Revisionist history: Tempo-free stats and Wisconsin's tournament woes

Kevin C. Cox

The last 11 years weren't so disappointing after all. Ken Pomeroy recently re-evaluated his formula, which makes the Badgers' NCAA shortcomings a bit easier to fathom.

Just before this season began, Ken Pomeroy dropped a bombshell: he was tweaking his algorithm, and he was doing it retroactively. The nature of the tweak suggests that it is a direct response to qualms about the Badgers’ unlikely dominance of his ratings: Pomeroy adjusted his algorithm to deemphasize blowout wins over terrible teams. Blowing out terrible teams is something that Bo Ryan’s Badgers do very, very well.

Of course, the Pomeroy ratings have almost always ranked the Badgers more highly -- usually much more highly -- than the pollsters and talking heads. As I’ve detailed elsewhere, the Badgers have almost always finished the season rated higher in the ratings than in the AP poll. So the Kenpom ratings have always been a bit of vindication for the Badgers, who take a fair amount of heat for their supposedly boring, plodding style.

But as the Kenpom ratings have become more and more famous, and the Badgers have become more and more a fixture in the Kenpom top five, a dark side emerged. First, Pomeroy has been a bit defensive about the Wisconsin problem, going so far as to write a long post defending his algorithm in response to teasing about the Badgers. Second, tempo-free adherents have become frustrated with the Badgers’ failure to make a deep run in the NCAA tournament, as the Kenpom ratings might predict. Likewise, the tempo-free doubters use this as fodder denigrate the tempo-free revolution writ large.

If their computerized profile was accurate, no one could deny that the Badgers have underachieved in the tournament. Last spring, I ran 10,000 simulations of every NCAA tournament since 2003 using the final ratings as the source for team quality. This allowed me to determine how many "expected wins" each team had over that period, which I compared to its actual wins to get a plus-minus against Pomeroy's expectations:

1 Pittsburgh -5.9 13 19
2 Clemson -4.2 0 4
3 Georgetown -4.1 8 12
4 Notre Dame -3.9 4 8
5 Duke -3.5 24 27
6 Brigham Young -3.3 4 7
7 Utah State -3.3 0 3
8 Wisconsin -3.2 15 18
9 Missouri -2.6 5 8
10 Creighton -2.5 2 5

Wisconsin had underachieved by 3.2 wins, good for eighth worst. Below is a year-by-year breakdown for the Badgers, including their probability each year of reaching the Final Four.

A negative number in the +/- column indicates Wisconsin underachieved based on its seeding; a positive number means the Badgers exceeded expectations according to Pomeroy.

2013 5 -1.58 12.6%
2012 4 -0.19 18.0%
2011 4 0.45 14.8%
2010 4 -0.87 14.9%
2009 12 0.11 3.1%
2008 3 -0.34 16.8%
2007 2 -1.30 19.5%
2006 9 -0.55 1.2%
2005 6 1.58 5.7%
2004 6 -.067 13.7%
2003 5 0.17 9.9%
TOTAL: -3.20 1.30
Final Four probability: 76%

Now history has changed. Under the new Kenpom algorithm, the Badgers just weren't quite as good as we thought they were. But, on the bright side, it turns out the last 11 postseasons weren't as disappointing as we've been led to believe. Here’s the same chart as above, with the results using the new Kenpom data:

2013 5 -1.3 8.2%
2012 4 0.1 13.2%
2011 4 0.4 14.7%
2010 4 -0.5 10.4%
2009 12 0.1 3.3%
2008 3 -0.2 13.9%
2007 2 -1.2 19.1%
2006 9 -0.6 1.3%
2005 6 1.6 6.3%
2004 6 -0.5 10.6%
2003 5 0.4 8.2%
TOTAL: -1.7 1.09
Final Four probability: 68.6%

To summarize, the Kenpom 2.0 Badgers were expected to win 1.5 fewer games in the NCAA tournament over the past 11 seasons than the Kenpom 1.0 Badgers. Here’s a chart that shows the differences year-by-year:

2013 -1.3 -1.6 0.3
2012 0.1 -0.2 0.3
2011 0.4 0.5 -0.1
2010 -0.5 -0.9 0.3
2009 0.1 0.1 0.0
2008 -0.2 -0.3 0.2
2007 -1.2 -1.3 0.1
2006 -0.6 -0.6 0.0
2005 1.6 1.6 0.0
2004 -0.5 -0.7 0.2
2003 0.4 0.2 0.2
TOTAL: -1.7 -3.2 1.5

So there was only one year, 2011, where the Kenpom 2.0 Badgers were expected to perform significantly better than the Kenpom 1.0 Badgers, and six years where the reverse is true, including three of the last four. All told, the Badgers are now only the 23rd-most disappointing tourney team over the past 11 years. In other words, not that disappointing at all. (For the record, the most disappointing team under Kenpom 1.0 -- Pittsburgh -- remains the most disappointing team under Kenpom 2.0, and even more so at -6.0.)

There is only one team in all of Division I basketball whose tournament expectations changed more than the Badgers’ using the new Kenpom data: Kansas. Here are the top five and bottom five in terms of changed expectations over the last 11 years:

Kansas 1.6
Wisconsin 1.5
Louisville 1.2
Duke 1.2
Maryland 1.1
Winthrop -0.4
S. Illinois -0.5
Davidson -0.6
Butler -0.6
Illinois -0.7

It is mostly blue-bloods who’ve been downgraded in retrospect, and mostly mid-majors who have been upgraded -- notably including Davidson, which was much better than anyone gave it credit for when the Wildcats whooped the Badgers in 2008. Presumably what the teams on top share is that they built a good Kenpom 1.0 rating by blowing out terrible teams; what the bottom teams share is decent (but not blowout) wins over middling in-conference teams that Kenpom 1.0 didn’t give them much credit for.

The notable exception is Illinois. Turns out Bruce Weber’s teams were better than Kenpom 1.0 gave them credit for. Alas, that means his reign was even more disappointing than Illinois fans thought.

More from Bucky's 5th Quarter:

Log In Sign Up

Log In Sign Up

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

By becoming a registered user, you are also agreeing to our Terms and confirming that you have read our Privacy Policy.

Join Bucky's 5th Quarter

You must be a member of Bucky's 5th Quarter to participate.

We have our own Community Guidelines at Bucky's 5th Quarter. You should read them.

Join Bucky's 5th Quarter

You must be a member of Bucky's 5th Quarter to participate.

We have our own Community Guidelines at Bucky's 5th Quarter. You should read them.




Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.