First, we collected fixture and results data for the Performance Buzz eligible leagues for the past 3 seasons. In cases where there were teams that will be playing in the top division this year that haven’t played in the top division for all 3 of the previous seasons, we replaced relegated teams with these teams accordingly.
Next, we collected two years of performance data for every player to have played in the top 5 leagues over the last 2 seasons. We also cross-referenced this data with each player's actual PB scores last season as well some data provided to us by @BuzzingPaul.
Then we estimated each player's expected score and variability for each game over the 3 seasons of fixtures based on their PB average, the 2 seasons of performance data (weighted towards last season), fixture results, their current club and the variability of scores last season.
Finally, we used this data to simulate each game day for the last 3 seasons 100 times and recorded the winners of each game day every time as well as whether it was a single, double or triple match day. We then aggregated these results to find each player's average dividend returns per season across the simulations.
We have carried out this simulation according to the methodology above however there are some factors that won't be taken properly into account in the results. The main one being is that it doesn't take into account injuries and likely team selections. For this reason, we suggest using the results only as a resource to identify players to do further research into and not a suggestion to buy players purely because of their simulation results. Whilst we made an effort to include transfers in these results it is possible that a) we didn't include every transfer and b) that transfers in the last week after we completed the data collection weren't accounted for. Another limitation is that we didn't have sufficient data on all the players (especially those that didn't play for a top 5 league team last year). In the specific case of Koke the simulation didn't work because of a formatting issue so he wasn't included in the results. We still believe that this is a very useful resource as you look to estimate the dividends of each player.