arbitrary, subjective and biased as you are suggesting my standard of fairness is. The only point that I am making is that there should be a rule that limits the selection criteria to the performance from the current season. Then go and make whatever model you want and make it as predictive as you can. I don't have that model, but if you're suggesting that it can't be done then you must not have much of an imagination.

You are trying to make counter arguments by saying that doing so would decrease predictive accuracy. I concede that it would, especially in the early season. I think by the final rankings, the difference would be negligible, but I am content to sacrifice some predictive accuracy to preserve my "very subjective" standard of fairness.