Utilizing wOCVA to find the undervalued and overvalued in the NBA

As one season goes away, another one shifts into fourth gear. Just a week ago, the college basketball season ended with the Connecticut Huskies hoisting the national championship in Houston. As you know, my website is based entirely around college basketball, but I also want to branch out a little bit into the NBA spectrum. With the NBA Playoffs getting set to start on Saturday, I feel like there’s no better time to release something that I’ve been building recently. In November of last year, a passion for analytics led me to the creation of a one-number statistic that was designed to measure a college basketball player’s value. 

The aforementioned statistic, Weighted On-Court Value Added (wOCVA), attempts to credit a player for an assigned value of each box-score outcome (steals, blocks, assists, etc.), instead of treating all box-score statistics as if they are on an equal scale. This statistic uses percentage-based metrics rather than cumulative box-score statistics for points, assists, rebounds, turnovers, blocks, and steals. 

The main reasons I built wOCVA were due to an inspiration in baseball’s sabermetrics movement (“Moneyball”), and because I felt there were some deficiencies with Box Plus/Minus across the college game. I won’t bore you with the details, but you are more than welcome to read about the nitty gritty of wOCVA here. Since that article was published, I’ve split up rebounding percentage into offensive and defensive rebounding percentages. I’ve also created offensive and defensive wOCVA ratings for the statistic, combining metrics already used in the main calculation for each side of the ball. 

Everything else remains the same, but we do have a bit more tracking data at our disposal since we are looking at the NBA. I also do not want to come off as an “analytics are the end-all, be-all type of person”, as I believe there is inherently true value in the eye test for player evaluation. I do hope you enjoy learning a bit more about wOCVA and the model I’ve built to translate it for NBA usage. 

---

THE MODEL:

The NBA is a constantly evolving league, with team strategies and player values changing year over year. Whether it’s on the court or off, organizations are constantly evolving through player evaluation. For years, people with no inside access to the association have struggled to understand the value of certain players. As normal basketball fans, we can be taken aback when a player is traded for a package that we view as “too much or too little”, but the reality is that NBA front offices think and operate under a completely different mindset. 

The same principles can be applied when we initially think a player is being “overpaid or underpaid” in the free agency market. Sometimes, we fail to understand the constant evolution of the NBA market, as teams are often paying players with an eye towards the future when payrolls are going to be higher. As basketball junkies, how do we forecast and project the values of NBA players? 

That’s where the model comes into play, as I first took the statistics of over 500 players who played at least one game in the NBA in 2022-23. I incorporated these players into the wOCVA system to calculate their wOCVA rating from the recently concluded regular season. Next, I built in each player’s contract data from the 2022-23 league year, utilizing the HoopsHype Database. 

To better separate players based on their actual roles for their respective teams, I turned to Crafted NBA’s database for player offensive and defensive roles. The creators of the database inform us that these roles “attempt to enhance all-in-one metrics by adding some context to them by grouping players via their offensive and defensive roles.” The Offensive Roles database is based on the work of Todd Whitehead, while the Defensive Roles database is based upon Krishna Narsu’s collection of data. 

Utilizing these databases, I assigned every player an offensive and defensive role. However, I rearranged a few things to make the model run more smoothly and efficiently. I removed several offensive archetypes, including domineering ball handler, tall ball handler, and runaround wing. I felt that some of these were too bland of descriptors or already fit into another offensive role. Below is a colorful graphic created by Whitehead that groups players from the 2018-19 NBA season by their offensive role. 

Those changes leave us with eight offensive roles: Versatile Offensive Big, Tall Spotup, Spotup Wing, Secondary Ball Handler, Primary Ball Handler, Pick and Pop Big, Roll and Cut Big, or Glue Guy. With the roles set forth by Narsu, I only removed the hidden role as I felt like there weren’t enough qualifying players to convey accurate contract expectations. This leaves us with four defensive roles: Versatile Defensive Big, Drop Coverage Big, On-Ball Stopper, and Off-Ball Wing. 

Some players may be assigned an insufficient data role for both offense and defense, due to them playing less than 100 minutes. Think of this role as representing replacement-level talent. Utilizing the eight offensive roles and four defensive roles, the database of players splits into 15 archetypes. 

---

EXAMPLE AND CALCULATION

For the example I’ll be working with throughout the rest of the article, Phoenix Suns reserve guard Josh Okogie is classified as a glue guy and off-ball wing archetype, while Orlando Magic reserve center Moritz Wagner classifies as a pick and pop and drop coverage big. 

The reason that I incorporated all of this contract and archetype data is to further improve the performance of the system of evaluation. Separating players into specific archetypes will allow the system to better project the contract of such a player based on how the NBA has recently valued their archetype on the open market. To calculate how the NBA values a certain archetype, I first added the salaries of every player in the archetype as well as the wOCVA of every player within it. To get the final value, I simply divided the total salary by the total wOCVA, giving me a valuation, which can be read as dollars (in millions) per 1 wOCVA. 

Going back to our example, we can run through the calculation for the valuation of Wagner’s archetype, the Pick and Pop / Drop Coverage Big. Keep in mind, there were nine players with this archetype in the 2022-23 season, including guys like Brook Lopez, Naz Reid, Myles Turner, and Christian Wood. The combined salaries of this archetype equal $102.6 million dollars, while the combined wOCVA comes in at 42.4. With this information at hand, the calculation looks like this:

Archetype Valuation for Pick and Pop / Drop Coverage Big: $102,645,656 // 42.4 = $2,420,888

As stated earlier, this implies that the NBA values archetypes such as Wagner as worth $2.4 million dollars per every 1 wOCVA they will accumulate in a season. For Okogie’s archetype, Glue Guy and Off-Ball Wing, there were 54 players who fit these descriptors this season, including players like Bogdan Bogdanovic, Donte DiVincenzo, Norman Powell, and Jae Crowder. The combined salaries of this archetype equal $272.4 million dollars, while the combined wOCVA comes in at 72.7. With this information at hand, the archetype’s valuation is $3.7 million dollars per wOCVA. 

Now that we have the valuations of these two archetypes, we can set the stage for what type of contract these two players should expect to receive when they hit unrestricted free agency this summer. These two players both recently wrapped up strong seasons for their respective teams, as they each put up 3.3 wOCVA on league minimum deals. 

Okogie is 24 years old, while Wagner is 25 years old, and the two players played similar minutes off the bench this season. The model projects that Okogie should command a contract with an average annual value around $16.1 million dollars, while Wagner’s contractual asks should sit around $11.6 million. Despite their similar resumes, there is nearly a $5 million dollar difference in projected salary, as the organizations within the NBA clearly don't value players like Wagner as highly as they do guys like Okogie. 

---

REVIEW

The formula to calculate a player’s projected salary is actually pretty simple at this stage. Within the formula, we utilize the player’s wOCVA for the season, alongside his age and the valuation of his archetype. As many basketball fans understand, the NBA places a premium on youthful talent on the open market, so not accounting for age would be a fatal flaw for the model. 

Projected Salary = wOCVA * $Archetype Valuation + ($50,000 * (100 - Player Age))

In retrospect, through utilization of contract data, wOCVA, and player archetypes, this model allows us to better understand which NBA players are overvalued and undervalued. Obviously, the model can project massive new contracts for players who are still on their rookie contracts, but it can also help in plenty of other instances. Boston Celtics superstar forward Jayson Tatum, making $30.3 million this season, actually is undervalued by $27.1 million dollars. His fellow Celtic teammate, Derrick White, is undervalued by $14.9 million. Sacramento Kings star forward Domantas Sabonis is an absolute bargain at his current $18.5 million dollar salary, making $25 million under his expected salary. 

While there are countless examples of beautifully aging contracts, there’s quite a few contracts that have aged horrendously. Aside from the obvious ones in John Wall, Russell Westbrook, and Kemba Walker, plenty of once-valued players have been diminished by the evolution of the game. As a Roll + Cut and Drop Big, Minnesota Timberwolves center Rudy Gobert is often referred to as one of the best defenders in basketball, but his actual salary of $38.2 million is $23 million dollars more than he should actually be making. A primary reason for this decline in expected salary is that the NBA is not valuing players with Gobert’s skill set very highly, as he’s been somewhat unplayable in the playoffs in recent seasons. 

Another overvalued player who has been commended for his defensive versatility is Memphis Grizzlies forward Dillon Brooks. It’s not that Brooks’ Glue Guy and On-Ball Stopper archetype isn’t valuable in the modern-day NBA, but rather that he has been genuinely awful this season, putting up a wOCVA of -0.8. That number is the sixth worst-rank amongst all NBA players this year. Brooks will hit unrestricted free agency this summer, and the model projects that no NBA team should pay him above the league minimum. 

Moving forward, I hope to continue adjusting and building with wOCVA as well as this contract projection system. By no means do I think this model is a finished product and I am continuing to work to find its flaws. One of my next projects is to work on adjusting the model to factor in injury history, more than a one-year sample size, and the existence of severe outliers. With 15 different player archetypes and over 500 players, predicting player values prior to free agency can be a daunting task. 

Of course, there are some flaws in this approach, such as injured players who miss significant portions of the season or veterans who are still making a lot of money and not playing much. Nonetheless, my hope is that these projections can provide a useful starting point for NBA teams as they look to build their rosters.

Thank you for reading! If you have any further questions or inquiries, please feel free to shoot me a DM or email me.

Twitter: @tbeckmann24

Email: tbeckmann24@gmail.com

Credits to Basketball Reference, FiveThirtyEight, KenPom, Crafted NBA, HoopsHype, College Basketball Reference, Seth Partnow’s book “The Midrange Theory”, and Synergy Sports.