Bracketology vs. the AP

[RegisteredUconn]

This analysis is based on the most recent AP poll and published bracket projections. Regions are designated by the #1 team in that region. E.G., Louisville region is called the UConn region, South Bend Region is Notre Dame region, etc. Top 25 teams in each region (AP rank in parentheses).

UConn Region: UConn (1), Louisville (5), North Carolina (6), Baylor (9), NC State (18), Iowa State (23)
Notre Dame Region: Notre Dame (2), South Carolina (7), Penn State (12), Texas A&M (17), California (21), Gonzaga (22), Middle Tennessee (25)
Duke Region: Duke (3), Tennessee (10), Okla State (11), Kentucky (13), Arizona State (15), Purdue (19), Florida State (23)
Stanford Region: Stanford (4), Maryland (8), LSU (14), Vanderbilt (16), West Virginia (20)

Assumptions:
1) AP poll offers a reasonably unbiased assessment of which teams will advance how far in post season play.
2) ESPN's Bracketology is a reasonably unbiased application of NCAA guidelines

Question: How well do NCAA bracket creation rules allow teams fair paths?

First Pass: Distribution of top 25 teams across regions. Reasonableness would suggest that regions have approximately equal numbers of teams.
Results:
UConn Region - 6 teams
Notre Dame Region - 7 teams
Duke Region - 7 teams
Stanford Region - 5 teams

Second Pass: Average ranking of Top 25 teams across regions. Reasonableness suggests approximately equal average rankings.
Results:
UConn Region - 10.3
Notre Dame Region - 15.1
Duke Region - 13.4
Stanford Region - 12.4

Third Pass: Distribution of Top (Sweet) 16 (AP poll) teams across regions. Expectation of relative equality.
Results:
UConn Region - 4 teams
Notre Dame Region - 3 teams
Duke Region - 5 teams
Stanford Region - 4 teams

Fourth Pass: Average ranking of Top 16 teams across regions. Expectation would be relative equality.
UConn Region - 5.3
Notre Dame Region - 10.5
Duke Region - 9.25
Stanford Region - 10.5

Fifth Pass: Distribution of Top Ten teams across regions. Would expect relative equality.
Results:
UConn Region - 4 teams
Notre Dame Region - 2 teams
Duke Region - 2 teams
Stanford Region - 2 teams

Prima facie evidence, namely the distribution of Top 25 and Top 16 teams across regions, exhibits commendable fairness. But comparative rankings and distribution of Top 10 teams shows bias bordering on malfeasance.

 

[UcMiami]

Part of the problem with the locations chosen and home teams is that most of the highly ranked teams are east coast and southern so the nearest east coast location is Louisville which is also the most southern location. California being the furthest location for all but a couple of top 25 locations will by proximity rule get the weakest teams as in each seed line the stronger teams will get distributed to the three central locations. And Louisville which everyone assumes is not going to be a #1 seed will perforce get the highest non-home #1 seed (if proximity is used and ND remains a #1 seed), plus the strongest of each of the following seed lines.

Sorry, but the more analysis one does on the choices the committee made for locations this year, the worse they look.

 

[ASweet3781]

Part of the problem with the locations chosen and home teams is that most of the highly ranked teams are east coast and southern so the nearest east coast location is Louisville which is also the most southern location. .

Which makes this committee priority on location in terms of seeing to be competitive malfeasance.

The only excuse to do it in the past was attendance. Well with Stanford, ND, and Louisville probably selling out their buildings what's the excuse this year to use it? None that I can see.

 

[Vowelguy]

Good stuff.
Though you're assuming that the AP rates teams the same way that the committee does -- not necessarily true. And other things come into play in the bracketing process.
Also, youre assuming Creme's bracketology is accurate.