Statistical Player Value, SPV

Statistical Player Value, SPV is a new approach to understanding the accomplishments of individual players. Creator of SPV metric, William Benton explains his metric thoroughly.

Basketball is rich with history, rich with talented players, and rich with data that helps to track player accomplishments in games.  But it has always lacked a cohesive measure that allows the value added by players of differing skill sets to be compared as apples to apples.  I am attempting to create just such a way of tabulating the value that a player provides, and keep it simple enough that people can understand the elements that affect the metric; in other words, keep it simple.  What I have come up with is SPV, or Statistical Player Value, which uses a combination of individual and team stats to quantify the accomplishments of each individual player.

What is value?  One of the problems with creating a system of measuring value is that you have data representing very different things that need to find commonality so that they can be measured on the same scale.  The other challenge is living only within the data that you have and applying a value that is properly represented by that data.  Basketball gives us a bunch of data to work with that has been consistently recorded for many years.  The farther back you go the less data you have, but there are some essentials that have been recorded as long as the game has had professional leagues.  I am using the following individual stats to calculate value.  I will explain with each one how I am using it.

ONE BIG LEAP – The most logical way to calculate value for stats is to turn them into an equivalent number of points, and this is exactly what I have done, except for the single largest contributing stat for every player: minutes.  I very arbitrarily assign 1 minute to equal 1 point.  I didn’t do any tricks with statistics or math to validate that equivalence, I simply declare it to be true and I move forward.  If there is a weakness to my methodology it is that, but this is my reasoning: Each team needs at least 240 minutes played during a game, and players compete to be awarded those minutes.  It is then up to the players to take the minutes they are awarded and create additional value for the team in the form of points, rebounds, assists, etc.  Players who get minutes are awarded those because their coach believes that is the best option for winning games.  So a defensive expert who does not put up points will accumulate value in playing long minutes, along with value from the stats that indicate strong defensive play: blocks, rebounds, steals.

SPV – Statistical Player Value is calculated for each game, and is the sum of several different elements from a player’s stat line for the game.  How each stat is converted to points differs, and each is explained below.  The stat elements included in the SPV value are Minutes, Points, Rebounds, Assists, Steals, Blocks, and Turnovers.  Each of those stats (except minutes)  is converted to points using a rational conversion factor, and then all of them are added up to create the SPV for that game.

TEAM & LEAGUE FACTORS – The modifiers for the individual stats are in part dependent on team statistics.  The individual impacts the team, so their contribution to a team metric can help or hurt their own numbers.  This means players who don’t get individual stats, but help a team overall get the benefit of the team doing well, and players on a team that performs poorly get less benefit from the team metrics. Note that the use of team data means that stats recorded in the beginning of the season will drift as the season continues and the team numbers used as factors change.  As a result, the numbers are not final until the season has ended.

PPB – Points Per Basket takes the total points scored by a team and divides it by the number of baskets they made.  This means that all the points scored from free throws are included in the PPB calculation.

PSPP – Points Scored Per Possession is the team’s Offensive Rating (points per 100 possessions) divided by 100.

PAPP – Points Allowed Per Possession is the team’s Defensive Rating (points allowed per 100 opponent possessions) divided by 100.

DB – Defensive Bonus is a team statistics based on a league statistic.  The team with the worst PAPP in the league is the base.  Each other team’s PAPP is subtracted from that to find the extra defensive bonus awarded to each rebound made by that team. This is not a huge number, but it helps defensive specialists who make their team better to get some extra value for each rebound that they make.

LPPB – League Points Per Basket is a league-wide stat used as the individual multiplier for blocked shots.  It is calculated using the sum of league points and dividing by the sum of league baskets made, similar to the PPB, but at a league level.

INDIVIDUAL FACTORS – this is how the player stat line is used to calculate the SPV.  Each stat has a different rationale for the number of value points it is awarded.  This is how each is calculated and why.

MIN_VAL – One minute = one value point.  At the most basic level minutes equals value.  Or you could say that minutes equals the potential to add value.  The more minutes a player has, the more opportunity they have to add value.  You can only add statistical value by being in the game.

PTS_VAL – One point = one value point.  This is the easiest to explain.  A player can score 2 point and 3 point field goals during active gameplay, and they can score 1 point free throws during stoppages for fouls.  You win by scoring more points than the opponent.  Scoring is always the first player stat that people look at.  No need to change what is not broken.

REB_VAL – One rebound = PSPP + DB.  A player scores a rebound by securing the loose ball after a missed shot.  What is a rebound worth?  It does not add points to the score in and of itself.  What I have decided to do is award the value of a possession and we know that is the PSPP for the team.  And I supplement that with the DB for individuals on teams with superior team defense.

AST_VAL – One assist = PPB.  An assist is awarded when a pass from a player directly leads to a field goal being made.  Assists indicate that the player is an effective part of the offenseThere are more points generated by players who are making assists than are measured by the number of assists.  How often is the player fouled on the shot either causing a miss or becoming an and-one opportunity at the foul line?  More assists mean more team basketball, and that is something to be encouraged.  For those reasons, I score the assist higher than you may have first considered, but I think the value of assists as an element of good offense deserves it.  The PPB is always higher than 2 and can be much higher than that on a team that is effective at getting to the foul line or scores more 3 point shots.

STL_VAL – One steal = PSPP + PAPP.  A steal is awarded when a player takes the ball from the opposing team and in doing so gains possession for their own team.  In making the steal, the player has removed the potential score of the opposing team and added the potential score of their own team.  This is typically something more than two points.  A steal is worth more to a poor defense than a good defense, and it is worth more to a good offense than a poor offense.  A steal is always good.

BLK_VAL –  One block = LPPB.  A block is awarded when a player cleanly stops a shot from being scored by an opponent by interfering with the trajectory of the ball before it reaches its apex.  A block is difficult to make and can be demoralizing to the shooting team and motivating for the defending team.  A block is difficult, and for the psychological value, I make it worth more than the points that may have been scored.  I score a block similar to an assist because I believe there needs to be an emphasis on blocking for those defensive specialists who have mastered the art.  Blocks are a minor stat, so emphasizing them really does not move the needle far except for those few players who are good at it.

TOV_VAL – One turnover = (-1) * (PSPP + PAPP).
A turnover is credited when a player makes a mistake that causes their team to lose possession of the ball.  There are also team turnovers, but those are not on a player stat line.  A turnover causes your team to lose possession and the other team to gain possession.  I score this the opposite of a steal.  This is the only stat that counts against rather than for a player, and can actually result in a low minutes player having a negative value if all they manage to do is get on the court and turn the ball over.

SPV – This is the sum of all the stats for each player for each game.  SPV = MIN_VAL + PTS_VAL + REB_VAL + AST_VAL + STL_VAL + BLK_VAL + TOV_VAL

COMPARING VALUE – Once the value for each player for each game has been calculated we calculate variations of the SPV as summary data for the player for that season.  These summary values are then compared to what we call the Average Starter Value, which is the baseline player value for each season.

SPVPG – Statistical Player Value Per Game takes the sum of a player’s SPV and divides it by the number of games they played in.  This shows the typical contribution by that player in games they play, regardless of if they start or come off the bench.  It does not factor for overtime games, so the extra value from extra minutes will impact a player’s SPVPG.

SPVPM – Statistical Player Value Per Minute divides the player’s SPV by the number of minutes they played.  This is most useful in showing the value of low-minute players to see how much they are contributing while they are playing, as they simply don’t get enough playing time to have an SPV of any consequence.  At the extreme low end, this can give statistical anomalies, such as players having a negative value if they come in for one minute and have a turnover.

SPVPTM – Statistical Player Value Per Team Minute divides the player’s SPV by the number of minutes each team has played.  A standard game is 48 team minutes, while overtime games go above that in increments of 5 minutes.  This helps account for overtime as well as different numbers of games being played in a time period.  Players on teams that played more games have more opportunities to generate value, and this helps to account for that rather than just measuring total value.  It also allows us to see the leaders in things like the MVP race during a season where teams have played differing numbers of games, and players have missed differing numbers of the games played by their team.

ASV – Average Starter Value is calculated from all the SPVs recorded by all the players who start each game.  What is calculated is both the mean SPV and the standard deviation from the mean.  All comparisons within a season are to the ASV for that season.  There are actually ASV’s calculated.  ASV per game, ASV per minute, and ASV per team minute.  These become the basis of how a player’s stat line compares to all other players within the same season.  This also allows for the comparison of players across seasons and eras by seeing how they performed versus the mean in their own era.  The ASV is also calculated for the SPVPM and SPVPTM.

WHAT THE MEASURES MEAN  – Each of the measures gives a different perspective.  For a player to be the highest value they should excel at every measure. For players lower on the totem pole, they may only have the opportunity to excel in one way.

SKILL and ACTIVITY – the bench players are best measured using SPVPM, or their value per minute.  Bench players don’t generally get starter’s minutes, so they are best measured by the value they provide during the minutes that they are on the court.  This allows the productivity of a 5-minute player to be measured against the productivity of a 10 or 15-minute player, still using the average starter as the baseline of performance.

CONDITIONING and DOMINANCE – starters are best measured using the SPVPG, or their per-game value.  Starters who don’t play heavy minutes are either out of shape or are not a step better than the bench players on the team.  The standard of measure is still the average starter.  Those players who have SPVPG above the average starter begin to show their ability to dominate a game.

DURABILITY and AVAILABILITY – for the most elite players this is the measure that

Below is the population of all SPV scores for the current season.  You can see it is a well-formed distribution even rounded to buckets of 1.  The orange lines are the mean and the standard deviations from the mean.  The green bar is the SPVPG of Nikola Jokic.  The ASVPG for this season is 61.57 with a standard deviation of 19.82.  The red bar is the current leader for Rookie of the Year, and the yellow bar is the current leader for Sixth Man of the Year.  Both Scottie Barnes and Tyler Hero are averaging better than the average starter, but both are far behind the MVP, Nikola Jokic.

To put Jokic’s SPV line into perspective, his average game this season is greater than 95.05% of all games by all starters.  Or conversely, only 4.95% of all starter stat lines are greater than the average stat line of Jokic.