Journalists, Vegas oddsmakers, and that one college friend who you only hear from this time of year all agree: The NCAA tournament selection committee did a god-awful job of seeding the tournament teams this year. But, did it really?
The seeding outrage flared up again Sunday, when, in a matchup The Los Angeles Times claimed marred the tournament, and somehow decided was the de facto national title game, an underseeded 8 Kentucky took down 1-seed Wichita State for a spot in the Sweet 16. This draw, along with the Shockers' placement in a Hulkbuster region that included 2-seed Michigan, 3-seed Duke, and, comically, 4-seed Louisville, caused many to openly speculate on a nefarious NCAA conspiracy against Wichita State.
There certainly were a few seeding errors by the selection committee this year, and the errors that did happen involved prominent teams in difference-making slots. But how did they do as a whole? Let's take a look.
I aggregated stats from Ken Pomeroy and Jeff Sagarin on the tournament field over the last 12 years to get a power rating of each tourney team for each season. I then compared what seed the stats said each team should be assigned and compared this to the seed the selection committee actually gave them.
Here's a graph showing the average error in seed assignment per team from 2003-2014:
As you can see, 2014 isn't all that bad compared to recent years, and is actually better than the average.
This year's worst misseedings were: Massachusetts (actual 6, expected 11); Tennessee (11, 6); and Iowa (11, 6). Those mistakes were only five slots off, which actually isn't too bad compared to some of the worst errors the committee's made since 2003:
|Team||Year||Actual Seed||Expected Seed||Error|
|San Diego St.||2012||6||13||7|
No respect for Steph.
When you look at this year's field, only a handful of abnormally good 4, 9, and 11 seeds really stand out, and six teams from those seeds have advanced to the Sweet 16.
One interesting tidbit is that the committee has severely underseeded 4-seeds in recent years, and as a unit, they've actually been better than 3s in every tournament since 2009. This may be why, after seeing an average of 1.76 1- make the Final Four from when the tourney first adopted the modern 64-team bracket in 1985 to 2009, that rate has declined by a full team the last four years, with only three 1-seeds surviving to the final weekend during that period for a rate of 0.75. The super 4-seed trend is certainly true again this year, with Louisville (should be a 1) and Michigan State (should be a 3) in the mix.
A quick note: Many cite SMU's snubbing as additional evidence of the selection committee's shortcomings this year, and this analysis only grades the committee on its seeding of the teams they selected—not the actual selection process of those teams. It's possible that adding this criterion would cast a more unfavorable light on the committee's job this year, but solely in terms of seeding the teams, 2014 really isn't all that shabby.
Finally, for what it's worth, Wichita State should have been a 2 and Kentucky a 5 by these numbers (Michigan should have been a 3 and Duke a 2), so certainly a bit off, but probably not so egregious as to give the bracket truthers much to chew on. It is, however, fair to note that the "value" of a seed isn't a linear progression, as we saw when we looked at alternate tournament formats.
For reference, here's the expected and actual seeds of all the teams in this year's tournament. Click here for the list back to 2003.