Tuesday, March 16, 2021

"Luck" and College Basketball

As we've previously discussed, Kentucky had an incredibly poor record in close games this year.  According to Ken Pomeroy, UK was the 49th best team in the country.  But they had a record of 9-16 because they lost so many close games.  Thus, he ranked them number 354 in the country in terms of "luck," which is his term for why some teams do better (or worse) than their statistics indicate.  Only three teams had a lower "luck" score.  One of them was Murray State, at 356.  Vandy ranked 352.  (Vandy was just about the only team to lose a close game to UK.)  I felt like I spent the whole year watching close losses, and apparently I was right.

Now Pomeroy's quite serious about the use of the term luck.  He argues that a team's past performance in close games is not indicative of how it will do in future close games.  And there are certainly strong points in his favor.  For example, in 2014 UK's luck score was terrible all year.  It finished at 240 in luck even after reaching the NCAA Final.  But I've never seen another UK team play better in late-game situations than those guys did in the NCAA Tournament.  Pomeroy would say that they had bad luck most of the year, and then it evened out.

This year, Western was on the other side of that equation.  The Hilltoppers were 25th in the country in luck, meaning that they were very successful in close games.  They beat Memphis by six.  They beat Alabama by two.  They squeezed out several close wins in the season, including a 64-60 victory over UAB in the C-USA semi-finals.  So when Western was in a late-game situation against North Texas in the final, you had to like their chances.  In fact, the Hilltoppers were up 48-43 with 1:15 left.  At that point, Western had a 95.3 percent chance of victory.  And then their luck turned.  Examples like that are why Pomeroy calls it "luck" instead of something like "close-game success."

But I'm not so sure.  We have data for 12 Calipari teams at UK.  Let's rank them by luck (final four teams in bold):

2010:  Kentucky (35-3) (43d in luck)
2015:  Kentucky (38-1) (68th)
2012:  Kentucky (38-2) (86th)
2019:  Kentucky (30-7) (98th)
2020:  Kentucky (25-6) (130th)
2017:  Kentucky (32-6) (175th)
2013:  Kentucky (21-12) (194th)
2018:  Kentucky (26-11) (199th)
2014:  Kentucky (29-11) (240th)
2011:  Kentucky (29-9) (285th)
2016:  Kentucky (27-9) (288th)
2021:  Kentucky (9-16) (354th)

There are certainly some facts here to support Pomeroy's thesis.  John Wall's team did really well in close games, and then bombed out against West Virginia.  Brandon Knight's team was awful in close games, and then beat OSU and UNC to reach the Final Four.  The 2014 team was unlucky but beat Wisconsin; the 2015 team was lucky but lost to Wisconsin.  

Still, there are patterns.  The "lucky" teams tend to have more veterans, and more talent, than teams that are less fortunate.  The top three teams in luck probably were the three best teams Calipari has had at UK, and those 2019 and 2020 teams are looking better in retrospect as their players do well in the NBA.

We'd have to look at a lot more teams to draw real conclusions here, and the safest thing is probably to trust Pomeroy for the most part.  But I'm struck at how wise the common sense of most fans turns out to be.  For years, the press made fun of UK fans for getting stressed about because we "only" beat (name of poor team) by 10 points.  Now it turns out that the most advanced analytical thinking studies basketball in the same way.  If you're a really good team, you will crush weak teams -- just as we always thought.

On the other hand, fans traditionally believed that teams with big stars like John Wall, Anthony Davis, and Karl-Anthony Towns were more likely to win close games.  They thought that teams with Final Four experience -- like UK in 2012 and 2015 -- would be tougher down the stretch.  And that turns out to be correct as well.

And finally, fans always understood that you should never give up -- that even an unlikely team can go on an amazing run.  As we saw in 2014, that's also true.

So I'm not convinced that it's all luck.  But I do think that Pomeroy has created a useful framework to distinguish between a team's record and its ability.  And that is a very helpful tool.


  1. I think it's a useful tool, too. But it feels to me that what his number measures is not so much luck as how quickly a team's players make decisions and act on them. Kentucky didn't look as unlucky to me this season as it its players didn't seem to know what to do. Amidst the wondering, somebody would take a tentative or impossible shot, or some bad thing would happen to them.

    Darius Miller, one of my absolute favorite Kentucky players of all time, seemed to arrive from Maysville knowing what to do, and then he grew in his time at Lexington into being willing to do it. By the time he left UK, he was a guy who could force his way onto professional-basketball rosters. He has played 267 career NBA games--and counting. That's 44th most in the history of the program. I love Darius Miller.