Throughout this three year effort I have always wanted to take time to make time to review the process and look for ways to improve the output while retaining the integrity of the End State (create an Index that matches, as close as possible, the League Table without using points earned).
A critical part of this has always been to ensure that the data points used within the Index had relevance (made sense) to how the game is played.
For three years my data points within Possession with Purpose have been:
- Passes Attempted across the Entire Pitch
- Passes Completed across the Entire Pitch
- Passes Attempted within and into the Final Third
- Passes Completed within and into the Final Third
- Shots Taken
- Shots on Goal, and
- Goals Scored
My new and improved PWP Family of Indices will continue to leverage these relevant data points but I am making a modification with respect to the measurement of quality given those data points. The new modifications end up seeing the overall measurement of PWP being:
- Possession Percentage
- Passing Accuracy across the Entire Pitch
- Passing Accuracy within and into the Final Third
- Percentage of Passes Completed across the Entire Pitch versus Passes Completed within and into the Final Third
- Shots Taken per Passes Completed within and into the Final Third
- Shots on Goal per Shots Taken, and
- Goal Scored per Shots on Goal (times 2)
The two categories making up the new Index composition are highlighted in boldface font…
Well for me – in how PWP has developed – I don’t think I quite captured the mroe significant intent of a team to penetrate (given any style of attack – direct, counter, or short pass type of engagement given conditions on the pitch) nor do I think I really captured the considerable value of a goal scored – in any fashion (be it in run-of-play or via set-piece).
I don’t think this violates the integrity of the general tendency of teams and their behavior – I think it actually better represents the importance (weight) of a goal scored as well as the considerable advantage some teams show in being mroe accurate (in passing) as space on the pitch diminishes.
Finally, in making this adjustment I don’t violate the integrity of the original data points collected – I just am finding a better way to translate that quantity of information into a different output relative to quality.
So how do these changes manifest themselves in the data outputs? I’ll let the diagrams and Correlation of Coefficient (R) speak for themselves.
Major League Soccer 2014: (Before and After)
English Premier League: (Before and After)
Bundesliga: (Before and After)
La Liga: (Before and After)
Major League Soccer 2015: (Before and After)