A Myth About Assists

One of my biggest pet peeves is when people just spout "conventional" basketball wisdom without bothering to check the most basic facts. Conventional wisdom in just about every field of knowledge is usually riddled with falsehoods and logic traps, and basketball analysis is no exception. Today I am going to tackle one that's really been getting to me: the myth that a player's teammates' shooting percentage plays a large role in the number of assists a player racks up. Whether it's in defense of Kyrie Irving or Deron Williams, or in denegration of Ricky Rubio, I hear it all the time: "player X's assists are only so low because his teammates cannot shoot.  If he played with better shooters, he'd have way more, I just know it."

Apparently, Rubio's assists are just because he plays with Love

This argument is horrifically flawed on so many levels. Let's start with the logical ones before I move on to plain old boring MATH.

First, by definition, assisted shots have 100% FG. An assist is the pass before a field goal. The reason we keep track of assists is because, in theory, the pass before the basket allows the player to take a better shot than he would have gotten without the pass. We don't keep track of "attempted assists" but the idea behind assisting in general is that good passes lead to open shots. Open shots have higher FG%. So we keep track of assists in an effort to track a player's ability to get his opponents open shots.

But when players make the argument "his teammate's can't shoot", they are using total FG% (if they bother to look at any stats at all). And what they really mean is "his teammate's can't hit open shots when player X gives them one." But of course, total FG% does not come close to measuring this. Most players make lots of field goals that weren't assisted on, which tautologically means that they attempt lots of field goals that are not the result of a pass. This means that players take lots of shots that are much harder than the open shots they get right after receiving a pass (the shot attempts that would lead to an assist). Using a player's (or team's) total FG% as a measure of how likely they are to hit an open shot when passed to is ridiculous; a random player's FG% when wide open under the basket and receiving a pass, or when wide open on the wing receiving a swing pass, is obviously radically different from his FG% when he's trying to break down a defender in an isolation play, or when he's grabbing an offensive board and trying a contested put-back, or when he is posting up against a defender. And a bad FG% might reflect something as simple as "this player takes lots of bad shots" or "Most of this player's shot attempts come from 1-on-1s" rather than "this player can't shoot well".

A classic example is Dwight Howard. He's a terrible shooter, as evidenced by his FT%. But that's irrelevant because when he shoots he is right under the basket, a lot of the time he's dunking or just laying it in, and a large amount of the time he's catching a pass when he does so. The flip side is Kobe Bryant. If you told any defender to just leave Kobe open to shoot whenever he catches the ball, then he'd ask to have some of what you're smoking. Kobe is a great shooter. But most of his shots come from isolations; he's rarely the recipient of a pass that leaves him wide open. So, naturally, Kobe's FG% is lower than Dwight's. And still, neither palyer's FG% reflects his ability to hit open shots when you pass to them, which is much higher.

Ok, so let's forget all of that. Let's assume instead that I buy your silly argument that player X's teammates cannot shoot. Let's instead use some MATH. The average PG last season recorded 8.8 assists per 48 minutes, and the average team's field goal percentage was 45.9%. The best was Boston, with 48.7% and the worst was Milwaukee with 43%. So this gives us a reasonable range of how well (or poorly) we could expect a random player's teammates to shoot.

The average team shot 81.2 FGA per 48 minutes, which gives us a good idea of the average pace so that we can hold this constant. BUT, of course, those 81 FGAs include those that are imaginary point guard is shooting (16.4 per 48 minutes), so an average point guard's teammates take 64.8 shots per 48. So teammates shooting 43% would get 27.8 field goals per 48 minutes, the teammates shooting an average FG% would get 29.7, and the teammates shooting 48% would get 31.5 field goals. Note that we're not measuring offensive efficiency here, because only the number of extra field goals has an impact on assists; we aren't counting the extra points from 3 point shots or free throws, since neither of those affect the number of assists (an assist that leads to a 3 doesn't count as 1.5 assists, and a pass that leads to free throws isn't recorded as an assist).

So, swapping out the average PG's average teammates for the best shooters in the league would mean that his tammates would score an extra 1.8 field goals per 48 minutes. Even if we made the patently ridiculous assumption that ALL of the extra field goals were assisted by our imaginary point guard, he'd get 1.8 extra assists per 48 minutes. This is a nice boost, but it hardly teleports Luke Ridnour into the same area code as Steve Nash.

But of course, it's obviously much worse than this for our poor imaginary PG. Of the 64.8 shots that his teammates take, only a portion of them are after passes from him. A good portion of those shots come from isos, or from passes from other teammates. And of course, whether or not those shots go in has no bearing on our PG's assist rate. Even if we are generous and assume as many as a third or even a half of those 64.8 shot attempts are off of passes from their point guard, and even if we bought the theory that raw total FG% was a good measure for this kind of thing, going from the worst teammates to the best teammates is only going to impact a point guard's assist rate by something like a half to one assist per 48. Again, a nice bump but hardly transforming a player into the next John Stockton.

In reality there are other factors, like pace (more shots means more possible assists), transition offense (faster pace does not always equate to lots of transition buckets, ask the 2010-11 Timberwolves), offensive schemes (point guards who play in schemes like Minnesota's or Phoenix where much of the offense is predicated on the point guard playing the high pick and roll will have many more opportunities for assists than point guards who play in a triangle or in isolation-heavy offenses like the Knicks' or Lakers'). And yes, how well your teammates shoot matters, just like playing next to good rebounders will impact your rebounding rate.  Just nowhere near to the extent that most people make it out to.

Another factor is the types of baskets a point guard assists on. I mentioned earlier that an assist that leads to a 3-pointer does not get "extra credit". Some bloggers have tried to adress this.  Deron Williams and Ricky Rubio for instance both lead the league in assists that lead to 3-pointers.  This means, logically, that many of the open shots that these point guards get for their teammates are 3-point shots. Many of the open shots that Chris Paul gets for his teammates, though, are dunks, because he's passing to Blake Griffin and DeAndre Jordan. So which is better? I don't think the answer to that is clear but what I do know is that trying to use FG% to answer that question is probably a pretty stupid idea.

Finally, ask yourself this: If teammates' shooting is so important, than why is Steve Nash leading the league by a huge margin in assists/48 when his teammates are shooting 42.7%? And why is Norris Cole so awful at assists when his teammates shoot north of 48%?

Loading...