04-09-2012, 11:53 AM
|
#1
|
Pro
|
Do you know how the AI calculates...
...the setup of the bullpen?
...the position alignment of the fielders?
...the lineup of the batters?
It was discussed last year and this year we have obvious the same problem.
The AI/CPU makes hilarious setups of the bullpens regarding who the closer/setup men/middle and long relievers are.
The AI/CPU makes ridiculous choices regarding who they let field the positions.
The AI/CPU makes unrealistic lineups.
This creates headache especially in franchise when you see right fielders playing center field and center fielders playing right field and nobody knows why. Or the leadoff hitter in reality is the fourth batter in the lineup ingame. Or the closer in reality is a middle reliever ingame. And so on and so on.
But did/do you know what comes into account for the AI to calculate those things?
I made some testings and this is what I found out:
1) Starting pitching/bullpen:
This is the easiest section. The starting pitchers are put in 1-5 simply based on their overall ratings. I think they at least must have a 70 something stamina rating also.
The bullpen is somehow similar created. At first here the overall rating comes into count. The highest rated pitchers of the foundation that are not in the starting slots 1-5 of the MLB team of the organization are the 7 pitchers that will be in the bullpen. After that the slots are based on the composure rating of the players. That means that the pitcher with the highest composure will be the closer, the 2nd one setup up man, and so on until the two pitchers with the lowest composure rating will be the long relievers. The only way a pitcher with a lower comp. rating can have a higher spot than another pitcher with a higher rating is when the lower comp. guy has a only slightly lower comp. rating but also another category (control) in which he is way ahead of the higher comp. guy.
2) Position alignment for fielding:
This is the first really ridiculous part. I didn't finally find out how it works. I only found out that the CPU makes sometimes really questionable decisions. E.g. Carlos Gonzalez is at first hand a LF, but he has a high RF and LF rating, too. He is the best outfielder in the Rockies organization. He will always play CF. The AI doesn't care what goes on in RF or LF, it simply puts Gonzalez into CF. Dexter Fowler is a mediocre CF and has a lower LF and RF rating. Cuddyer is a RF and has a lower LF rating a very low CF rating. The CPU puts it always like this:
LF Cuddyer
CF Gonzalez
RF Fowler
Normally it should be:
LF Gonzalez
CF Fowler
RF Cuddyer
When you look into the position alignment of the management section ingame the 2nd option is the option in which the highest overall fielding rating (when you sum up the ratings of the 3 players) is the highest. The first option has a lower overall rating (when you sum up), but it has Gonzalez in CF which means that it seems that the CPU at first looks at the best CF available. But that is not the case, since in other teams the AI screws that over also.
So we must deeper look into that. Here comes into account what range a fielder has and how good his arm and glove is. Sometimes the CPU doesn't put the outfielder with the best arm and range into the CF slot, also his CF fielding ability is not the highest, that means he would be better at another position. To come over this you have to set the fielding abilities of the players secondary positions very, very low. For some players this is very very unrealistic when you compare that to reality, but there is no other way to fool the AI.
For example a problem was that the AI often switches around the Dbacks outfield.
In reality it looks like this:
LF - Kubel (Parra)
CF - Young
RF - Upton
The AI set it up like this:
LF - Kubel (Parra)
CF - Upton
RF - Young
Now I lowered Uptons fielding ability at CF to 60, but the AI still set him up in the CF and Young in RF. After that I trimmed his CF ability down to 50 and NOW the CPU sets it up the correct way of LF Kubel, CF Young and RF Upton.
It always depends on the best outfielder. Here we come to Gonzalez again. I tried to lower the ability of Fowler for LF and RF so low that the CPU doesn't set him up at LF or RF, so to say I expected the CPU to put him into CF since there would be no other option available for him. But the CPU didn't, they finally benched him (with a 80 overall rating) and called up a AAA-LF (Blackmon) which was set up in RF. The only option here would be to lower the fielding ratings of Gonzalez significant, at least for CF.
Another problem is e.g. the Giants outfield, here the CPU always puts Cabrera into CF and Pagan into LF, although Pagan is the better fielder.
Somehow I think the arm rating is also very importan.
It is all so difficult to see through...
So finally we have two choices:
- when we want to have a chance of realistic fielding position alignment we have to cut down some fielding abilities of players ingame, which would be unrealistic when compared to their real life abilities
- when we want to have real life abilities of the players also ingame, then we must accept that the CPU will screw up the positioning.
3) Lineups:
Well, this is the most difficult part. Here I don't understand nearly anything. Sometimes it looks like the lead off hitter is always the fastest guy available in the lineup, in other lineups it looks like it is a mishmash of the combined highest sum of speed and contact. Sometimes it seems to be depending on who has the highest power rating, because this guy is always number 3 or 4, again depending what contact he has, and so on and so on.
I think we have a chance of getting the Bullpen and Positions of the fielders right if we want to. But I think we NEVER will have a real chance of getting realistic lineups when playing a franchise.
If somebody wants to post any comments here you go. A discussion would be very appreciated, perhaps somebody knows something which I didn't find out.
Last edited by wudl83; 04-09-2012 at 11:58 AM.
|
|
|