Freelance Friday – Grading 2015 NBA Free Agent Salaries

Freelance Fridays are back. This is an opportunity for us to open our platform up to new voices and really make the site into an interactive community, instead of just a one-way portal where our writers toss their opinions at you. Once or twice a month we’re going to take an entire Friday and share with you a collection of guest posts written by talented writers and analysts from around the web. If you have an interest in participating, now or in the future, get in touch TheNylonCalculus@gmail.com.

This post is an examination of contract values from this summer’s free agency period from Max Schimanski. Max is a senior at Davidson College pursuing a degree in Economics with a minor in Applied Mathematics. Max hails from the Madison, WI-area, and is a lifetime fan of the Wisconsin Badgers and the Milwaukee Bucks. Through attending Davidson, he has naturally become a huge Golden State Warriors fan (All hail, Steph Curry). You can find him on Twitter at @schimax. 

Deal or No Deal? Grading the Quality of 2015 NBA Free Agent Salaries

Every summer, NBA free agency takes place over a frenzied period of weeks (or even days, as was the case this summer). When the dust settles, the accounting begins, with the assignment of grades for each signing, and debates over teams “winning” or “losing” in Free Agency. These discussions often focus on which teams added or lost the most talent. Less often are deals dissected in terms of whether a deal represented good reflection of a player’s market value.

A player’s “market value” is determined by a number of factors relating to both the market and the player himself. Market conditions can often artificially influence a player’s value. For example, if free agent class has fewer quality point guards than usual, a team looking for a point guard is likely to “overspend” for a quality point guards relative to a more normal market, as that position of need to the team is scarce. Other times, teams may project players to take on a bigger role when switching teams, pricing expected improvement into their valuation. Lance Stephenson was perhaps an example of this when moving to the Charlotte Hornets last summer. Player-specific factors can include age, usage, contribution, personal attitude, among many other attributes.

However, the simplest reason as to why NBA teams sign players is that players help NBA teams win games. Assigning value in terms of wins to individual players is always tricky, but coming up with some measurement of these contributions is necessary if we are going to compare dollars and cents to wins and losses. In this case, I’ve selected Win Shares as the metric which best combined the approximation of that value in a relatively simple to calculate form.

Below, you can see the Win Shares produced during the 2014/15 season by each of the summer 2015 free agents signees plotted against their respective salaries for the upcoming year. In an attempt to model the prices of premium free agents on the market (that is, those that will likely play significant minutes next season), only players who signed for $2.5 million or more in the first year of the new contract are included. Players who signed for $2.49 million or less aren’t meant to be seen as irrelevant players, but the goal of the model is to attempt to mimic front office executives’ decision making process when evaluating talent that will significantly impact their team’s season. Most players, though not all [1. E.G., Josh Smith], under that $2.49 million salary threshold will likely have minimal impact on their respective team’s success next season.

As we can see, Win Shares are a decent predictor of a premium player’s salary, as shown by an R2 of 0.57 between the two variables. However, as noted above, prior production isn’t the only variable in teams’ decisions when deciding upon the amount of money to offer a free agent. A more comprehensive model is needed to predict actual salaries received to see what a player’s true value might have been on the free agent market this summer.

For this exercise, three variables were used to enhance the model, alongside simple Win Shares, to predict “expected contract value”.  The model accounts for age for 2015-2016, games played in 2014-15, usage in 2014-15, as well as Win Shares. With this additional information, the model is able to better predict durability, role, and where a given player might fall on the aging curve.

These variables account for over 80% of a newly signed free agent’s salary, far superior to Win Share totals alone. While this model allows us to see how much over or under prediction each new contract fell in total dollars, a more instructive way to interpret the data is to look at the differences between the two sets of in terms of percentages. “Missing” by a million on a max contract for Dwyane Wade is a far smaller benefit or loss than being off by that same million for a role player such as Leandro Barbosa.

Over half of the newly signed premium free agents signed a contract that pays them within 22% of the market value predicted for 2015/16. However, there are some free agents that have been significantly overpaid or underpaid. For example, this model suggests that Tim Duncan will be worth a salary of about $12.4 million to the San Antonio Spurs next season. Duncan, of course, took a massive “hometown discount” to sign a contract worth approximately $5.4 million next season, in order for the Spurs to re-sign Kawhi Leonard to a max contract and still have room to chase LaMarcus Aldridge.

In order to compare these deals a bit more comprehensively, the percentage point difference between predicted salary and real salary and a player’s real salary are then clustered, based on distance from each other, into five contract value categories: Bargain, Good, Average, Poor, and Overpaid. These clusters represent what kind of a deal a respective team got when signing a specific free agent. By grouping this data into clusters, we now can see which teams “won” free agency in terms of making smart, economical decisions with the money that they spent on players. In some cases, these clusters reflect financial flexibility not readily available to other teams, such as with Duncan and the Spurs. Those sort of factors must be accounted for when extrapolating from this analysis into “grading” signings. Some teams simply don’t have as many bargains available to them, and not grading on that curve is possibly an unfair evaluation of managerial performance.

Looking at the overpaid players, Joel Anthony and Kevin Garnett were the most overpaid players in this 2015 Free Agent Class by a wide margin. Both were likely given a higher salary than market value out of respect for their veteran presence in the locker room. Other questionable signings include Omer Asik, Aron Baynes, and Rajon Rondo. This classification includes mostly backup big men and once-star players who had a down year last season, like Rondo and Kevin Love.

Any player with an estimated salary within 15% of their true salary falls into the Average category, a category that includes stars like LeBron James (who is even a bit undervalued by the model due to the time missed during the early January LeBattical), Draymond Green, and LaMarcus Aldridge.

Good contracts include many players that are fringe starters: guys that haven’t quite proven themselves to be consistent starting players, like Lou Williams, or aging veterans taking a smaller role within a system, like Leandro Barbosa. Duncan is the biggest name in the Bargain bin, a category populated a number of veteran big men.

This model and its results are not meant to be a system of valuing NBA Free Agents that can be applied each and every year, especially given the impending NBA Salary Cap increase in coming years. However, looking at free agency a year-to-year basis, an exercise like this one allows us to judge the free agent market in a specific year, make an educated guess regarding the efficiency of contracts signed in the summer of 2015, and possibly give insights into the types of bets NBA teams are making on players in the future.