View Single Post
Old 01-08-2015, 07:53 PM   #13
teachermike
Bat Boy
 
Join Date: Dec 2013
Posts: 6
Quote:
Originally Posted by RchW View Post
I didn't say it was too easy to change positions.
Quote:
Originally Posted by RchW View Post
hence it is too easy to learn a position
You are splitting hairs, my friend. I don't even know what the distinction is that you're trying to make.

Quote:
Originally Posted by RchW View Post
The threshold level of defensive ratings that allows the AI to start a player is too low . Further to that it is too easy to learn a position contrary to the defensive spectrum...In this same league there are several other OF with multiple but bad infield ratings playing as MI. They exist because the AI sees only the bat and starts them over qualified 2B and/or other infielders.
And I already agreed with all of this in my previous post, much of it almost verbatim:

Quote:
Originally Posted by teachermike View Post
I do agree that the AI doesn't do a good job of determining when it should prioritize defense when determining lineups (it's pretty much geared for maximizing offense), and that it has too many players switching in and out of positions when it doesn't make sense. All in all, the positional ratings system feels incoherent, and often doesn't seem to adhere to the defensive spectrum.
I'm not sure, but I think you're incorrect about where this problem comes from. To be clear, I agree with you that part of the problem comes from the AI making poor lineup choices, and that for SS/2B/CF (I'd add catchers) the AI should do more to prioritize defensive players, when available. But, and this is the crux of the change I'm suggesting, I don't think these guys initially developed their 2B rating because the AI erroneously played them there to get their bat in the lineup. If they didn't already have a 2B rating (or at least infield ratings), I'm convinced from my Mookie-to-the-OF trials that they wouldn't develop a 2B rating at all. I think the AI, when generating new players, tries to account for the fact that some players move around throughout their careers, or are simply utility players, by randomly assigning a low rating at a secondary position when the player is generated. I see a lot of young OF prospects with a 2/20 SS rating; I think these are the same prospects you're seeing play out of position.

The solution to this problem is two-fold, and it's probably a pretty significant change to the engine. First, the root defensive ratings should be more general and tool-based, as they represent ability; positional ratings should come from a combination of root defensive ratings and positional experience--to an extent intelligence should be factored in, and to an extent experience might go up in "clusters" (when you play a lot at SS, it's going to make it easier to transition to 2B later; if you can master 3B, you can probably play 1B; etc). Second, the AI should have some basic limits set that keep the defensive spectrum in mind, such that it won't start a player with a 5/20 SS rating and a great bat if there's someone with a 10/20 SS rating or higher on the bench. These might become customizable, either having an offense vs. defense slider for each position for the user team, or having a league-wide setting for computer teams, but that's another layer of complexity on top of what's already becoming a pretty significant suggested change.
teachermike is offline   Reply With Quote