Hi.
Please doublecheck your entries - if I made a mistake, I will correct it!
It's an interesting experiment but unfortunately the procedure is not clear or something since I can't replicate these results. Mine are different. Perhaps you could post an example using the scores of a player.
Well, I don´t want to spend much more time than I did and I thought the rules are quite obvious (just the 33% were stupid, better would have been 33,3% to reduce some rounding errors). But if you insist, send me an example of a different outcome per Discord, not here.