Quote:
Originally Posted by bigmonk
Well, this is a frightening concept! Are you claiming the computer manager knows about and uses managing techniques of which the game player is ignorant? 
Does the computer manager know how to juggle the abilities of unrated players to properly use them in the game and I have no way of knowing how or why that works? Crackers in a barrel! 
When we started playing APBA or Strat, the cool part was looking at the cards and seeing how the company's ratings of the player compared with our idea of how the player should be rated. Then you played with the card and the ratings directed the course of play. That's what made the game fair. I knew what was on your cards and you knew what was on mine. Then the dice decided the game. 
But if the computer uses ratings we don't know about ... |
I am not sure I would go that far. However, the exact workings of the AI are unknown to anyone that I know of other than the AI team. In the past Markus has explained the AI logic is often different in some ways than what we see. For instance the AI using a scouting system but it is not necessarily the scouting reports a player sees. It is a streamlined version.
I never said the AI uses ratings we don't know about it. That has been alleged but I don't think it is true. What I said is that the position rating is a combination of ratings you can see. For OF, OF range, OF arm, and OF error make up the position rating. For IF, IF range, arm, TD, and error make up the position rating.
The position rating has minimum ratings for 2B, 3B, SS, and CF for a player to get a rating. So a guy who has a 3B rating but no SS rating often doesn't have the minimum TDP for a SS. Doesn't mean he can't play SS with low errors the error rating determines that. It does mean he won't turn many DPs though.
What I think is the AI uses the ratings we see error, TDP, range, and arm to make decisions on who to play where. In addition hitting has a weight as well. I think the evidence is that the position rating which is nothing more than a combination of experience at the position and the arm, error, range, and TDP ratings is not heavily weighted and maybe not used at all by the AI.
I would be careful about the AI not using position rating at all because experience at the position is important. Experience can only be seen in the editor. I don't think the AI uses anything we can't see. I do think the position rating has less effect on who plays were than a comparison of arm, range, error, and TDP do.
For instance if I wanted to rank my CFs. I may not use the fielding ratings. I may put say range as 50% of my metric, error as 45% and arm as 5%. In my metric for evaluating CF I might rank a CF with no rating above a guy with a rating and he would do fine even though he is below cutoff for arm. For a 1B I might weight error at 75%, range at 25%, arm at 0%, and TDP at 0%. I might find an inexperienced first base with a low rating gets in over someone with a higher rating because my metric is different from the fielding rating.
The important thing to remember is that fielding rating is a metric that is comprised of several factors. A low TDP SS may have great fielding stats but you won't see in those stats all the DPs that didn't happen. The game won't rate him at SS because of his lack of TDP. What I think is that the game may still see that the guy has good range and good error and place an unrated SS in the position. I can't say I do this often but in a pinch I might.
Part of the problem with this happening is also not how the AI chooses players but the roster logic. If you have 3 1B in a non-DH league you are going to have play people out of position if you make subs. I typically keep a backup C, 1B/3B, 2B/SS, and CF who can play all OF positions in DH leagues in non-DH I will add another OF or utility infielder. The AI doesn't always do this.