The analytic perils of dual-efficiency fetishism



We’ve reached the time of year when good teams are being praised on the basis of showing up “in the top-[highest applicable number divisible by five] for both offensive and defensive efficiency at KenPom.” Specifically, Michigan State’s getting a lot of this variety of love at present.

(Virginia qualifies for this treatment too, surely, but the Cavaliers in 2019 are destined to be a special case. There’s a lingering UMBC effect 10 months after the fact that seems to be inhibiting a more full-throated chorus of bedazzlement.)

The Spartans are indeed destroying opponents, of course. Tom Izzo’s guys could well win the national title. (Heck, I’ve sung their praises too.) Not to mention it’s a clear basketball benefit to be one of the best teams in the country at offense at the same time that you’re also one of the best teams in the country at defense.

There’s no searing indictment to be filed against such common-sense notions, goodness knows, but a warning label may still be in order. “Top-X in both offensive and defensive efficiency at KenPom” isn’t as predictive of tournament success as you probably think it is, and, in particular, dual-efficiency essentialism can’t shed much light on whether a team so blessed — even if it’s a top seed — will reach the Final Four.

At this time a year ago, the dual-efficiency praise was being lavished on Duke. The No. 2-seeded Blue Devils went on to reach the 45th minute of the Elite Eight before falling to No. 1 seed Kansas, so that can be classed as performing to or even exceeding expectations.

Past the Blue Devils, another team that looked good in these terms was (well, this is a coincidence) Michigan State. That didn’t pan out as well as expected come March. Past the Spartans, still another logical candidate for this brand of compliment in 2018 was (seriously, this is getting eerie) Virginia. Full stop.

It’s good to be good at both offense and defense, but the only thing that’s better is to be good at outscoring opponents, period. The hoops gods really don’t care how you do that, and, anyway, ordinal rankings stand at one remove from this kind of scoring-margin ding an sich. If you’re No. 5 in the nation for defense but you’re a relative country mile behind the top two or three, the ranking itself can fog up our best appraisal glasses ever so slightly.

Besides, a glance at Ken’s overall rankings illustrates how teams can reach effectively identical levels of basketball excellence through very different mixes of offense and defense. At this writing, Purdue shows up in the top 10 despite having a defense that ranks outside the top 50. Meanwhile, running right alongside Matt Painter’s guys in the rankings is Kentucky, which displays far more uniform rankings on both sides of the ball.

Does this mean UK is destined for greatness while Purdue’s doomed because of its defense? Perhaps. Nevertheless, in a world where we all saw South Carolina’s offense play basketball in the 2017 Final Four, there’s limited traction to be gained in insisting as a matter of irrevocable decree that this or that offense or defense is simply too weak to get the job done. The odds are long, sure, but never say never.

Every year, there’s a team that looks the best on Selection Sunday in terms of dual efficiency. Most years, of course, that team does not win the national championship, for the good and sufficient reason that any team that ranks best for any quality we care to name is less likely to win it all than is the rest of the field.

Actually, most years our dual-efficiency champion doesn’t “even” (ha) reach the Final Four….

Best combinations of offensive and defensive efficiencies, 2002-18
KenPom adjusted efficiencies on Selection Sunday

                      OE rank  DE rank    Result
2018  Duke               3        7       Elite Eight
2017  Gonzaga           10        2       National runner-up
2016  Kansas             7        4       Elite Eight
2015  Kentucky           6        1       Final Four
2014  Louisville         7        6       Sweet 16
2013  Florida            6        4       Elite Eight
2012  Kentucky           2        6       National champion
2011  Duke               6        5       Sweet 16
2010  Kansas             2        4       Round of 32
2009  Gonzaga           11       11       Sweet 16
2008  Kansas             1        3       National champion
2007  North Carolina     3        2       Elite Eight
2006  UConn              8        9       Elite Eight
2005  Illinois           2        5       National runner-up
2004  Duke               3        3       Final Four
2003  Kentucky           8        4       Elite Eight
2002  Duke               1        2       Sweet 16  

That sums to a pretty strong collective record, to be sure. If you expect this year’s dual-efficiency champion to make the second weekend, you’re certainly on solid ground, the occasional Farokhmanesh notwithstanding.

(Also, take a good long look at that line item for 2008. Now, join me anew in gobsmacked shock and incredulity over the fact that the nascent national champion posted those numbers and went 37-3 yet was never once ranked No. 1 in the AP poll that season. Said team in fact clocked in at No. 4 on Selection Sunday. Pollsters, am I right?)

Still, the table above pulls together not only 17 years’ worth of dual-efficiency champions but also no fewer than 13 No. 1 seeds. These teams were supposed to get to the second weekend.

As a group stretching back to the dim mists of KenPom year one when mastodons roamed the earth and people still used rebound margin, these 17 teams only slightly outperformed their seed expectation. The dual-efficiency special sauce hasn’t added much kick in March in the KenPom era.

Bonus methodological suggestion. When looking at offensive or defensive efficiencies from past seasons in an effort to try to predict the future, always use Ken’s handy pre-tournament data sets.


Be wary of deploying final end-of-tournament numbers for this particular endeavor. Saying Villanova or Michigan ranked this or that for offense or defense at the final horn of the 2018 national title game is akin to saying the winner of a marathon had traveled the greatest distance of any of the runners at the moment he or she finished the race. Winning four or five or especially six games in the field of 68 can help your whole-season numbers. Conversely, if we want to try to parse out in advance which teams will receive that boost, pre-tournament figures are most helpful.