Average Margin Of Victory, Average Margin Of Defeat, and Comparison of Actual To Expected Points

Discussion in 'Statistics and Analysis' started by EvanJ, Jun 28, 2010.

  1. EvanJ

    EvanJ Member+

    Manchester United
    United States
    Mar 30, 2004
    Club:
    Manchester United FC
    Nat'l Team:
    United States
    I hypothesized that if a club's average goal differential per game in wins was greater than the club's absolute value of average goal differential per game in losses, the club would obtain less points than a person would expect using their goals scored and goals allowed and if a club's average goal differential per game in wins was less than the club's absolute value of average goal differential per game in losses, the club would obtain more points than a person would expect using their goals scored and goals allowed. For each club in the 2009-2010 EPL, I used the following data:

    1. Points
    2. Goals Scored
    3. Goals Allowed
    4. Goal Differential
    5. Point % (Points for the club divided by points for all the clubs in the league)
    6. Wins
    7. Draws
    8. Losses
    9. Goal Differential in Wins
    10. Goal Differential in Losses (note that the sum of this and Goal Differential in Wins will be the overall Goal Differential)
    11. Goal Differential Per Game in Wins
    12. Goal Differential Per Game in Losses
    13. Absolute Value of Goal Differential Per Game in Losses (I did this to compare to Goal Differential Per Game in Wins)
    14. Goal Differential Per Game in Wins - Absolute Value of Goal Differential Per Game in Losses (this will be positive for the best clubs)
    15. Goal Differential z-Score (how many standard deviations above or below the mean each club's Goal Differential was)
    16. Point % z-Score
    17. Goal Differential Per Game in Wins - Absolute Value of Goal Differential Per Game in Losses z-Score
    18. Point % z-Score - (Goal Differential Per Game in Wins - Absolute Value of Goal Differential Per Game in Losses z-Score)

    The correlation between 17 and 18 was -0.5062. While not an especially strong correlation, the correlation is negative like I expected and closer to -1 than to 0 (correlation ranges from -1 to 1). Using Goal Differential z-Score may not be the best method of determining how many points a club should should have based on their Goals Scored and Goals Allowed, but I wanted to do something simple to calculate and the correlation was strong enough that I'm confident using another method would also produce a negative correlation.
     

Share This Page