Quote:
Originally Posted by SittingDuck
Hmm... I used the search function on this thread for 'AI Evaluation' and couldn't get any results.
Anyhow, I am curious as to the settings in the AI Evaluation for player value (from the NABA). I am assuming these are SkyDoggie's values, no? And what I am particularly interested in is what findings were made about using minimal settings for prior year stats, focusing the majority of value on ratings and also on current year-only stats?
I would be obliged...
|
See posts 128 & 129:
Quote:
Originally Posted by Mittal
Great work skydog and Ronco 
Quick question though, what exactly do the AI weights (Towards ratings, current stats, previous stats etc) effect? Basically, does changing those affect the league totals/mess up my stats?
I'm wondering because I do like to have some weight on previous year and 2 year stats, because if a player had a monster season 2 years ago or last year, but stinks this year (And his ratings have dropped), in real life some teams still will keep putting him out there expecting a bounce back and so on. |
Quote:
Originally Posted by SkyDog
1. Changing them won't mess things up drastically. It will just cause the AI to use a few players longer than it probably should.
2. The problem with that is that it makes the game easier (if you're using ratings). I look more at ratings in making my roster/lineup decisions, because I know that performance follows ratings. If a guy who had an .850 OPS last year for me had a big ratings drop-off, I'm using someone else. I want the AI teams not to be at a competitive disadvantage due to sticking with guys that I never would stick with.
|
I'd say SD is right in his reasons. However, if you're going to base any of it on this year's stats, then I'd base more than just 1% on the previous stats. I could see going with 100% on ratings, but going with 65/33/1/1 doesn't make much sense to me.
Players do have up and down years so if you're going to go by stats at all then I don't think all of your stat weight should be on the current year. Yes, it probably makes a lot of sense to put more on the current year than previous years, but that much? I can't remember what I used to go with, but if you want 65 on ratings, which sounds like a good idea, 65/20/10/5 sounds better to me than 65/33/1/1.