Introducing wOCVA: A college student’s attempt at creating a college basketball valuation stat
With college basketball’s opening night just 10 days away, I feel like it is the perfect time to release something that I’ve been working on in my free time. I’m very interested in analytics and I think that there is a lack of data in college basketball, compared to the NBA. We don’t have the tracking data that the NBA teams do, but we still have a plethora of data at our disposal for research and analysis. Inspired by baseball’s sabermetrics movement and disappointed with some of the deficiencies of Box Plus/Minus, I decided to create a one-number statistic to measure a college basketball player’s value. While I know this isn’t the be-all, and end-all, I hope you enjoy learning about this statistic.
Weighted On-Court Value Added (wOCVA) is a rate statistic that attempts to credit a player for the value of each box-score outcome (steals, blocks, assists, etc), rather than treating all box-score statistics equally. However, wOCVA strays away from using cumulative box-score metrics for points, assists, rebounds, turnovers, blocks, and steals. Rather, wOCVA uses percentage-based metrics for these statistics to ensure that the disparity between high-minute and low-minute players is not out of control in terms of impacting the metric. Minutes disparity is already addressed in the statistic, as a player’s percentage of possible minutes played (MP%) is calculated within the formula.
Instead of turnover percentage, wOCVA utilizes a metric called true turnover rate (TTR). True Turnover Rate is proposed by Seth Partnow in his book, “The Midrange Theory”, as a metric that adjusts for a player’s offensive involvement by accounting for any potential assists. In turn, it more accurately reflects the ability of a player to avoid turnovers, based on their offensive role.
Rather than utilizing traditional shooting efficiency statistics, wOCVA moves forward with the era and uses advanced statistics in True Shooting percentage (TS%) and Effective Field Goal Percentage (eFG%). As more and more teams across all levels of basketball realize the value of the three-pointer, Effective Field Goal percentage becomes inherently more important. In studying the shooting efficiency of a player, this metric factors in the extra point for making a three, while also acknowledging that it is typically a more difficult shot than those inside the arc. True Shooting percentage is related to this metric, but it adds in the value of free-throw attempts to create an estimate of a player’s points per shooting possession.
To understand a player’s ability to score the basketball, I decided to use the simple points per possession (PPP) metric over points scored. Rather than telling me that a player scored 23 points, points per possession tell me how efficiently those points were scored based on the usage rate of the player. Not all 23-point scoring performances are created equal, and wOCVA attempts to account for that.
With all of the aforementioned information, one would assume wOCVA is a stat that goes a little deeper than the box score but ultimately reflects similar data to it. This assessment would be incorrect, as wOCVA utilizes a few other player-based value ratings in accordance with each other. wOCVA uses Basketball Reference’s updated version of Box Plus/Minus (BPM) in an attempt to estimate an individual player’s contribution to his team while he is on the court, based on the data provided by a traditional basketball box score.
Essentially, box plus/minus estimates this player’s contribution to his team per 100 possessions, with team adjustments being accounted for. The team adjustment segment of the metric is critical, as it attempts to accurately reflect that player’s value on his own. This part of the metric attempts to quantify a player’s real contributions, removing the importance of team strength and focusing solely on the individual. This adjustment is key in differentiating box plus/minus from regular plus/minus, as regular plus/minus can be heavily impacted by being the fifth wheel on a great team or the first option on a bad team.
Box plus/minus doesn’t account for the playing time that a player receives. This deficiency brings in the next metric that plays a role in wOCVA, which is that of value over replacement player (VORP). More commonly used in the NBA as a mechanism similar to baseball’s wins above replacement (WAR), VORP converts the rate of the box plus/minus into a player’s estimated overall contributions to his team, compared to a player of “replacement-level” value. Basketball Reference and Tom Tango dive further into VORP and replacement-level value, but the general consensus was that -2.0 was replacement-level value in the NBA per 100 possessions. I slightly adjusted this value to -2.05 for college based on the average VORP of 300 Mid-Major Division I players in the 2021–22 season.
After I find the college basketball player’s VORP, I multiply that number by 2.7. This number is provided by Basketball Reference, described as a translation of “a player’s efficiency differential approximately into wins, using the conversion rate near league-average.” The resulting number is a metric that I will refer to as unweighted wins above replacement (uwWAR), and this metric will play a significant role in the final calculation of wOCVA.
THE WEIGHTS AND FORMULA
This detailed introduction finally brings us to the nitty gritty of the statistic at hand. Aside from uwWAR and MP%, every statistic that goes into the formula for wOCVA has an assigned weight. Most of the weights come together based on an article written by Benjamin Morris, a former sports writer for FiveThirtyEight. Morris created a regression analysis that used each player’s box score statistics in an attempt to predict how different teams were when missing a key player.
The results are incredibly intriguing, as a steal comes out to be worth 9.1 points per game, a block comes out to be 6.1 points per game, and a turnover comes out to cost a team 5.4 points per game. Additionally, each assist is worth 2.2 points per game, while one rebound is calculated to be worth 1.7 points per game. Mathematically and logistically speaking, none of these numbers make sense, but Morris explains that steals and blocks are much more valuable because these two actions don’t just happen for a player, whereas rebounds and assists can happen for anyone.
Replacing a player who creates steals and/or blocks is very difficult because they make these plays happen, and are very skilled at them. Rebounds happen with or without a certain player, as the replacement player doesn’t force these. Steals or blocks may not happen without a certain player, because his replacement may not be quite as adept at forcing them. In summary of this portion, these exact weights are used for the wOCVA formula. These weights are not used for the cumulative numbers for their respective statistics, but rather for the rate-based metrics that we discussed earlier.
Three more statistics come into play for the calculation of wOCVA, and those are TS%, PPP, and eFG%. I didn’t want to overvalue or undervalue shooting efficiency metrics, so I assigned a middle-ground weight of 0.60 to the true-shooting percentage, avoiding overvaluation of free throws. I assigned a weight of 0.75 to an effective field goal percentage in an attempt to avoid the super-gaudy valuation of the three-pointer. My reasoning for this was to make sure big men who didn’t stretch out and attempt threes were not completely disregarded by wOCVA. The last metric is points per possession, to which I assigned a weight of 1.50 because I believe that efficiency in scoring on a possession-by-possession basis is very critical to the success of the team.
With all of the weights established and the statistics defined, the formula for wOCVA is:
((uwWAR+(1.7*REB%)+(2.2*AST%)-(5.4*TTR)+(6.1*BLK%)+(9.1*STL%)+(1.5*PPP)+(0.6*TS%)+(0.75*eFG%)*MP%)/100)*GP
Most of the required data can be found on College Basketball Reference, but you will have to do some of your own calculations for statistics such as points per possession, true turnover rate, unweighted wins above replacement, and percentage of possible minutes played.
THE CALCULATION
Let’s run through a calculation to provide an example of what you can expect. We’ll use last season’s wOCVA leader, Keegan Murray, as an example. Murray, a sophomore forward for the University of Iowa, experienced a breakout campaign that eventually helped him become the fourth overall selection in the 2022 NBA Draft. Here is the full wOCVA formula with Murray’s inputs added. Let’s go step by step to reach the final number.
Basic Formula:
((38.10+(1.7*0.149%)+(2.2*0.101%)-(5.4*4.95%)+(6.1*0.064%)+(9.1*0.023%)+(1.5*1.44)+(0.6*0.638%)+(0.75*0.614%)*0.797%)/100)*35
Step 2:
((38.10+(0.2533)+(0.2222)-(0.2673)+(0.3904)+(0.2093)+(2.16)+(0.3828)+(0.4605)*0.797%)/100)*35
Step 3:
((38.10+(0.2533)+(0.2222)-(0.2673)+(0.3904)+(0.2093)+(2.16)+(0.3828)+(0.3670185)/100)*35
Step 4:
(41.8177185/100)*35
Step 5:
0.418177185*35 = 14.63
As you can see, Murray provided an estimated value of 14.63 points when he was on the court. This number is in reference to his value compared to a replacement-level player. In other words, Iowa was 14.63 points better with Murray on the floor compared to a replacement-level player. It should not be a surprise to anyone that Murray is the leader of the pack from last season, as he was as close to an all-around contributor as it gets, playing in arguably the toughest conference in college basketball.
Murray was clearly an excellent player, as any look at the box score and his trophy case would tell you. However, Murray did not win either the Naismith Player of the Year or the Wooden Award. Kentucky center Oscar Tshiebwe won both awards after a great season. Tshiebwe is a terrific player in his own right, but Murray’s wOCVA was over two points better. wOCVA makes a case that Murray should’ve at least won one of these awards.
LOOKING BACK
When we take a look back at the 2021–22 college basketball season, we see that six of the top 20 players in terms of wOCVA are returning to school. This group is headlined by the aforementioned Tshiebwe as well as Indiana’s Trayce Jackson-Davis. The other 14 players are either playing in the NBA or in the G-League. Eight of them heard their names called in the 2022 NBA Draft, including Chet Holmgren and the previously mentioned Murray as top four selections.
The national champion Kansas Jayhawks were responsible for two of the players who finished inside the top six in wOCVA. Christian Braun finished fifth with 11.26 wOCVA, while Ochai Agbaji finished right behind him with 11.10 wOCVA. Both of these players were a significant part of how Kansas was able to capture its first national title in nearly 14 years.
Since the start of the 2010–11 college basketball season, only four players have recorded a single-season wOCVA of 16 or better. Those four players are Anthony Davis, Frank Kaminsky, Kemba Walker, and Zion Williamson. In fact, Davis is widely regarded as the greatest college basketball player of all time, and wOCVA agrees. Davis’ single season wOCVA of 18.36 is the highest mark in the modern era. Amongst players with 25+ minutes per game, the worst wOCVA in a single season belongs to former Towson guard Kris Walden at -6.11.
Moving forward, I hope to continue adjusting and building with wOCVA. By no means do I think this stat is a finished product and I am continuing to work to find its flaws. One of my next projects is to work on adjusting wOCVA for the strength of schedule and league difficulty, as well as attempting to measure offense and defense separately. 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, College Basketball Reference, Seth Partnow’s book “The Midrange Theory”, and Synergy Sports.