With the (in)famous cap spike in the summer of 2016, the NBA saw a huge increase in the base level of contracts shelled out to free agents. Look at what Solomon Hill ($12mm) and Allen Crabbe (about $19mm) got. Both are solid players, the latter superior to the former, but they raked home deals that would be unthinkable for a 2015-or-prior deal. Moving into this year, with free agency mere hours away, it can be tough to try to gauge the market. Last year teams spent money because they had it, and when the cap for 2017 was less than expected, they found themselves hurt by these lavish deals, I’m looking at you, Lakers. I’m sure teams have learned their lesson, but the fact remains: the cap is higher, so contracts will be too.
I tried to anticipate what these deals would be. Using data form over 120 last summer, I built a multi-linear regression model trying to anticipate what a player’s salary would be on the open market with a variety of different predictors. The various predictors I used were: age, career VORP, VORP from last season, RPM from last season, and WinShares (statistic built off if RPM). I also tried to incorporate injury status and position into the model, yet those predictors were not significant, and were removed from the final model. VORP is based off of Box-Plus minus, which, like PER, calculates a value of rating based off of box-score statistics. RPM (and WinShares) uses advanced techniques to track team ratings when a player is on and off of the floor by adjusting for his teammates and competition. By using both of these metrics, I am combining box-score stats with a metric that tries to capture the “intangibles” a player may have.
Interestingly (at least I think as a student of statistics), there was a significant interaction term included in the model, between WinShares and CareerVORP. While that is not too interesting in it of itself, the coefficient in front of this term was negative. This means that as both of these predictors increase in value, they depend on each other, and in fact combine to lower a player’s salary. My explanation for this is that as players age and pass their prime, they just amassed a lot of CareerVORP. As their production declines in the following seasons, so does their WinShares, as they are declining. This would make sense, as along with age, the model will devalue older players.
There were some challenges with the model, most of which created by the NBA Salary Cap rules. With systems in place such as the Mid-Level Exception, the minimum contract, the max contract, there are bars and limits created that prevent a player from getting paid his true market value. For instance, David West has taken the minimum for a few years now, but he is more talented than a minimum player. However, the model sees what he was paid and reads it into the model. Similarly, a player like LeBron or Kevin Durant is being paid less than what he is worth with the maximum contract, so the model may read in those trends and devalue certain strong predictors these players have. With that said, here is the output from my model:
Because of this, I tried to build in a “premium” factor that adds salary for players who are at least a certain age and have a certain skill value. This value increases as a player’s supposed skill increases. I thought this was a solid way of also incorporating the fact that some players will be over payed by teams to entice them to sign for that team.
Well, here are my projections:
LATE EDIT: I was planning on posting this mid-day on Friday, but heard news of the Ricky Rubio trade. As some may know, I am a big Ricky Rubio guy. The Wolves shipped him off to Utah to clear cap space, presumably to find another point guard. Rubio is making about $14mm this year, which is very cheap for his production. They are receiving a top-14 protected pick owned by OKC. If the Wolves were able to land Kyle Lowry, this would be a good deal. Lowry is a great player and makes this team a serious threat.
However, it looks like the Wolves are going to get Jeff Teague. This is not a good deal.
I think Thibs is too worried about Rubio’s fit with the Wolves composition after the Butler deal. Teague is a better shooter from 3 than Rubio, so it’ll supposedly space the floor better for Butler and Wiggins. Having said that, Rubio (like prime Rondo) is a player that still forces teams to guard him close because of his unique skill set in setting others up, finding driving lanes, and throwing magical passes.
I don’t even think Teague fits that well with this team! Teague shot league average from 3 last season. He is by no means a sharpshooter; so don’t label him as that. His best asset is his scoring, but we already have an All-NBA player, probably the 2nd best offensive center in the NBA (who is going to be 21 next season), and Andrew Wiggins, who probably tops out as an overqualified 3rd option in the NBA, but that is all he needs to be. I don’t think he can be your 2nd scorer, especially considering his poor efficiency and poor defense.
That makes Teague fourth on the totem pole. Teague’s best skill is his scoring, and he won’t even get to use it on this team! He doesn’t set up players like Rubio did, he doesn’t defend like Rubio did, he doesn’t rebound like Rubio did, and he doesn’t bend the laws of physics like Rubio does. To top it off, he’ll make more than Rubio will, my model projects him around $20mm this summer, about $6mm more than Rubio’s deal.
The wolves could have had Rubio and a better free agent (since they would have had more cap), yet instead go for a inferior players in both cases. This is why a coach and a POBO should be different roles. A coach can overvalue a player or be overly concerned with a player’s immediate fit, and make an unchecked decision that can harm a team’s long-term future.
I think that’s all I got in me know on this topic. If the Jazz can keep Hayward, that is an interesting team. Defensively stout with Rubio, Gobert, Hayward, and others (assuming they lose George Hill in free agency).
Happy trails to Ricky Rubio. In some ways, this is a perfect ending to Rubio’s time in Minnesota. He was underappreciated by the fan base and the front office, and this results in the team trading him at a loss.
I just wish Ricky would go to a winning team, and it looks like he might finally have that. The Wolves didn’t deserve Ricky, and when the team finally got substantially good enough to compete, they deal the one player who competed harder than anyone else in the franchise the past 6 seasons. Fitting isn’t it?
Excel File: freeagency