Defense Through the Non-Conference
Following Duke’s November 24th 91-90 win over Vermont, Duke’s defense was horrid enough to justify asking the following question: Is Duke’s defense good enough to make then a reasonable Final Four contender. At the time, the easy answer was no. Duke’s KenPom defensive ranking had ballooned to over 100, which is considered awful (if it needs any explanation). Although I was optimistic that Duke’s defense would improve, I was admittedly concerned about Duke’s chances going forward. On one hand, Duke’s defense could not get any worse, and its Vermont performance was likely an outlier to some degree. On the other hand, Duke has little size on the interior and lacks elite quickness on the perimeter, and no reinforcements were coming.
Five weeks later, Duke’s 2014 prospects seem quite brighter. Following the team’s best defensive performance of the season against Elon, there at least seems to be a reasonable probability that Duke’s defense will be good enough. With the #2 ranked KenPom offense, Duke’s defense needs to be just good enough to prevent the opposing team from regularly getting easy baskets, which was very much the case early in the season. If you buy into the concept that Duke can make the Final Four with a top 5 offense and a top 50 defense, you only have to look back to last season for comparables (ranking-wise, not necessarily personnel or style-wise).
If you are trying to make the argument that an elite offense can carry a lackluster defense to the Final Four, the best evidence for that argument would be last year’s Michigan team. That squad lost to Louisville in the National Championship. The 2012-13 Wolverines had the #1 KenPom offense and the #48 KenPom defense. The next best comparable is the 2012-13 Duke squad, which made the Elite Eight. That team had the #5 KenPom offense and the #31 KenPom defense. Going further down the line, Gonzaga and Indiana were the 2 other top teams that were primarily carried by offense. This year’s Duke team has the #3 KenPom offense and the #48 KenPom defense. The preceding examples are not proof that Duke will make a long NCAA tournament. They are only data points. However, we do know two things, the first with certainty and the second with strong likelihood. The first: There have been examples of similarly designed teams that have had success in the tournament. The second: Duke’s defense is improving.
There is research that suggests that if you want to have a strong chance of reaching the Final Four, you want to be ranked in the top 20 in both offense and defense. There is no hard rule. It is not as if there is some cliff at #21. It comes down to common sense. Teams with strong offenses and strong defenses are difficult to beat. Even if we can look to last year’s Michigan team for hope, it is better to change the narrative altogether by just playing better defense. I do not mean to suggest that playing defense is simple. However, there is strong evidence that Duke’s defense is improving.
The above graph shows the number of points per possession that Duke has allowed on defense in each game this season. The blue line across the middle is the unweighted average of Duke’s defensive performance this season. The bottom two lines are the Top 20 threshold and Ohio State, the #1 KenPom defense. The size of each blue circle is determined by the quality of the opponent’s offense: the better the offense, the bigger the circle.
The Vermont game stands out immediately. Duke allowed more than 2 points per possession to a mediocre Vermont offense. In no other game this season has Duke come close to playing that poorly on defense. Since then, Duke’s defense has been better, especially against Michigan, UCLA, and ELON (and maybe Arizona, depending on your standard). Presently, Duke’s defensive performance (alongside KenPom’s various adjustments) put the team on the edge of a top 50 ranking. Although earlier I provided evidence of teams of similar relative defensive quality to Duke that had NCAA tournament success, ideally that Michigan comparison would stop being applicable as we get close to the tournament. If the recent trend is in fact a trend, then Duke’s team defense on a game by game basis is approaching top 20 quality.
If we think that Vermont is so much of an outlier that we can remove it from consideration (something I am wary of doing), Duke’s defense looks considerably better. In the above graph, I have excluded the Vermont game. After doing so, Duke’s defensive performance improves all the way from the high 40s overall to the edge of the top 20. I would not worry too much about the cardinal value of the number. I do not use all of KenPom’s adjustment, so the numbers don’t align perfectly. KenPom uses pre-season and strength of schedule adjustments that I do not have access to. The point is that by excluding the Vermont game, I can make Duke look like a top 20ish defensive team. I do not like to pick and choose data points. However, I think it is useful to see how much better Duke’s defensive rating would be in an alternate universe in which that game never happened.
As of 1/3, KenPom ranks Duke 13th but has the team within shouting distance of the top 10. Without the Vermont game, Kenpom would very likely rate Duke much closer to its poll ranking than it does now. More importantly, if Duke continues playing above average defense, that Michigan comparison may become inapplicable by March, and Duke will be one of the top teams entering the tournament.