How Often Does the Nation’s Best Team Win the National Championship?
Posted by Neil Paine on March 30, 2010
In a variation on a running theme (one that’s especially pertinent given the early departures of Syracuse, Kansas, & Kentucky from this year’s tourney), I wanted to know how often the “best” (i.e., most talented, most dominant over the entire season, etc.) team wins the NCAA Tournament. We know that the NFL’s best team wins the Super Bowl about 24% of the time, that the best team in baseball wins the World Series about 29% of the team (or at least, they did back in the 1980s when Bill James studied the issue), and that the NBA’s best team wins the Finals almost half of the time… So what’s your guess for college basketball?
On the one hand, the NCAA tourney seems to have more in common with football, since it’s a string of single-elimination games (made even more random by the games being played at neutral sites); however, on the other hand it’s still basketball, and the structure of the game itself is set up such that the better team wins more often than in other sports (in no small part due to the massive number of possessions per game — unlike football — as well as the ability to have your best player carry a disproportionate share of the team’s offensive chances — unlike baseball).
But rather than just make a guess, I once again set up a Monte Carlo simulator and ran a series of “test seasons”. To do this, I used the 2009-10 D-I basketball schedule, excluding postseason & conference tournament games, as well as games involving non-D-I schools. For each school, I assigned them a random “strength rating” from a normal distribution based on the average SRS scores of their conference members since 1980:
Conf | avgSRS | stdvSRS |
---|---|---|
A10 | 2.96 | 7.01 |
ACC | 12.18 | 7.08 |
AMEA | -8.87 | 5.70 |
ASUN | -7.79 | 7.02 |
BG10 | 11.54 | 7.09 |
BG12 | 9.23 | 7.37 |
BIGE | 10.06 | 6.52 |
BSKY | -4.49 | 6.63 |
BIGS | -11.15 | 5.62 |
BIGW | -1.93 | 6.35 |
CAA | -3.43 | 6.81 |
CUSA | 3.01 | 7.43 |
GWC | -9.58 | 7.28 |
HRZN | -2.21 | 6.44 |
IND | -19.09 | 9.72 |
IVY | -7.93 | 6.62 |
MAAC | -5.15 | 6.03 |
MAC | -0.93 | 6.24 |
MEAC | -14.77 | 6.41 |
MVC | 2.53 | 5.90 |
MOUW | 5.08 | 6.93 |
NEAS | -10.69 | 5.65 |
OVC | -6.94 | 5.91 |
PC10 | 9.22 | 6.87 |
PATR | -8.36 | 6.52 |
SUNB | -2.83 | 6.86 |
SOUC | -6.14 | 6.26 |
SEC | 10.76 | 5.91 |
STHL | -8.13 | 6.34 |
SUML | -5.73 | 5.57 |
SWAC | -13.93 | 7.51 |
WAC | 1.48 | 6.01 |
WCC | 0.26 | 6.70 |
So for instance, you might end up with an ACC that looks like this:
Team | Conf | Strength |
---|---|---|
North Carolina St | ACC | 21.61 |
Virginia Tech | ACC | 21.36 |
Wake Forest | ACC | 19.58 |
Duke | ACC | 16.01 |
Maryland | ACC | 12.81 |
Boston College | ACC | 11.65 |
Florida St | ACC | 7.82 |
Miami FL | ACC | 7.80 |
Georgia Tech | ACC | 4.18 |
Virginia | ACC | 3.39 |
North Carolina | ACC | 3.21 |
Clemson | ACC | -1.13 |
…And a Patriot League that looks like this:
Team | Conf | Strength |
---|---|---|
Holy Cross MA | Pat | 0.07 |
Colgate | Pat | -5.19 |
Lehigh | Pat | -5.54 |
Bucknell | Pat | -6.11 |
Navy | Pat | -7.10 |
Army | Pat | -7.14 |
Lafayette | Pat | -9.75 |
American U | Pat | -12.34 |
Do this for every team in the country, and estimate the chances of one team beating another at home using this equation:
p(homeWin) = 1 / (1 + EXP(-0.7390072 – 0.179593 * SRSdiff))
(Note: For neutral games, I averaged p(homeWin) and p(roadWin) in a given matchup.)
From this, you can generate standings for the 2010 regular season, like these:
Team | Conf | W | L | W% | confW | confL | confW% |
---|---|---|---|---|---|---|---|
California | P10 | 28 | 2 | 0.933 | 16 | 2 | 0.889 |
Washington St | P10 | 23 | 6 | 0.793 | 12 | 6 | 0.667 |
Washington | P10 | 23 | 7 | 0.767 | 12 | 6 | 0.667 |
Oregon St | P10 | 20 | 10 | 0.667 | 11 | 7 | 0.611 |
Oregon | P10 | 20 | 10 | 0.667 | 9 | 9 | 0.500 |
Stanford | P10 | 17 | 13 | 0.567 | 8 | 10 | 0.444 |
UCLA | P10 | 17 | 13 | 0.567 | 8 | 10 | 0.444 |
Arizona St | P10 | 15 | 16 | 0.484 | 7 | 11 | 0.389 |
Arizona | P10 | 8 | 22 | 0.267 | 5 | 13 | 0.278 |
Southern Cal | P10 | 6 | 24 | 0.200 | 2 | 16 | 0.111 |
You can also use these simulated W-L results to calculate RPI, the Ratings Percentage Index that the selection committee uses to inform their picks. The formula is 25% team winning pct., 50% opponents’ winning pct., and 25% opponents’ opponents’ winning pct:
Team | Conf | W% | Opp% | Opp^2% | RPI |
---|---|---|---|---|---|
Arkansas | SEC | 0.968 | 0.542 | 0.538 | 0.648 |
Kentucky | SEC | 0.871 | 0.554 | 0.544 | 0.631 |
LSU | SEC | 0.767 | 0.568 | 0.546 | 0.612 |
Mississippi | SEC | 0.800 | 0.528 | 0.548 | 0.601 |
South Carolina | SEC | 0.700 | 0.550 | 0.543 | 0.586 |
Vanderbilt | SEC | 0.690 | 0.547 | 0.544 | 0.582 |
Florida | SEC | 0.548 | 0.567 | 0.539 | 0.555 |
Georgia | SEC | 0.517 | 0.575 | 0.545 | 0.553 |
Alabama | SEC | 0.467 | 0.593 | 0.535 | 0.547 |
Tennessee | SEC | 0.467 | 0.589 | 0.539 | 0.546 |
Mississippi St | SEC | 0.419 | 0.559 | 0.530 | 0.517 |
Auburn | SEC | 0.367 | 0.580 | 0.525 | 0.513 |
Now we have all the components in place to put together the field for a simulated NCAA Tournament. The selections go like this: the 32 conference regular-season champions get automatic bids (ties are broken by RPI); the remaining 32 at-large slots are filled by the 32 best non-conference champs by RPI. Overall seedings are then made, ranking all tourney selections by RPI, and the regions are seeded according to an S-curve.
As you can see, I tracked the global “talent ranking” of the eventual champion for each simulated season; in this case, it was Kentucky, who ranked 9th in team talent (in case you’re wondering, NC State was #1, but they were toppled by Arkansas in the Round of 32). Repeat this process 1,000 times, and how often does a team of each talent ranking win it all?
Ranking | Championships |
---|---|
1 | 336 |
2 | 165 |
3 | 103 |
4 | 79 |
5 | 62 |
6 | 35 |
7 | 37 |
8 | 27 |
9 | 34 |
10 | 18 |
11 | 15 |
12 | 16 |
13 | 13 |
14 | 9 |
15 | 6 |
16 | 9 |
17 | 6 |
18 | 2 |
19 | 2 |
20 | 4 |
21 | 1 |
22 | 2 |
23 | 5 |
24 | 2 |
25 | 1 |
26 | 1 |
27 | 0 |
28 | 4 |
29 | 0 |
30 | 1 |
31 | 2 |
32 | 0 |
33 | 1 |
34 | 0 |
35 | 1 |
36 | 0 |
37 | 0 |
38 | 1 |
The nation’s most talented team wins the National Championship approximately one-third of the time in this simulation, which is slightly higher than we observed for football and baseball, but much lower than the results of a similar simulation of NBA results. Going back to our original hypotheses, it was clear that a single-elimination tournament of 6 rounds creates many opportunities for favored teams to slip up along the way to a title, in much the same way that the NFL playoffs’ one-and-done format encourages parity. At the same time, though, the better team wins a basketball game more often than it wins a football or baseball game simply because the random variance of luck is reduced with higher-scoring games and more possessions.
In the end, we saw basically what we should have predicted before the experiment — by mixing high-variance elements of other playoffs, including a single-elimination format, the NCAA Tournament will never see the best team prevail as often as we see in the NBA Playoffs, but the nature of basketball offsets these elements to a certain degree, resulting in a winning percentage for the best team that falls between those of the NBA and the NFL/MLB.
And a less abstract take-home message? Well, when the 2010 NCAA champs are crowned on Monday night, they’ll be the most deserving team, having outlasted 64 others over the past few weeks… But will they be the nation’s best college basketball team? The odds have spoken — and they say probably not.
March 31st, 2010 at 9:03 am
[…] CBB Blog at Sports Reference » Blog Archive » How Often Does the Nation’s Best Tea… […]
March 31st, 2010 at 11:58 am
How do the results from your simulation compare to the observed frequency of the best regular season team winning the championship? Also, it might be interesting to see whether some measure of the “dominance” of the best team has an effect. Maybe only 30% of your simulations produce a dominant #1 team, and in the rest of the simulations, the #1 team is closer to the rest of pack.
March 31st, 2010 at 2:25 pm
I want to know the answers to Bryan’s questions, as well.
Also, I note that nearly 75% of the time, one of the best five teams won the simulated championship. I don’t know that the format – the six-round, 64-team tournament – needs any kind of validation, but I think if the eventual champion is in the 98th percentile or higher that often, the format “works”.
And maybe it’s just me, but I was far more interested in the sample tournament you showed than I probably should have been.
March 31st, 2010 at 11:32 pm
What I thought was fascinating was that the most talented team in your example simulation only received an 8 seed. I’d be interested to see a distribution of the NCAA seed received by the most talented team. I would have never guessed that the best team could be as low as an 8.
March 31st, 2010 at 11:35 pm
OK, I can actually run this quickly, at least the first question… Here are the champs from 1980-2009 and their SRS rankings:
So the observed rate is about 27%, a bit lower than in the simulation, but certainly not outside the realm of possibility in a 30-season sample.
As for the SRS of the top team, I’ll have to run some sims again and see how they compare to the top scores of the real-life teams. I’m pretty sure they’ll both top out around 24-ish on avg.
April 6th, 2010 at 10:15 am
[…] WordPress Planet « How Often Does the Nation’s Best Team Win the National Championship? […]
June 15th, 2010 at 12:39 pm
How did you come up with the equation you used to predict each game?
p(homeWin) = 1 / (1 + EXP(-0.7390072 – 0.179593 * SRSdiff))
August 3rd, 2010 at 12:56 am
Logistic regression.
December 28th, 2011 at 6:24 pm
I like this excellent website much, rescued to bookmarking. “To hold a pencil is to be in warfare.Inches by simply Francois Jessica Arouet Voltaire.|
January 4th, 2012 at 4:54 pm
I appreciate your data in this article. It’s smart, well-written and straightforward to recognize. You’ve my attention on this topic. I will likely be back.
January 5th, 2012 at 2:42 am
Great items from you, man. I have be mindful your stuff previous to and you are simply extremely wonderful. I really like what you’ve acquired right here, really like what you are stating and the way in which during which you assert it. You make it enjoyable and you continue to care for to stay it sensible. I can’t wait to read much more from you. This is really a great web site.
January 29th, 2012 at 5:34 pm
hey there and thank you for your info – I’ve definitely picked up something new from right here. I did however expertise some technical points using this website, as I experienced to reload the website a lot of times previous to I could get it to load correctly. I had been wondering if your hosting is OK? Not that I’m complaining, but sluggish loading instances times will very frequently affect your placement in google and can damage your quality score if ads and marketing with Adwords. Anyway I am adding this RSS to my email and could look out for much more of your respective fascinating content. Make sure you update this again very soon..
February 7th, 2012 at 11:57 am
Thanks , I’ve just been searching for info about this topic for ages and yours is the best I have found out so far. But, what concerning the bottom line? Are you positive about the supply?
March 8th, 2012 at 8:55 pm
Vitamix Coupon…
[…]How Often Does the Nation’s Best Team Win the National Championship? » CBB Blog at Sports Reference » Blog Archive[…]…