Nate Silver's bracket uses 5 computer modeling simulations, of which Pomeroy is one, and still has Louisville as his highest percentage for the title. It would be interesting to look more into those numbers and see if there is a consistent pattern as to the major difference between KenPom's and RPI. As others have suggested, it may be margin of victory. Louisville and Wichita State have very similar resumes. According to KenPom they both sport very good offensive and defensive efficiency numbers, but both teams played a particularly weak schedule.

[superdave]

Lowest combined odds (1 in X to Win) for top 4 seeds, each of the four regions via KenPom's chart:

MIDWEST = 99: WSU 13, Lou 8, Duke 31, Michigan 47
EAST = 166: Virginia 9, Nova 20, Sparty 41, Iowa St 96
SOUTH = 170: Florida 8, Kansas 29, Cuse 55, UCLA 79
WEST = 173: Zona 6, Creighton 27, Wisco 43, SDSU 97

[Deslok]

If you go off the 538 projections, taking the top 4 probabilities in each region of teams winning the national championship you get:
23% chance of the top teams in the South winning the title
17% chance of the top teams in the East winning the title
20% chance of the top teams in the West winning the title
28% chance of the top teams in the MidWest winning the title

Just further confirmation that the MidWest is stacked compared to the other brackets - and btw, Kentucky was perhaps the only >1% team not factoring into those calculations(since Wichita St, Louisville, Duke,and Michigan were all ahead of them on probabilities) which makes the odds even more skewed(adding another 2% to the MidWest).

[DavidBenAkiva]

At the least, I am encouraged that computer models seem to consistently rate Duke with a better chance of getting to the Elite 8 than Michigan.

KenPom has Duke with a 33.2% chance of getting to the Elite 8 versus Michigan's 31.2%.

Silver's FiveThirtyEight is even more in favor of Duke's shot at getting to the Elite 8, 42.8% to Michigan's 37.0% chance. FiveThirtyEight, which likes Louisville quite a bit, has Duke with a better shot at the later rounds than does Wichita State.

This is pretty encouraging to me. As we saw in the Virginia game, we have a player that can really step up in a tight game. If Parker can keep 'it' on for long stretch, Duke has a pretty good shot at going far in the tourney this year.

[ice-9]

Silver also takes into account preseason rankings in his model -- this would favor Louisville by a lot.

I have Michigan St and Creighton advancing far in my bracket, primarily because they have essentially the same team as last year back. As we Duke fans know, Michigan St went to the Sweet 16 and Creighton lost in the round of 32. Those teams will be hungry to do better this year and will have the experience to know what it takes to win. Each player knows their role.

Similarly, I think Louisville is a little overrated at this point (given all the love shown by the press), and that's because this year's team is fundamentally different from last year's. The Cardinals are a tough team, sure, but they get too much credit for being the defending champions. Without Siva, this Louisville team isn't as potent, and their good-but-not-great performance against their weak schedule hasn't shown otherwise. The high KenPom ranking is a reflection of their dominant performance against weak teams, but a better indicator is performance against tournament-caliber teams and they're good but not elite there.

I wouldn't be surprised if Wichita St beats them. We fear Louisville because of the press but Wichita St doesn't share that fear. Louisville is worse than last year while Wichita St is better this year.