Duke's Defense, Opponent Shot Selection, and Game Theory
Thomas and I have discussed at some length how we each of us expects Duke’s defense to perform over the remainder of the season. While both of us see defense as a major concern for this Duke squad, we have different expectations about how Coach K’s defense will perform going forward. While Thomas was (and I assume still is) relatively pessimistic about Duke’s chances of substantial improvement, I was somewhat more optimistic about Duke’s ability to produce consistently, acceptable defensive performances. My optimism was supported by Duke’s improved post defense against Arizona. It is with this optimism that I began delving through freely accessible play-by-play statistics. Unfortunately, what I found makes me somewhat less confident in Duke’s ability to improve than I was previously.
It is no secret that opposing offenses have aggressively taken advantage of Duke’s sub-par interior defense by attacking the paint possession after possession. Opposing coaches have said as much in post-game press conferences. However, if we were to take the opposition at its word, we might expect to find that opposing offenses were taking an outlandish number of shot attempts from inside the paint. Consider this: Excluding FTA, shot selection can be broken down very simply into 3PA, 2 point jumpers, and shots at the rim. Similarly, we could also break down shot attempts into 3’s, 2’s from outside the paint, and 2’s from inside the paint. For now, we will stick with the former.
Although not all three shot types are equally likely to occur (which would have a distribution of 33.3% for each type), it would be surprising to see a defense give up 50% or more of its shot attempts at the rim. It is very likely that such a defense would be terrible (although not guaranteed). Fortunately, Duke’s defense does not come close to approaching 50%.
Of the total shots it surrenders, Duke allows 38.0% of them to come at the rim. Above, I have provided the same information for teams commonly associated with Duke. As you can see, when compared to other elite programs, Duke does not permit an outlandish percentage of shots at the rim. Although 38.0% is on the high end, Duke is similar to Kentucky, Mich. St., and Ohio State and well below Kansas in terms of this particular measure. Clearly, Duke’s porous defense cannot be simply blamed on the frequency with which the opposition gets to the rim. However, the bright red box highlights an obviously terrible fact: the opposition shoots 63.8% from the field at the rim. When measured in this way, Duke is far worse at protecting the rim than any other team I’ve listed. Although Duke doesn’t give up an outrageous percentage of chances at the rim, when the team does get to the rim, the results are not pretty.
Interestingly, if we perform the same exercise for 3PA, we get completely different results. As highlighted by the bright green boxes, Duke surrenders the lowest percentage of 3PA (by a considerable margin) and the 2nd lowest 3P%. In terms of results, this could be considered a good thing. 3PA are possibly the most efficient shot in the game and limiting attempts is the best way to stop long distance shooting. I say that limiting attempts is the best way because Ken Pomeroy has shown pretty convincingly that defenses have little control over 3P% over time. Duke 27.0% 3P% against, while great, is likely unsustainable. This snapshot of Duke’s defensive performance tells us that the opposition is taking an unusually low percentage of 3’s and is making an unusually low percentage of its attempts.
So far we have 2 pieces to Duke’s defensive puzzle. Teams shoot a very high percentage at the rim but do not get to the rim insanely frequently. Teams shoot a very low percentage from 3 and subsequently do not take many 3PA. Finally, we have 2PT Jumpers.
Duke gives up the 2nd highest percentage of 2PT jumpers and is in the middle of the pack in terms of FG%. Depending on the length of the jumper, you might want the opposition to take as many jumpers as possible (if you have to pick among the various shot types). Therefore, Duke’s high percentage may not be such a bad thing. Similarly, Dukes FG% is pretty close to the mean. The caveat is here that it is possible that the opposition is taking very short jumpers. If that is the case, then these figures might be misleading.
Taking this all into account, we can say the following about Duke’s defense: Opponents have devastated Duke at the rim, shooting nearly 64%, albeit at a somewhat reasonable frequency. On the other hand, Duke has not suffered at the 3 point line, giving up both a low percentage of attempts and a low percentage of makes. Finally, 2pt jumpers have not hurt Duke more so than they have hurt comparable teams. My instinct is to say that Duke is going to see a regression to the mean, particularly at the 3 point line. Duke simply cannot sustain a 3P% against of 27.0%. Feel free to disagree, but if so, you are going to be woefully disappointed with the 33ish% Duke sees the rest of the season. If this happens, Duke could be in a lot of trouble if the team cannot correct its various defensive flaws. This is why my skepticism concerning Duke’s defense has grown. Duke’s defense has been shredded while opponents have only shot 27.0% from 3. Is it really reasonable to expect this continue? I find it unlikely that Duke has some special skill that keeps long distance shots from settling at the bottom of the net.
However, Duke’s over-achievement at the 3 point line might be indicative of a potential strategy that Coach K could use to correct his defense. By definition, there is a tradeoff between preventing 3PA and keeping the offense out of the paint and away from the rim. It is possible that it is worth it for Duke to back of the 3 point line, allowing a higher percentage of 3PA, while stuffing up leaks in the paint. As I said earlier, I expect that 27.0% to rise, so allowing more 3PA will come both more attempts and likely a higher percentage of makes. Normally, this result would be very, very bad. I say normally, because it is possible that Duke’s interior defense is so bad (hence the 63.8% FG%) that the boost from clogging leaks to the rim will be greater in value than the loss from allowing more 3PA. Although this idea is simple and typically counterproductive, it might be the best Duke can do, apart from unilaterally knocking that 63.8% down to 50%.
In a more general sense, what I’m referring to is a type of game theory. We have two actors: the offense and the defense. The outcome of their interaction is dependent on the choices they make along the way. If Duke is actively trying to suppress 3PA, they’d be doing so at the expense of interior defense. If as a result of Duke’s 3P strategy, opponents are attacking the rim and giving up decent or better 3P looks, then Duke’s low 3P% against is more related to the offenses’ choices than it is to Duke’s defense prowess. Maybe teams have so little respect for Duke’s interior that they have all but given up shooting 3’s and only do so when Duke can manage to keep them out of the paint. If this is the case, that 27% tells us nothing good about Duke’s offense. All of this impacts what we can expect from Duke going forward. Although I still optimistic about the team going forward, I am a little less so today.