Possession-adjusted or PAdj statistics is a more robust method to calculate the true measure of defensive statistical contributions by a player in the game. As it takes into account the measure of possession, we are able to more accurately determine the value of defensive contributions. Think about it this way, any DSTV Premiership team that has a lot of possession means that the opposition does not have the ball very often during the course of a game, therefore the team is highly likely to not accumulate defensive statistics. As a player, you cannot carry out defensive contributions for your team unless the opposition has the ball. Yes, you can block channels, passing angles but unfortunately, these are not measurable (now at least) with only event/on-ball data.
Let’s say a team of high possession such as Cape Town City FC or even Mamelodi Sundowns are in this position, you have to (as best possible) adjust these basic numbers of tackles, interceptions, challenges and so forth for each player. The logic is based on ‘you are only able to make defensive contributions when you are not in the possession of the ball‘. Therefore, the players from low possession-based teams would generally have higher defensive values as they are forced to make more actions (fouls, interceptions & tackles etc). Adjusting these values using another unit of measurement (normally possession, touches or time) will give further insight into the frequency of defensive actions that are made by both sides of the spectrum.
How?
There are numerous ways one can tackle this problem (time, touches, possession etc). We have decided to use possession based on the data we have access to. Using the average duration of football matches which are somewhere between 90 and 95 minutes (minus the time for the ball in play) we are left with about 50 to 60 minutes of actual football being played. This number is taken as the value to adjust PAdj statistics to.
Example:
In the recent match between SuperSport Utd and Cape Town City (2:1), the possession statistics read (40%) or 18:54 (mins) of possession time vs (60%) or 28:53 (mins) of possession time. Two random defenders of each team; Thatayaone Ditlhokwe from SuperSport Utd and Craig Martin of Cape Town City were chosen.
Ditlhokwe, who played for SuperSport Utd made 7 interceptions. His PAdj interception statistic would be 11.33.
Martin, who played for Cape Town City FC made 10 interceptions. His PAdj interception statistic would be 10.52.
Therefore, while Ditlhokwe made 3 fewer interceptions than Martin, their possession adjusted interception figure for the game is virtually even as we take into account the level of possession that Cape Town City had versus SuperSport Utd.
Application
Whilst the raw numbers’ now look like they have been ‘packaged’ very differently, this process has been the industry standard in terms of defensive evaluations for a long period of time. We felt it was only right to make our process transparent as much as possible without giving away all of the secrets, especially around how our data process is designed and handled.