As player-tracking data has caused a proliferation in the statistics available to evaluate NBA players, part of the challenge is figuring out exactly what the new metrics mean. That's particularly true at the defensive end of the court, where traditional box score stats have always come up wanting. But the new metrics aren't always an upgrade or might not have the assumed meaning.
Consider the case of deflections, tracked by the NBA the past few seasons as part of their "Hustle Stats" on NBA Advanced Stats. Long a favorite of coaches who want their players to be active on defense, deflections have been cited as part of the case that James Harden is underrated defensively and Paul George is one of the league's best defenders. (They rank 1-2 in total deflections.)
However, as I noted on Twitter recently, the interesting thing about deflections is that they tend to track steals almost perfectly. So if Harden and George are already getting credit for their steal rates -- typically also mentioned in these arguments -- adding in deflections is essentially double-counting the same skill. (Deflections are useful in situations where players are turning an unusually low percentage of their deflections into steals. That's not the case for either George or Harden; they're getting more steals than you'd expect based on their deflections.)
So, alas, you won't find any references to deflections in the breakdown of my All-Defensive teams, based on a combination of stats and watching players throughout the season. Instead, I rely more heavily on the defensive ratings of ESPN's real plus-minus, how teams perform with and without the player on the court, block and steal rates and opponent shooting percentages inside five feet (also featured on NBA Advanced Stats using camera-tracking data from Second Spectrum). With that in mind, let's take a look at my picks.