It’s been a while since we had a sports post on SomeDisagree, so why not have another one now?
Last night, someone made the argument to me that Russell Westbrook was “the premier defensive point guard in basketball.” Being a rational observer of basketball, I disagreed; to me, it is obvious that Rajon Rondo is the best defensive point guard in the league. When I pressed him on it, he cited a poll of NBA GMs in which 11% said Westbrook was the best perimeter defender in basketball. Interestingly, the plurality (15%) said Rondo was the best perimeter defender, leading me to wonder aloud why he cited information saying Westbrook is second-best to support the assertion that Westbrook is the first-best. But that’s neither here nor there. The supremacy of Rondo had been challenged, so I decided researching pointless basketball stuff was preferable to doing productive work set out to defend his honor. It was time to prove my point with statistics.
(Note: Am I biased in Rondo’s favor because I’m a Celtics fan and Rondo happens to be my favorite player in the league? Yes, of course. But that doesn’t mean I’m wrong.)
The first method of measuring how good they might be defensively that came to mind was to compare how the top point guards in the league fared in games when matched up against Rondo or Westbrook. So I devised a list of the top point guards in the league, consisting of Chris Paul, Deron Williams, Tony Parker, John Wall, Russell Westbrook, and Rajon Rondo. I excluded Derrick Rose and Steve Nash from this list because they were not common opponents of Rondo and Westbrook – Rondo had not faced Nash and Westbrook had not faced Rose. From that list, I took down the points and assists statistics that each top point guard had put up in their games against Rondo and Westbrook, and compared the average of those contests against the player in question’s season average. So, for instance, in three games against Rajon Rondo this season, John Wall put up point/assist lines of 19/8, 11/8, and 12/9, for average production of 14/8. This is 2 points per game less than his season average of 16, but 1 assist per game higher than his season average of 7.
From those average differentials, I compiled an average of averages to see how the top point guards, as a group, performed against Rondo and Westbrook. Here are the results:
Average point differential: -4.6
Average assist differential: -1.4
Average point differential: +1.8
Average assist differential: -0.8
So, by that metric, Rondo is a significantly better defender against top point guards than Westbrook is. But there was a pretty obvious flaw in my methodology: When computing the average point differentials, Chris Paul’s production average against Westbrook, compiled over four games, counted for just as much as his production “average” against Rondo, which was really just one game. There was a clear need to weight the numbers. So here are the results after averaging by game rather than by player.
Average point differential: -4.5
Average assist differential: +0
Average point differential: +2.2
Average assist differential: -0.4
Recalculating the differential in this way doesn’t change things all that much – Rondo is approximately 6.5 points and 0.5 assists better per game than Westbrook in both sets.
So, question settled, right? Well, no, not really, because there’s another glaring problem with this analysis: the sample size is miniscule. Only 10 games of data is not really enough to be drawing wide-ranging conclusions. So, what if I expanded the sample to include all starting PGs on playoff teams? That gives us a player set of Derrick Rose, Mario Chalmers, George Hill, Rajon Rondo, Jeff Teague, Jameer Nelson, Jrue Holliday, Jeremy Lin, Russell Westbrook, Tony Parker, Derek Fisher, Chris Paul, Mike Conley, Jason Kidd, Ty Lawson, Steve Nash, John Wall, and Deron Williams.
But if I were to do that big of a player set, I might as well just do the entire league. And further, that wouldn’t even solve all of my problems. How could this model avoid counting production that was put up while each guy was on the bench and the opposing point guard was feasting on the second stringer? How could it measure how well Rondo and Westbrook did guarding the second stringers on the other teams? How should I deal with the differences in minutes played? And does reducing defense to the raw number of point scored and assists tallied by your opponent really capture everything? What about things like limiting field goal percentage or maximizing turnovers?
Answering this question was starting to look like a monumental task that I’d probably not finish for weeks if at all when I remembered, “Oh yeah, there’s people on the internet who do this for a living. I wonder what they think.”
And, sure enough, the good people over at 82games.com publish a stat section for each player called “Opponent Counterpart 48-Minute Production.” It is what it sounds like: a measurement of what numbers opposing players at the same position are putting up against a player while he’s on the floor, adjusted to correspond to a 48 minute game. So, here’s how our boys did.
Effective Field Goal %: 42.0
Effective Field Goal %: 47.2
So, in conclusion, opposing point guards are putting up a 3.2% higher effective field goal percentage, getting 1.4 more assists, committing 1 less turnover, and registering an extra 3.3 points of PER against Westbrook than they are against Rondo.