The Weekly Nichols: Introducing Composite Score, Position-Adjusted Classification, and Value Rating
Over at Basketball-Statistics.com I have been updating my original statistics for a little while now. Those include Composite Score (CS), Position-Adjusted Classification (PAC), and Value Rating (VR). I have created each of these statistics for a different purpose, and while they will remain perpetual works in progress, I feel that each measurement has some value. Today I’m releasing the final 2008-09 data for every player in the NBA. You can find it all here: http://spreadsheets.google.com/ccc?key=r1lEAryxcq1hiqzsjGpuJAA.
There’s a decent amount of data on that page, so allow me to give a brief explanation of each, starting with Composite Score. For full explanations of everything, you can always go to my web site, Basketball-Statistics.com.
Composite Score
Composite Score is my first original statistic and has been around for the last two years. It is a player rating system that includes data on all players going back to the 2003-04 season. It actually started as just a defensive measure, known as Defensive Composite Score (DCS). Eventually I altered it to include Offensive Composite Score (OCS), and the combination of the two (CS). OCS and DCS are each just combinations of three different metrics.
For DCS, those metrics are Defensive Rating (developed by Dean Oliver and available at Basketball-Reference.com), Counterpart PER (which can be found at 82games.com), and defensive plus-minus (available in many places, including 82games.com). To calculate DCS, I first find a player’s z score in each of those three categories based on their position. I then sum up the z scores and multiply them by -10 to get their DCS. A higher DCS is better, and a DCS of zero is considered average. I also include a player’s DCS percentile rank to give you a better feel of how they compare to the rest of the players in the NBA.
OCS is calculated in the exact same way as DCS, except with the offensive counterparts to the three ratings used for defense. That means that OCS includes Offensive Rating, PER, and offensive plus-minus. Composite Score is simply the sum of a player’s OCS and DCS.
For a more detailed explanation as well as a discussion of CS’s limitations, go to http://basketball-statistics.com/aboutcs.html.
Position-Adjusted Classification
PAC was developed just a few months ago with the purpose of being a scouting tool. Although most people are familiar with the styles of NBA players, PAC provides a snapshot description. This can be useful when trying to find a list of certain players that play a certain way, or when trying to learn about an unfamiliar one.
Like with most of my systems, the way I calculate PAC is quite simple. For each player, I gather their statistics in five categories: Pure Point Rating, Jump Shot %, Rebound Rate, FTA/FGA, and Usage Rate. I then determine if they are high or low in each of these categories based on their position. Depending on how they rate, they are assigned to one of 48 classifications (divided by usage rate: high, medium, or low).
For a more detailed description and a list of limitations, go to http://basketball-statistics.com/howpacworks.html. If you go to http://basketball-statistics.com/whatpaccandoforyou.html, you can read about the potential uses of PAC.
Value Rating
The final statistic I will discuss is called Value Rating. It is a measure of a player’s “value” based on their Composite Score and current salary. To calculate it, I simply subtract a player’s Composite Score rank from their salary rank. For example, Paul Millsap has a very high salary rank (something in the upper 200’s because he is paid so low). However, his Composite Score rank is quite low (one of the best in the league). Because he is paid so little but plays so well, his Value Rating is number one in the NBA. Meanwhile, a player like Allen Iverson, who has a high salary but low performance, has a very low Value Rating. VR is expressed as a percentage. 100% is the best, while 0% is the worst.
Conclusion
As is commonly known, statistics have plenty of limitations. It is practically impossible to replace the value of solid scouting and basketball wisdom. On the other hand, numbers can be used as an objective tool to supplement the knowledge we already have and in some cases help us make decisions. I have developed these statistics because it never hurts to add to your toolbox. With that being said, I understand my numbers aren’t perfect. They’re just another metric to be added to the world of adjusted plus-minus, PER, Win Shares, etc. Feel free to comment with any suggestions, because I’ll be the first to tell you that none of these numbers are without their own weaknesses.
