View Single Post
Old 05-19-2017, 03:16 PM   #2
NoOne
Banned
 
Join Date: Apr 2015
Posts: 7,273
Infractions: 0/1 (3)
'they' have said that the current year does consider sample size. nothing specific was stated beyond that, though. have to search for it for exactly what was given... likely markus or lukas posted the info, if i recall? within last 1-2 years in forums... (google search can cut those 1,000's of threads down)

so, if you have X% in current year and it's game 2, it's likely bumping that portion back into the other 2 years, or ignoring it, or ... ??? something else that makes sense.

nothing was specifically said about previous years and how much they were used affecting weighting, however.... could assume it follows the same rules, but maybe not, too. neither if it's about playing time or rather # of days out of days of season length. my money is on playing time (if it wasn't specifically stated, this is by memory*)

you'd have to find some distinct examples in your league and adjust ai-eval and such to see for sure. switch to 100% ratings and 100% accuracy for a baseline, then turn ai eval back on and adjust % stats etc etc... comapre what you see and you can deduce what's going on with certainty.

*Remember, it only affects overall (i assume potential?)... hopefully overall is not something you, as a human, use much. this should be more about how it affects the AI, if you have a concern about it. don't worry about it from your perpsective... you should be mostly ignoring overall on it's own.

Last edited by NoOne; 05-19-2017 at 03:19 PM.
NoOne is offline   Reply With Quote