Finding top NBA Free Agent targets

Feb 27, 2016; Oklahoma City, OK, USA; Oklahoma City Thunder forward Kevin Durant (35) takes the floor prior to action against the Golden State Warriors at Chesapeake Energy Arena. Mandatory Credit: Mark D. Smith-USA TODAY Sports
Feb 27, 2016; Oklahoma City, OK, USA; Oklahoma City Thunder forward Kevin Durant (35) takes the floor prior to action against the Golden State Warriors at Chesapeake Energy Arena. Mandatory Credit: Mark D. Smith-USA TODAY Sports /
facebooktwitterreddit

In the days that precede free agency, the conversation around the NBA is always filled with questions: Who will get a big pay day this season? Why is Team X interested in Player Y? While Andrew Johnson, the Master of NBA Modeling, has already answered the first question, the NBA community does not always fully consider the second question. Why do teams take certain interests in a certain NBA free agent? Some free agents bring upgrades to a position of weakness, while others may fill spots for players that have retired or moved on to other teams.

This exercise is an attempt to estimate which free agents each specific team should be targeting, based on their need at each position. However, in today’s NBA, classical position labels are becoming less and less valuable. Many players in today’s league are able to switch among multiple positions. A more fluid position estimate is needed to accurately understand the value of a specific player. Using Seth Partnow’s Production by Position statistics, players in this year’s free agent class can be divided up into 7 different positions: Pure PG (1-1.2), Combo Guard (1.2-2.05), Small Wing (2.051-2.55), Large Wing (2.451-3.1), Stretch 4 (3.101-4.25), Flexible Big Man (4.251-4.79), Center (4.791-5). Most of these categories are pretty self- explanatory. Pure PGs and Centers represent players that primarily one position. Combo Guards, Small Wings, Large Wings, and Flexible Big Men, represent players that play one position but can alternate between multiple positions in certain situations.

After determining positions for each free agent, teams must account for what production at what position they need to replace. To do this, this exercise uses Basketball-Reference’s Win Shares value to estimate the average WS production at each of the 7 positions. Once the position average is determined, the averages are normalized to 0. By looking at how far above or below the WS average a team is for each position can determine what players they should target in free agency. The areas where teams do not have a returning player (and no production) or have below average production are indicated as areas of need.

More from Nylon Calculus

Once a team’s needs have been determined, a player-placement algorithm is applied to the available free agents at each position to determine which free agents each team should target. The algorithm includes things like team construction, team culture, free agent age, player positional VORP, and more. For example, a veteran role player that is searching for a ring probably will not want to sign with the rebuilding Sixers, even if they offer that player the most money. The algorithm also includes player remarks about free agency. Because LeBron James has explicitly stated that he will return to the Cleveland Cavaliers, the model assumes that the Cavs have already locked down LeBron. Trades up through the morning of 6/29 have been accounted within the model and algorithm.

The Tableau above shows three visuals. The upper visual contains a box and whisker plot of each team’s normalized average WS production at each position. When filtering by team, if a team does not have any production returning next season in a position, that position will not show up on the graph. The middle visual is a table containing all of this year’s available free agents, their position, their former team, their free agent status, their market value, and their Win Shares from the 2015-2016 season. Finally, the bottom graph shows multiple salary cap figures. The first bar on the left shows each team’s estimated maximum available cap room. Each of the bars on the right show the team’s available cap room after signing their first, second, and third target free agents. When hovering over each value for each team, you will see the suggested first, second, and third target free agents, based on Win Share production at each of the seven positions. You will also see the estimated average contract offer to each of the targets. These contract offers are based on the free agent’s market value, shown in the upper-right table, and the amount of cap room a team has remaining, so some free agents may receive larger offers from certain teams with more cap room.

Free agent market value was determined with my Post-Free Agency model from last season, which I used to rank free agent signings. The model has been altered from determining dollars to determining percentage of cap, something that is particularly useful for the new cap increase. It’s not a perfect model, but using WS, Usage, Age, and Games Played, it does a pretty decent job.

Looking at results, we see a few patterns. First, players with injuries over the previous season, like Joakim Noah, are likely undervalued in terms of market value, as the model doesn’t capture that effect. However, the model does a pretty good job of valuing young guys who may not have gotten their chance to shine, like Allen Crabbe or Evan Fournier. Finally, the free agent suggestions are dependent on the others, so after Team A signs Free Agent 1, then they will move on to signing Free Agent 2. If a team does not sign Free Agent 1, then their free agent needs could change and they could move on to signing a different Free Agent 1. They would not necessarily have to move immediately to Free Agent 2. If you happen to play around with the dashboard or graphs and find something particularly interesting, feel free to mention it in the comments below.

Unless you work for an NBA franchise, you will probably never know what kind of intense planning and decision-making goes into deciding on what free agents to target every season. However, with tools like this one, the public can get a sense of what types of players may fit with teams based on their previous year’s production, a team’s salary cap space, and a free agent’s market value.

Next: A model for NBA Free Agent earnings