# NBA Dollars & Sense — Part II: Player value by role

This is the second piece in the multi-part series of posts, Dollars & Sense. This model is the first of a few models that will be combined to create the final NBA Dollars & Sense model. The model described and shown below investigates player value in relation to the game roles that players have on their respective clubs.

When constructing an NBA team, there are a lot of things that a front office must consider. One of the most important aspects of team construction that must be remembered is the different roles that different players will play on a team. Unless you’re the Golden State Warriors, it’s very hard to construct a team with multiple ball-dominant stars. Even if a team can be created with only stars, that team’s depth is likely going to suffer. As a result, teams must be willing to fill in their rosters with solid supporting players around their high-paid stars.

As Part I of this series referenced, different types of NBA players are evaluated in different ways. Role players are coveted based on their ability to do two or three things very well. Additionally, bench players may be valued for their experience or potential, while stars will likely be asked to at least be high-usage scorers, if not more. These same patterns seem to appear in this exercise’s statistical analysis, as well.

More from Nylon Calculus

This portion, and every portion, of the Dollars and Sense Model uses a mixture of every player’s basic (games, minutes played, position, etc.) and advanced statistics (AST%, STL%, TOV%, USG%, DBPM, etc.) from each season since the beginning of this CBA (2011-12). Each player is clustered into one of three clusters, Bench, Role, or Star, based on statistical production. From there, each cluster is regressed on a player’s Percentage of Salary Cap Received.

As referenced, many of the common ideas about player evaluation, based on types of players, are true. The Star cluster equation is most heavily influenced by both field goals made and usage percentage. A Role player’s value is determined by a host of equally-influential portions of a player’s game, like Defensive Box Plus-Minus, assist percentage, defensive rebounding percentage, and true shooting percentage. Finally, Bench players’ value is generally determined by age, age2, and games played. Age in this equation indicates that very young players are paid for their youth and potential, while age2 indicated older players are valued for their veteran leadership.

Once a player’s estimated Percentage of Salary Cap Received is determined, the Actual Percentage of the Salary Cap that player is receiving is divided by the estimated Cap%. This value is known as the player’s Role-Adjusted Salary Ratio.

Actual Cap% / Estimated Cap% = Salary Ratio

This ratio represents the percentage of a player’s estimated salary he is actually receiving, according to the Player Role equations. From here, that player’s VORP (Value Over Replacement Player) is then divided by this ratio. The resulting number represents the player’s Role-Adjusted Salary Value.

VORP / Salary Ratio = Role-Adjusted Salary Value

The reason that a player’s VORP is divided by the ratio is an attempt to model many feelings about bad contracts. According to the model, LeBron James was actually slightly overpaid this season. That isn’t terribly surprising, as this model uses regular season data, and LeBron had one of his less-impressive regular season statistical outputs this season. Additionally, models like these often find the outliers at the top of the model to be slightly overpaid. Nevertheless, David Griffin will lose a lot less sleep over giving LeBron 5 percent extra cap room as opposed to giving a bench player like Mo Williams or Dahntay Jones 5 percent extra cap room. By dividing the player’s VORP by their Salary Ratio, great players who are slightly overpaid will still be seen as good contracts. On the other side of the spectrum, if below-average players are underpaid, they will still come out as bad contracts, but not AS BAD of contracts as large contracts spent on low-producing players.

Looking at the Tableau visual above, you will see a few things. Each column of dots represents a team’s collection of salaries. The salaries can be toggled by team or by year of the current CBA. The dots are colored based on the role in which the model clustered them. The role groupings aren’t perfect. For example, some role players like J.R. Smith are firmly placed in the Bench grouping. Some of these role mismatches are likely due to certain players losing or gaining usage on a new team, injuries to them or their teammates, or any other event that may change player roles on a team.

Taking a look at some of the values, many of the NBA’s often-mentioned high-value contracts are present near the top of the graph, such as the Nuggets’ Nikola Jokic, the Warriors’ Stephen Curry, and the Jazz’s Rudy Gobert. We also see some well-known poor contracts among the Star group, like Derrick Rose and O.J. Mayo. Additionally, we see that some less prominent bench and role players were also on great contracts this past year, including the Clippers’ (now T-Wolves’) Cole Aldrich, the Pelicans’ (now Grizzlies’) James Ennis, and the Hornets’ Marvin Williams. All three of these players were recognized for their production last season and were subsequently signed to much larger free agent deals this offseason. Other free agent deals like these will be investigated in a future part of Dollars and Sense.

More from NC:: Player monetary value

By looking at some of these values, in relation to players with other similar roles on their teams, general managers and other front office personnel can see what kind of value they are receiving for the salary they are paying certain players for that specific season. They can also use past values to try to decide on what values to offer players during contract negotiations in the future.

Up Next: Part III – Player Value by Position