When Matt Moore did a behemoth of a post a couple weeks back on player usage and PER for every team in the league, I got this nagging feeling, like there was something I was supposed to have done. But what was it? I had paid all my bills for the month, I had brushed my teeth, and I had even remembered to bump the thermostat down for the night. And then it hit me: I was supposed to have done something with the player swap tool I created back in the off-season. Sometime. Something. I had never really decided on what was the first improvement I wanted to make or when I should get around to it. But Moore's terrific post got me motivated (and the upcoming trade deadline was another slight push) and provided the inspiration for the focus of the change.
Efficiency and usage are the key (and practically only) components of the player swap tool. I created it to crudely calculate how the potential addition of Allen Iverson might have impacted the Bobcats performance and then to calculate the impact of the actual addition of Tyson Chandler/subtraction of Emeka Okafor. It's purpose and design were simple: If one player uses this many possessions at this efficiency, how will the team's numbers change if a new player with a different usage and efficiency comes in? And for an off-the-cuff number, it did ok it reasonably reflected the difference in how the Bobcats played with D.J. Augustin on the court for Raymond Felton and it came close to calculating how Orlando slid a bit with Rafer Alston coming in to replace Jameer Nelson. But…
It is kind of a big but it thinks the Bobcats would be worse with LeBron James than Gerald Wallace this season. And as big of a Crash fan as I am, that is preposterous. Now, it is not the tool's fault: It just does not know any better. It was designed to say that if a new player comes in, the rest of the team will continue to perform at the same efficiency they have previously. But when you add a player like LeBron James, things are a little different for one, he uses about 15 more possessions per 100 team possessions than Gerald. All of a sudden the rest of the Bobcats are responsible for splitting just 65 possessions, instead of 80 hmm, wonder if that may make their jobs a little bit easier?
No surprises: It does. A great example is Raymond Felton's performance from last year to this year. Last season, his usage ([fga+0.4*fta+turnovers]/100 team possessions) was 24.7 and his efficiency on those possessions (points per 100 possessions used) was just 82.0. This year, with the addition of Stephen Jackson, Ray's usage is down to 20.7 and his efficiency up to 95.0. Looking across the league, this relationship holds true as usage goes up, efficiency goes down. This is not to say that high usage players are the lowest efficiency players that is not the case. No, what I mean is that as an individual player is called on to shoulder an increased burden, his efficiency drops. The chart below shows the usage versus efficiency on a per game basis for every player who played at least 8 minutes in a given game.
Because it is a big blob of dots and a bit incomprehensible (there are over 11,000 data points in there) I have the line of best fit displayed, along with its equation, indicating the negative relation between the two.
So, now, thanks to Moore's indirect motivation, the player swap tool estimates the impact of shifting responsibilities on efficiencies a team's and a player's. If you replace LeBron with Gerald, the rest of Cleveland has some heavier lifting to do, including Gerald. Both of their efficiencies are expected to go down. However, if you place LeBron in Charlotte (or New York or whatever other fantasy fans may have), is he really going to subjugate his game? I'd be surprised so, his usage will not change in the new tool. The change in usage and efficiency is only done against the rest of Charlotte. A similar but opposite effect is seen when moving a player like Ben Wallace: If Ben Wallace leaves Detroit to go help the Nuggets win the West, is he going to take on a larger offensive load? Doubtful again, his usage and efficiency are predicted to stay static, while the acquiring team would see their usage go up, and their efficiency drop proportionally. The cutoffs for those kind of decisions are based on the top 10% and the bottom 20% in usage guys who use possessions like a primary option are expected to maintain it, while guys who are currently the 6th option of the 5 men on the court remain that way (not a typo an exaggeration just to head off the comments).
With those changes made, sanity is restored LeBron James coming to Charlotte in place of Gerald Wallace makes the Bobcats better. How much better? Go to the improved Player Swap Tool and find out there is a bit less clutter on the page and a nice summary of the projected outcome. And if you want to compare the new results to the old I still have the old tool available here: Original Player Swap Tool
(Oh, and if you see a result that makes you scratch your head, please say so I have done a lot of testing…but bugs may have snuck through).