Defensive adjustments to Moreyball

The NBA is an ecosystem, everything is connected and related. Changes in one area inevitably affect outcomes in another. Over the past few years, the biggest systemic change has been the proliferation of three-point attempts and two weeks ago Tom Haberstroh crunched some numbers indicating the ceiling could be moving even higher this season, at an unprecedented rate[1. Through two days of the regular season teams are averaging 27 three-point attempts per game, up from 22.4 last season]. This has had myriad effects from the way players are evaluated, drafted, and compensated, to the way offenses and defenses are structured. One of the other changes that’s long been expected is that at some point the value of three-pointers will decline. If every team is gunning for threes, defenses should adjust and those threes should, theoretically, become harder to make.

Interestingly enough, last season’s league average three-point percentage of 35.0 percent was the second-lowest of the last decade. If we compare the difference in the expected value on three-pointers[1. League average 3PT% * 3] with the expected value of two-pointers[1. League average 2PT% * 2] we see that threes were less valuable (relative to two-pointers) than at any other point in the last decade.

The decline in the marginal value of three-pointers was dramatic last year and there has been a general downward trend over the past decade. This fits with what we’d expect as defenses adjust to the new dominant shot selection patterns. One complicating factor is the general nature of defensive contests. Research has shown that defenses have far more control over an opponent’s three-point attempts than an opponent’s three-point percentage, and that regard NBA defenses seem to be failing spectacularly. The other side of that coin is that offenses may be helping defenses by forcing more contested three-pointers that they wouldn’t otherwise take.

The trend may be complicated by a few other factors as well[1. In discussing this article Seth also raised the possibility that teams like the Sixers and Rockets who hunt threes relentlessly but are below average percentage-wise may be skewing the averages. However, removing them only raises the league average three-point percentage from last season by one-tenth of a percent]. We’re comparing the relative values of each shot and just as three-point percentages are theoretically lowered by taking more of them, two-point percentages should go up as teams take fewer long two-pointers and are generally more selective inside the arc.

The trend shown on the graph below is certainly not definitive and there are too many factors at play identify exactly what is causing what. Also, we’re still talking about a significant margin between shots inside and outside the arc — more than eight points per 100 shot attempts in additional expected value. Still, in two years, as the trend has really taken hold of team offenses, the margin has declined by nearly 50 percent. As we watch teams shatter records and set new precedents for three-point attempts this season it will be interesting to track how that affects the value. This is the nature of advantages, they are discovered, popularized, and eventually degraded as everyone tramples to climb aboard.