Quote:
Originally Posted by mgoetze
But you can start off in the right direction by reducing scouting accuracy and scrapping the 20-80 scale in favour of, say, 1-5.
|
I'm of the opinion that the rating scale doesn't matter as long as you are having the AI evaluate solely on production.
When I add even 5% of the evaluation towards ratings, I see the issue with the AI seeing a decline before it actually manifests statistically (the crystal ball.)
I use the 20-80 but I have 0/67/22/11 as my evaluation settings, and evaluate across the entire league instead of by positions. Everyone seems to have even information, and you don't have the mystical genie scouts.
Just my two cents...
(Also is a way to bridge that initial stats-only "shock" to the system that The Wolf spoke of...)