@Jeb I think that (Pa+Pb+New)/3 is a bad (and slightly naiive) way to implement stat breeding since if both parents have perfect stats the expected value for the offspring is 5/6 (if perfect is 1 and worst possible is 0) and getting a perfect stat again would be impossible, as others have pointed out.
A better approach wold be to average the parents stats and add a small random deviation (that can be negative or positive).
An even better approach wold be to take a randomly weighted average of the parents stats and add a small random deviation to the result. This way some offspring may take after one parent more than the other (for each stat).
Pseudo code example: w=randn(); //random number between 0 and 1 r=(randn()-0.5)*some_smallish_number; Pbaby=Pa*w+Pb*(1-w)+r
@Jeb I think that (Pa+Pb+New)/3 is a bad (and slightly naiive) way to implement stat breeding since if both parents have perfect stats the expected value for the offspring is 5/6 (if perfect is 1 and worst possible is 0) and getting a perfect stat again would be impossible, as others have pointed out.
A better approach wold be to average the parents stats and add a small random deviation (that can be negative or positive).
An even better approach wold be to take a randomly weighted average of the parents stats and add a small random deviation to the result. This way some offspring may take after one parent more than the other (for each stat).
Pseudo code example:
w=randn(); //random number between 0 and 1
r=(randn()-0.5)*some_smallish_number;
Pbaby=Pa*w+Pb*(1-w)+r
Hope that made sense.