In order to account for the difference in importance of predictions (it is far more important that a prediction is correct for a player that gets 100 points over a player that gets 20 points), each statistical category was divided into an equal number of players.

  • Offensive statistics were divided into 12 equal groups of 25 players.
  • Defensive statistics were broken up into 8 equal groups of 5 players. 

When splitting the players into their different groups, statistical categories (rank and point totals) were sorted alphabetically first and then in ascending or descending order (depended on whether it was rank or point total).   Then the equal groupings were made.  By arranging the players alphabetically, no preference is shown in dividing up the players. This is because at some division points, players with the same rank or point total were put into two different player ranges (Figure 1).  Sorting alphabetically keeps the division of players fair for all forecasters.  

Figure 1. Division of players into player ranges

Individual player point totals were also calculated for determined of total group statistical errors as well as for use in analysis.