It's a Mismatch: NCAA Tournament Seedings and the Ineffective Polls
NCAA selection chair Dan Guerrero attempted to defend the field he and his colleague pieced together for the 2010 NCAA men's Division I basketball tournament. Problem is, he has no reason to defend himself as he was simply following precedent.
Let's relive what happened last weekend in Atlanta.
Mississippi State makes a strong run to the SEC championship game, pushing Kentucky to the brink by taking the Wildcats to overtime. During that run, the Bulldogs defeated a Florida team that lost their last three regular season games and 5-5 in their last 10, excluding the SEC tourney (to be fair, MSU was also 5-5).
So, how is it that despite having a higher RPI ranking, Mississippi State is in the National Invitation Tournament while Florida is in the Big Dance as a No. 10 seed?
Certainly the Bulldogs suffered a “bad” loss down the stretch at the hands of Auburn, but it is quite confusing how a team with a better record and a higher RPI ranking—a measure of a team’s strength—is left out of the tournament.
Scenarios like this play out every season; which teams were snubbed and which teams made the field that probably should be in the NIT. It is a debate that plays out the minute the field is announced.
But there is a greater issue here beyond the teams that are snubbed: seeding mismatch.
Every season there are mismatches between the teams in field and their proper seeding. This issue trickles all the way down and leads to teams being left out that deserve to be in—at least according to the rankings.
Now, let's take a trip back to the 2006 tournament.
George Washington finished the regular season with only one loss, but lost in the Atlantic 10 tournament (won by Xavier). At 26-2 and ranked No. 14, George Washington should have been a four seed. Instead, they were given the eighth seed and subsequently lost in the second round to the one seed Duke.
This seeding mismatch put the Colonials at a disadvantage because they faced a more difficult matchup. They started with the nine seed UNC-Wilmington, a game that went to overtime before GW won it 88-85, and then faced the No. 1 overall seed. Meanwhile, fourth-seed LSU, a team that based on their No. 19 ranking should have been seeded fifth, made it to the Final Four.
But how often do seeding mismatches occur?
Well, based on the pre-NCAA tournament rankings, the answer is fairly often. And, the tendency is to under-seed teams.
Here is the average seeding matching for Top 25 teams from 2003 until this season:
- 2010: -0.47
- 2009: -0.27
- 2008 : -0.2
- 2007: -0.56
- 2006: -0.2
- 2005: -0.08
- 2004: -0.583 (-1.0)*
- 2003: -0.208*
A quick explanation about how this was calculated before I analyze this further.
Because there are four regions, the top four teams in a poll should receive the four No. 1 seeds; the next four teams in a poll (ranked five through eight) should receive the four No. 2 seeds; and so on.
Based on this, the projected seeding is subtracted by the actual seeding resulting in a discrepancy. If the discrepancy is zero, then the seeding matches the expectation. But, any deviation from zero means there is a mismatch. A positive number means a team is seeded higher than expected; a negative number means a team is seeded lower than expected.
For example, Butler is ranked eighth in the coaches poll and should be a two seed.
However, the Bulldogs received a fifth seed, which gives them a discrepancy of -3. Or, to put it another way, they are seeded three spots lower than where they ranking suggests they should be. Meanwhile 21st-ranked Baylor is seeded three spots higher than where they should be.
Thus, what the trend indicates is that on average, a Top 25 team is slightly under-seeded. But there is more to it than the overall average.
But, if you look at a conference breakdown since 2003, what becomes clear is that certain schools get under-seeded while others are over-seeded.
The following is the net seeding discrepancy, with the average in parenthesis:
- ACC: 4 (0.148)
- Atlantic 10: -2 (-0.25)
- Big 12: 7 (0.292)
- Big East: 1 (0.024)
- Big South: -5 (-5)
- Big Ten: -10 (-0.476)
- Big West: -2 (-2)
- Conference USA: -16 (-1.23)
- Horizon: -12 (-1.6)
- Missouri Valley: -12 (-3)
- Mountain West: -5 (-0.714)
- Pac 10: -4 (-0.222)
- SEC: 5 (0.263)
- WAC: -8 (-2)
- West Coast: -7 (-1.167)
What is surprising is how under-seeded the Big Ten tends to be. But it is not close to how grossly under-seeded the WAC, West Coast, and Missouri Valley conferences tend to be.
And if you include the 2004 snubbing of Utah State, then the WAC’s numbers become even worse.
What is not surprising is that the ACC usually has their members seeded higher than their rankings indicate. But over-seeding among the Top 25 seems to be marginal.
However, when we group the conference together, a clearer pattern emerges.
The Big Six tend to be over-seeded by 0.0199 spots. In other words, teams from the Big Six—ACC, Big 12, Big East, Big Ten, Pac 10 and SEC—are usually seeded correctly or slightly over-seeded.
Teams from all other conferences tend to be grossly under-seeded. On average, a team from a non-Big Six conference is given a seeding that is 1.55 lower than their ranking would suggest.
What this means it that the selection committee believes that teams from outside of the Big Six are ranked too high.
It is this type of mentality that upsets coaches and teams from outside of the Big Six. In an article written by Ray Melick of The Birmingham News , UAB head basketball coach Mike Davis expressed disappointment in the NCAA selection process. His frustration was not because the Blazers were left out of the tournament.
UTEP gets a 12th seed? That’s a pretty strong message. They were the best team in our conference, a top 25 team, and they get a 12th seed?
Again, this speaks not necessarily to who is in and who is out, but this statement speaks to the mismatch that occurs in the seeding process.
With the seeding mismatch for this season being -0.48, it is worth looking deeper at the 2010 seedings for all teams receiving votes in the coaches’ poll.
For the 38 teams that at least received votes in the pre-tournament poll, the average discrepancy is -0.82. The conference breakdown should not surprise anyone. Here is the net seeding discrepancy for all conferences with teams receiving votes.
- ACC: 0
- Atlantic 10: 2 (0.67)
- Big 12: 6 (1.2)
- Big East: 5 (0.83)
- Big Ten: -2 (-0.4)
- Conference USA: -5 (-5)
- Colonial: -2 (-2)
- Horizon: -3 (-3)
- Ivy League: -4 (-4)
- Metro Atlantic: -4 (-4)
- Missouri Valley: -3 (-3)
- Mountain West: -7 (-2.33)
- Pac 10: -3 (-3)
- SEC: 0
- WAC: -4 (-4)
- West Coast: -6 (-3)
Once again, the non-Big Six programs tend to be under-seeded (-2.4 average) while the Big Six schools are on target or slightly over-seeded (0.261 average).
The point of all of this is that rankings and polls are useless in college basketball.
What good is being ranked eighth when you are not seeded accordingly? It makes no sense to have BYU ranked 16th and seeded seventh if teams ranked below them— Pittsburgh (17th), Wisconsin (19th), Maryland (20th), Baylor (21st), Vanderbilt (22nd), and Texas A&M (23rd)—are seeded higher.
What is worse, is that a team that is not even receiving votes—Notre Dame as a six seed—is seeded higher than both BYU and Gonzaga (ranked 18th; seeded eighth).
These types of mismatches turn polls into jokes. But how does a ranking system like the RPI measure up?
Well, if looking only at NCAA tournament teams in the RPI top 65, then there is actually a slight over-seeding (0.235). But, due to upsets and conference champions with extremely low RPI ratings, not every team in the RPI top 65 gets in the Big Dance.
The average RPI of all NCAA tournament teams is 114.57. This is because there are 14 teams with an RPI greater than 65.
Therefore, those 14 teams take the place of 14 teams in the RPI top 65 that are not conference winners. This means that the last four into the tournament should be UNLV, Kent State, UAB, and Wichita State.
Of those four, only UNLV made it in as an eight seed.
Accordingly, the last four out should have been Notre Dame, Marquette, Memphis and Mississippi State. So, while perhaps MSU still should not have made it in, why is Notre Dame and Marquette in?
Not only that, but why are those two teams graced with six seeds while a team like Northern Iowa (RPI 17) is a nine seed?
It is worth noting that in the 2010 NCAA women's tournament, teams are more likely to be under-seeded as well. In this case, however, the Big Six schools are either on target or slightly under-seeded (-0.091) while all other schools were grossly under-seeded (-2).
Therefore, it seems as though under-seeding is common phenomenon.
In the end, the problem rests on perceptions.
If polls and RPI do not matter, then what does seem to count is conferences. Some will say that strength of schedule offsets the polls and that would be fair to argue why Northern Iowa or BYU are low and a team like Texas A&M is higher.
But, then again, what of Kentucky and their 31st S.O.S.?
So, it is still all about the conference. Even though the Big Ten does tend to be under-seeded, the ACC and Big 12 make up for that loss, along with the SEC and the Big East. Meanwhile, conferences like the Missouri Valley and Mountain West continue to be under-appreciated and under-seeded.
And, if we are going to simply go by the perception of conferences, should Division I be subdivided like D-I football?
I mean, the likelihood of Robert Morris making even the Sweet 16 is slim, let alone the title game. Subdivide it so teams from the Atlantic Sun or Patriot League have a shot a title and the “big boys” can play amongst themselves.
That is what the selection committee wants, right?
The selection committee is human and we should at least be somewhat appreciative of that. If the tournament seedings were simply conducted by computers, then we would have no drama, no suspense, no debate, and people like Joe Lunardi would be out of the job.
But because of the human factor, we are left with biases that allow a mediocre team like Wake Forest in on the perception of their conference while the UAB's and Wichita State's of the college basketball world are left sitting in the corner wondering when someone will ask them to dance.
* – Only 24 of the 25 ranked teams made the NCAA tournament in 2003 and 2004. In 2003, Georgia withdrew from all post-season tournaments while investigating then-head coach Jim Harrick. In 2004, Utah State was not extended an invitation. If they are included in the discrepancy measure and their actual "seeding" is 17 (i.e. not in the tournament), then the average discrepancy in 2004 is -1.0 rather than -0.583.
This article first appeared at Uncle Popov's Drunken Sports Rant on Monday March 15, 2010.
What is the duplicate article?
Why is this article offensive?
Where is this article plagiarized from?
Why is this article poorly edited?