NCAA Bracket Analysis

by Huck L Berry on Mar 17, 2009 2:08 PM CDT in Texas Longhorns and Other Basketball

Using my college basketball ratings, I ran an analysis on the NCAA tournament. I only used the power and standard deviation numbers, not the ones that account for wins and losses, so teams that make a habit of winning close games will not be adjusted in this run. This is the first time I've run such a look at the tourney, and the extent to which favorites are favored surprised me somewhat, but we'll have to see how it goes. First things first, here's the table (HTML Gods willing it will look decent):

Reg	Team	1R	2R	S16	E8	F4	NC	Title
M I D W E S T	Louisville	100.00%	99.84%	87.27%	64.51%	40.42%	21.20%	9.66%
	Alabama St.	40.75%	0.03%	0.00%	0.00%	0.00%	0.00%	0.00%
	Morehead St.	59.25%	0.12%	0.00%	0.00%	0.00%	0.00%	0.00%
	Ohio St.	100.00%	78.00%	11.73%	4.20%	1.03%	0.14%	0.02%
	Siena	100.00%	22.00%	1.00%	0.14%	0.01%	0.00%	0.00%
	Utah	100.00%	54.85%	20.42%	4.70%	1.34%	0.23%	0.03%
	Arizona	100.00%	45.15%	15.20%	3.13%	0.80%	0.12%	0.02%
	Wake Forest	100.00%	92.19%	63.10%	23.25%	10.29%	3.33%	0.90%
	Cleveland St.	100.00%	7.81%	1.29%	0.07%	0.00%	0.00%	0.00%
	West Virginia	100.00%	93.98%	52.60%	34.62%	17.95%	8.04%	3.08%
	Dayton	100.00%	6.02%	0.60%	0.08%	0.01%	0.00%	0.00%
	Kansas	100.00%	93.65%	46.30%	29.00%	14.12%	5.84%	2.06%
	North Dakota St.	100.00%	6.35%	0.50%	0.06%	0.00%	0.00%	0.00%
	Boston College	100.00%	29.16%	6.56%	1.00%	0.16%	0.02%	0.00%
	Southern Cal	100.00%	70.84%	26.00%	7.02%	1.98%	0.37%	0.06%
	Michigan St.	100.00%	98.06%	67.36%	28.22%	11.89%	4.10%	1.19%
	Robert Morris	100.00%	1.94%	0.07%	0.00%	0.00%	0.00%	0.00%
W E S T	Connecticut	100.00%	99.99%	80.10%	58.75%	30.71%	18.40%	8.30%
	Chattanooga	100.00%	0.01%	0.00%	0.00%	0.00%	0.00%	0.00%
	BYU	100.00%	79.82%	18.85%	8.84%	2.54%	0.93%	0.22%
	Texas A&M	100.00%	20.18%	1.06%	0.20%	0.02%	0.00%	0.00%
	Purdue	100.00%	93.89%	49.68%	16.04%	5.06%	1.99%	0.52%
	Northern Iowa	100.00%	6.11%	0.53%	0.01%	0.00%	0.00%	0.00%
	Washington	100.00%	85.66%	47.00%	15.95%	5.27%	2.17%	0.59%
	Mississippi St.	100.00%	14.34%	2.79%	0.22%	0.02%	0.00%	0.00%
	Marquette	100.00%	78.60%	26.01%	6.95%	2.24%	0.79%	0.18%
	Utah St.	100.00%	21.40%	2.60%	0.19%	0.02%	0.00%	0.00%
	Missouri	100.00%	96.66%	71.05%	32.75%	17.47%	9.48%	3.91%
	Cornell	100.00%	3.34%	0.34%	0.01%	0.00%	0.00%	0.00%
	California	100.00%	66.01%	9.65%	2.88%	0.68%	0.19%	0.03%
	Maryland	100.00%	33.99%	2.95%	0.58%	0.08%	0.01%	0.00%
	Memphis	100.00%	99.30%	87.33%	56.64%	35.89%	22.63%	11.12%
	Cal St.-Northridge	100.00%	0.70%	0.07%	0.00%	0.00%	0.00%	0.00%
E A S T	Pittsburgh	100.00%	99.69%	86.37%	77.60%	46.58%	23.73%	13.95%
	East Tennessee St.	100.00%	0.31%	0.02%	0.00%	0.00%	0.00%	0.00%
	Oklahoma St.	100.00%	49.47%	5.74%	3.13%	0.66%	0.11%	0.02%
	Tennessee	100.00%	50.53%	7.87%	4.28%	1.04%	0.21%	0.05%
	Florida St.	100.00%	42.94%	16.65%	1.69%	0.22%	0.02%	0.00%
	Wisconsin	100.00%	57.06%	25.84%	3.41%	0.58%	0.08%	0.01%
	Xavier	100.00%	94.85%	57.11%	9.88%	2.16%	0.38%	0.08%
	Portland St.	100.00%	5.15%	0.40%	0.00%	0.00%	0.00%	0.00%
	UCLA	100.00%	94.94%	65.44%	32.97%	16.28%	6.98%	3.45%
	Virginia Commonwealth	100.00%	5.06%	0.51%	0.03%	0.00%	0.00%	0.00%
	Villanova	100.00%	99.53%	34.05%	10.96%	3.37%	0.88%	0.27%
	American	100.00%	0.47%	0.00%	0.00%	0.00%	0.00%	0.00%
	Texas	100.00%	65.28%	12.61%	3.74%	0.89%	0.18%	0.04%
	Minnesota	100.00%	34.72%	4.13%	0.79%	0.12%	0.02%	0.00%
	Duke	100.00%	99.93%	83.27%	51.50%	28.10%	13.37%	7.30%
	Binghamton	100.00%	0.07%	0.00%	0.00%	0.00%	0.00%	0.00%
S O U T H	North Carolina	100.00%	99.96%	93.75%	69.73%	56.32%	37.78%	26.17%
	Radford	100.00%	0.04%	0.00%	0.00%	0.00%	0.00%	0.00%
	LSU	100.00%	60.12%	4.97%	1.17%	0.32%	0.05%	0.01%
	Butler	100.00%	39.88%	1.28%	0.18%	0.03%	0.00%	0.00%
	Illinois	100.00%	91.14%	26.01%	4.26%	1.63%	0.38%	0.10%
	Western Kentucky	100.00%	8.86%	0.34%	0.01%	0.00%	0.00%	0.00%
	Gonzaga	100.00%	98.05%	73.43%	24.65%	15.54%	7.17%	3.52%
	Akron	100.00%	1.95%	0.21%	0.00%	0.00%	0.00%	0.00%
	Arizona St.	100.00%	86.50%	44.05%	22.99%	6.35%	2.17%	0.81%
	Temple	100.00%	13.50%	1.79%	0.24%	0.01%	0.00%	0.00%
	Syracuse	100.00%	98.33%	54.09%	28.20%	7.49%	2.49%	0.90%
	Stephen F. Austin	100.00%	1.67%	0.07%	0.00%	0.00%	0.00%	0.00%
	Clemson	100.00%	72.06%	30.46%	13.70%	3.23%	0.95%	0.30%
	Michigan	100.00%	27.94%	5.45%	1.27%	0.13%	0.02%	0.00%
	Oklahoma	100.00%	99.86%	64.09%	33.60%	8.96%	3.02%	1.12%
	Morgan St.	100.00%	0.14%	0.00%	0.00%	0.00%	0.00%	0.00%

As expected, North Carolina is the favorite heading in and is expected to win the tournament a quarter of the time. That they are nearly twice as likely to win as the next most likely team, Pittsburgh, was very surprising to me.

And, of course, the draw affects a team's outlook. Louisville is 6th in the power ratings but the #1 seed moves them to 4th most likely to win the tournament. Pittsburgh is calculated as the most likely team to reach the regional final as the other half of their 8-team subregion is weaker than North Carolina's (mostly due to the Zags being in the South).

The good news for Texas fans is that the odds of the Longhorns winning the tournament are roughly the same as the odds of a PGA tour golfer making a hole-in-one on any given Par 3 hole. So it could happen.

Tweet Comment 14 comments | 0 recs |

Do you like this story?

Comments

This is awesome, Huck. Thanks.

The weakness of the Midwest is notable. I really like Louisville to walk out of there easily.

UConn almost as likely to win as…Duke? Wow.

by Scipio Tex on Mar 17, 2009 2:16 PM CDT reply actions

Your ratings suck. This dude’s will be better.

http://dehn.slu.edu/sports/tournament/index.html

Even though he can’t spell ‘coming.’

by HenryJames on Mar 17, 2009 2:48 PM CDT reply actions

You could probably go ahead and round those numbers to integer percentages – I think presenting two significant digits lends a level of accuracy that can most kindly be described as hopeful.

by Ag_in_TX on Mar 17, 2009 3:33 PM CDT reply actions

Except for Memphis why is there such a large discrepancy between #1 and #2 seeds? For example: Pittsburgh has a 12.8% better chance of winning it all than Michigan St.?

by Art Vandelay on Mar 17, 2009 4:17 PM CDT reply actions

HenryJames – that Leschner dude is all wrong. The percentage of ESPN voters who pick a given team to reach the final four is not a “perceived probability.” That’s a cross-level inference error. I could perceive that Team A has only a 25-30% chance of making the final four, but when I have to make an actual pick, the odds ratio for that team might be 2-1 over the next best competitor, so it makes sense that people pick them 2/3 of the time.

by Fuzzy on Mar 17, 2009 4:29 PM CDT reply actions

I pay attention to these things not at all. I just like to mock Huckleberry’s statistical faggotry.

by HenryJames on Mar 17, 2009 5:01 PM CDT reply actions

Did I hear my name called?

by statistical faggotry on Mar 17, 2009 5:03 PM CDT reply actions

So you’re sayin’ there’s a chance…

by Bob in Houston on Mar 17, 2009 5:11 PM CDT reply actions

HJ – while I might listen to a math geek who can’t spell “coming”, I am less inclined to believe one who can’t spell “two”.

by Longhorn in Canada on Mar 17, 2009 5:29 PM CDT reply actions

Ag -

Actually, they were rounded once upon a time, but that got lost when I copy-and-pasted into notepad so I could format the table. And there’s no way I’m going through it to erase all the decimal places.

by Huckleberry on Mar 17, 2009 7:31 PM CDT reply actions

No problem – in my line of work, the number of significant digits presented in data has meaning.

Hence, while some denizens of the Intrawebs are “Grammar Nazis”, I am a “Significant Digit Nazi” – a small and detestable lot to which I belong.

by Ag_in_TX on Mar 18, 2009 8:23 AM CDT reply actions

Well, to be fair, my ratings are accurate to at least the millionths digit, so I think I’m okay anyway. Excel’s calculation precision is the only possible limiter on the number of significant digits I am able to display.

The very notion that I may introduce some sort of error into the process is nothing short of preposterous!!!

by Huckleberry on Mar 18, 2009 8:57 AM CDT reply actions