View Single Post
Old 08-27-2015, 02:38 PM   #20
Questdog
Hall Of Famer
 
Questdog's Avatar
 
Join Date: Jun 2006
Location: In a dark, damp cave where I'm training slugs to run the bases......
Posts: 16,142
Quote:
Originally Posted by NoOne View Post
using stats to some extent will make decisions a little more life-like. GMs will have a chance to be fooled by randomness(stats) instead of using the concrete ratings, which determine actual probability of success within the game. this assumes 'normal' scouting accuracy or better. i have no experience with lesser accurate scouting in regards to these settings.

if you resim 2015 over many times (50+), and then repeat that process while keeping all settings, players, coaches, the year(2015) the same except for Scouting/AI Eval options, you will see that 100% accurate or normal accuracy for ratings and no stats ai eval results in elevated league-wide offensive numbers when compared to 100% stats evaluation. the proof is in the pudding. this is based on empirical evidence, not personal preference or emotional attachment to one settings or another.

i am not suggesting you use any particular method, but understanding how it influences the ai might help you decide how you want to play the game and how you want it to work.

stats are a derivative of ratings and numerous other factors in the game, including luck. so, if you use stats for evaluation, they do not paint as clear a picture - this is incontrevertable fact. they are mucked up by many factors outside of the player's actual ability. whether or not you prefer that element in the game is a matter of opinion.

lowering scouting below 'normal' and having really poor scouts/scouting budgets may make what i said false. i have never tested this setting in that environment. so, there is possibly a break-even point where stats will start to become a better indicator of future sucess as scouting gets less and less accurate/reliable.
This would be 100% true if the AI evaluated the ratings 100% accurately, which it does not do.

Having a healthy chunk of stats in the AI evaluation helps compensate for the shortcoming of the ratings evaluation routines. It let's the AI see for itself that the fella it thought would do well as starting pitcher really is worthless.
Questdog is offline   Reply With Quote