How's the Trading AI??

ravinhood · 05-02-2006, 11:18 AM

I've enjoyed all the OOTPBB series, but, have always wished the trading AI was better and even the free agency purchase AI. I still find over all the series that the AI will trade for aging dying stats players and pay tremendous amounts for the same during free agency and give these players extended years way past their 40's. They will also accept trades with me for players who's performance has been awful in that year and I'm just trying to dump them because they are making 10 million plus and not worth the expense. A trade as a human player I would never accept, why does the AI?
Many times even with the setting on "Highly favor Prospects", they will put up some of their best prospects or at least a very decent one for a player I'm just trying to dump because of salary or because of performance.

Does the AI even look at their performance for the season at hand or just their attribute points that are clearly high and worth the price, but, their performance for that year has been horrible. For instance I'll have a player with 62 75 51 yet batting .182 for 2 months of play and only 4 HR's. And the AI will break it's back trying to make a trade for that player with me. Of course that is with me cycling thru the teams looking for the best deal, heh, it's sorta hard not to since in the real world I would get on the phone and do the same if I were an owner/manager. But, in the real world I don't think I would get the deals that I do vs the AI.

I don't want to have to set the AI on HARDEST just to stifle myself, I want the AI to make practical and analyzed decisions based on what is in front of it and really think out the deal, ask itself why should I trade my 5 star SS for this crappy 10 to 20 million dollar player who just happens to have a 6 year contract with 62 75 51 attributes and yet batting .182 for 2 months and only 4 HR's???

That's also something else, when a player is having a bad year and has a fairly low salary already but yet good attributes I can offer an extension for 7+ years sometimes for a very low salary. Perhaps there should be more extention year caps. I've secured Pujols until he's 35 for a mere 12 million from his now 28 age at which time I will dump him anyway and probably even before he reaches that age. It's all too easy to keep star players many times at low salary where they would bring mucho millions if they could hit the free agency scene. The way OOTPBB 6.5 has been playing players with stats much LESS than Pujols are getting near 20 million plus. Especially if I try to get them. lol I guess that's one good part of the AI it will force the player to pay extreme prices in the free agency, but, this can backfire as well as I will or can pull out when the salary is way up there and then that AI team is forced eating a player who doesn't perform up to his salary. It always seems pretty much any decent player I want off of free agency I have to pay 18 million plus for them. Then if they don't perform immediately to my liking I can dump them on the team that offers the best 5 star prospect. Even these 35 year old players that the players knows is going to bomb out in just a few years even if they do have 7+ year contracts.

Just wondering if this has been improved in OOTPBB 2006? What have you done to improve this?

andymac · 05-02-2006, 11:25 AM

Everything is adjustable. What do you have your player evaluations percentages set at? If most of the percentage is set to ratings then that is basically all the AI will look at. Also, if you are using scouts it is quite possible that the AI is seeing very different ratings than you are.

It is impossible to make a perfect AI. There are many things a human evaluator can see that is very difficult to input into the computer, much less input at the correct percentage.

Sure, there are improvements that can be made (and I'm sure are being made), but it is impossible to make it perfect. If you are this worried about the trade AI maybe you should try an on-line league against fellow human owners. That is where the toughest competition is going to be.

draven085 · 05-02-2006, 11:49 AM

Adjusting all the settings in the world isn't going to improve the trading AI as long as the AI lacks a concept of market value. As Aadik and others have pointed out the reason that the AI makes such poor decisions is that it appears to have hard coded concepts of value that are not affected by any league specific conditions such as finances, demand for players at a given position, etc. I don't know how one would go about programming AI that could recognize and analyze these issues but until someone can we'll always see relatively weak AI with regards to trading and financial management.

Gorilla Shakespeare · 05-02-2006, 11:56 AM

Also, the AI is really, really stupid when it comes to dealing with players from eras not dominated by the longball. That supposedly has been taken care of with the equivalencies and all, but I am taking a wait and see attitude.

To a point, though, you're always going to be limited by a. the fact that it's artificial, not real intelligence you're dealing with and, more importantly, b. the fact that the AI is modelled after real baseball. Baseball is not an industry dominated by rocket scientists. Even with Billy Beane and Theo Epstein, it's still an industry dominated by people like Joe Morgan, people who will trade for your 20-game winner with the 5.00 ERA because he wants a guy who "knows how to win." The best way I can see to get around this is to handicap yourself.

Sven Draconian · 05-02-2006, 12:54 PM

Just turn off ratings and talents.....sure helps with difficulty

james17 · 05-02-2006, 02:22 PM

Gee. I always thought the A/I did a pretty commendable job of trading. Signing mediocre free agents to outlandish contracts IS a problem I've seen.

As stated above, no A/I is going to be as good as most humans but all in all I think OOTP does a much better job than any other sports sim I've played. Having difficulty levels for trading helps too.

sfeldkamp · 05-02-2006, 04:28 PM

I personally am hopeful that VORP is an important factor to the AI in determining a players trade value. Other things I think are important in no particular order

Players Salary
Players Contract status (i.e. pending free agent)
Organizations Payroll Needs
Players Age (chance to improve)
Players Potential (according to AI scouts)
Organizations positional needs

Malleus Dei · 05-02-2006, 05:01 PM

Quote:

Originally Posted by Sven Draconian

Just turn off ratings and talents.....sure helps with difficulty

Exactly! I do that and it really increases the difficulty.

Oh, and turn off stars as well.

05-02-2006, 11:18 AM	#1
ravinhood Major Leagues Join Date: Apr 2004 Posts: 374 Infractions: 0/1 (1)	How's the Trading AI?? I've enjoyed all the OOTPBB series, but, have always wished the trading AI was better and even the free agency purchase AI. I still find over all the series that the AI will trade for aging dying stats players and pay tremendous amounts for the same during free agency and give these players extended years way past their 40's. They will also accept trades with me for players who's performance has been awful in that year and I'm just trying to dump them because they are making 10 million plus and not worth the expense. A trade as a human player I would never accept, why does the AI? Many times even with the setting on "Highly favor Prospects", they will put up some of their best prospects or at least a very decent one for a player I'm just trying to dump because of salary or because of performance. Does the AI even look at their performance for the season at hand or just their attribute points that are clearly high and worth the price, but, their performance for that year has been horrible. For instance I'll have a player with 62 75 51 yet batting .182 for 2 months of play and only 4 HR's. And the AI will break it's back trying to make a trade for that player with me. Of course that is with me cycling thru the teams looking for the best deal, heh, it's sorta hard not to since in the real world I would get on the phone and do the same if I were an owner/manager. But, in the real world I don't think I would get the deals that I do vs the AI. I don't want to have to set the AI on HARDEST just to stifle myself, I want the AI to make practical and analyzed decisions based on what is in front of it and really think out the deal, ask itself why should I trade my 5 star SS for this crappy 10 to 20 million dollar player who just happens to have a 6 year contract with 62 75 51 attributes and yet batting .182 for 2 months and only 4 HR's??? That's also something else, when a player is having a bad year and has a fairly low salary already but yet good attributes I can offer an extension for 7+ years sometimes for a very low salary. Perhaps there should be more extention year caps. I've secured Pujols until he's 35 for a mere 12 million from his now 28 age at which time I will dump him anyway and probably even before he reaches that age. It's all too easy to keep star players many times at low salary where they would bring mucho millions if they could hit the free agency scene. The way OOTPBB 6.5 has been playing players with stats much LESS than Pujols are getting near 20 million plus. Especially if I try to get them. lol I guess that's one good part of the AI it will force the player to pay extreme prices in the free agency, but, this can backfire as well as I will or can pull out when the salary is way up there and then that AI team is forced eating a player who doesn't perform up to his salary. It always seems pretty much any decent player I want off of free agency I have to pay 18 million plus for them. Then if they don't perform immediately to my liking I can dump them on the team that offers the best 5 star prospect. Even these 35 year old players that the players knows is going to bomb out in just a few years even if they do have 7+ year contracts. Just wondering if this has been improved in OOTPBB 2006? What have you done to improve this?

05-02-2006, 11:25 AM	#2
andymac Hall Of Famer Join Date: Feb 2002 Location: Effingham, IL Posts: 5,725	Everything is adjustable. What do you have your player evaluations percentages set at? If most of the percentage is set to ratings then that is basically all the AI will look at. Also, if you are using scouts it is quite possible that the AI is seeing very different ratings than you are. It is impossible to make a perfect AI. There are many things a human evaluator can see that is very difficult to input into the computer, much less input at the correct percentage. Sure, there are improvements that can be made (and I'm sure are being made), but it is impossible to make it perfect. If you are this worried about the trade AI maybe you should try an on-line league against fellow human owners. That is where the toughest competition is going to be. __________________ June Madness: Links FTB: andymac

05-02-2006, 02:22 PM	#6
james17 All Star Starter Join Date: Nov 2004 Posts: 1,305	Gee. I always thought the A/I did a pretty commendable job of trading. Signing mediocre free agents to outlandish contracts IS a problem I've seen. As stated above, no A/I is going to be as good as most humans but all in all I think OOTP does a much better job than any other sports sim I've played. Having difficulty levels for trading helps too. Last edited by james17; 05-02-2006 at 06:12 PM.

05-02-2006, 04:28 PM	#7
sfeldkamp Minors (Double A) Join Date: Mar 2006 Posts: 184	I personally am hopeful that VORP is an important factor to the AI in determining a players trade value. Other things I think are important in no particular order Players Salary Players Contract status (i.e. pending free agent) Organizations Payroll Needs Players Age (chance to improve) Players Potential (according to AI scouts) Organizations positional needs

05-02-2006, 11:49 AM	#3
draven085 Hall Of Famer Join Date: Apr 2003 Location: New Jersey Posts: 2,074	Adjusting all the settings in the world isn't going to improve the trading AI as long as the AI lacks a concept of market value. As Aadik and others have pointed out the reason that the AI makes such poor decisions is that it appears to have hard coded concepts of value that are not affected by any league specific conditions such as finances, demand for players at a given position, etc. I don't know how one would go about programming AI that could recognize and analyze these issues but until someone can we'll always see relatively weak AI with regards to trading and financial management.

05-02-2006, 11:56 AM	#4
Gorilla Shakespeare Banned Join Date: Mar 2006 Posts: 405	Also, the AI is really, really stupid when it comes to dealing with players from eras not dominated by the longball. That supposedly has been taken care of with the equivalencies and all, but I am taking a wait and see attitude. To a point, though, you're always going to be limited by a. the fact that it's artificial, not real intelligence you're dealing with and, more importantly, b. the fact that the AI is modelled after real baseball. Baseball is not an industry dominated by rocket scientists. Even with Billy Beane and Theo Epstein, it's still an industry dominated by people like Joe Morgan, people who will trade for your 20-game winner with the 5.00 ERA because he wants a guy who "knows how to win." The best way I can see to get around this is to handicap yourself.

05-02-2006, 12:54 PM	#5
Sven Draconian Major Leagues Join Date: Jul 2005 Posts: 432	Just turn off ratings and talents.....sure helps with difficulty