Success: remaking the Trade AI
Moderator: puresimmer
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
Padres, that is correct. All based off the v1.18 XML file, so as Taco Bell commercials say....
RE: Success: remaking the Trade AI
Awesome 
(p.s. Sitting Duck was my online screename for Battlefield 1942
)

(p.s. Sitting Duck was my online screename for Battlefield 1942

- PadresFan104
- Posts: 1147
- Joined: Mon Aug 08, 2005 3:29 am
- Contact:
RE: Success: remaking the Trade AI
Files are now on my site as a single download. Thanks Duck!
PadresFan's Text Sim Mod Website: http://www.padresfanmods.net
Follow Me On Twitter: http://twitter.com/padresfanmods
Follow Me On Twitter: http://twitter.com/padresfanmods
RE: Success: remaking the Trade AI
So is this file getting played with by folks? Is the general consensus good? I don't want to download it only to cause problems later.
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
It's like 1.5KB big, rowech. I'd suggest just d/l it and store your own favorite, in-use puresim.xml in a safe place and play with it on a copied league that you have (archive your current league so you can restore if you don't like it). It will take a few days before I expect much feedback.
I'd still like to hear from Shaun because I am not sure how the program recognizes my expansion of some lookup tables.
I'd still like to hear from Shaun because I am not sure how the program recognizes my expansion of some lookup tables.
RE: Success: remaking the Trade AI
I'm not up for testing at the moment, for obvious reasons, but I wanted to say that your changes look good - very similar to changes I made to the PS-2005 puresim.xml in pursuit of the same goal.
The 7/3 change in DYNAMIC_TALENT_EVAL seems almost mandatory; I think that's the solution to 'nobodies starting in the world series'.
Similarly, adding a non-zero value for Endurance to the IN_SEASON_EVAL_WT causes the AI to value starting pitching; very good.
I'd like to see both of those folded back into the 'official' .xml file, honestly, once we have community sign-off on the play balance.
One thing for folks to keep in mind is that this will result in an AI which values Power more than the current AI does; thats a good thing after 1920 or so, but it won't help those of you playing in the dead-ball era.
Shaun, I'd noticed that 'Salary' and 'Endurance' aren't present in the END_OF_SEASON_WEIGHTS section - would they be respected, if added?
The 7/3 change in DYNAMIC_TALENT_EVAL seems almost mandatory; I think that's the solution to 'nobodies starting in the world series'.
Similarly, adding a non-zero value for Endurance to the IN_SEASON_EVAL_WT causes the AI to value starting pitching; very good.
I'd like to see both of those folded back into the 'official' .xml file, honestly, once we have community sign-off on the play balance.
One thing for folks to keep in mind is that this will result in an AI which values Power more than the current AI does; thats a good thing after 1920 or so, but it won't help those of you playing in the dead-ball era.
Shaun, I'd noticed that 'Salary' and 'Endurance' aren't present in the END_OF_SEASON_WEIGHTS section - would they be respected, if added?
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
Thanks, Am. Looking forward to when you are more up to tweaking and we can bounce stuff off one another.
I am hoping that the tweak to the Endurance weight along with the historical reliever adjustment of v1.19 will really make historical leagues much better, as most v1.18 teams are always looking for closers and relief pitching. Here's hoping. I do know it at least lowers the RPs on the priority list.
So we'll see.
On power, I have noticed that in past versions that power was a real issue for a significantly high BA. This may not be the case now, wherein I would be tempted to draw the CH and PH levels even. I didn't think the CH5, PH3 factors were very good and caused the big sluggers to go unappreciated. This isn't the case now.
You'll note, and here's the big issue, that range, hands, and arm (as well as speed) really don't work well in the whole 'evaluation' thing. Because pitchers get rated for those areas, too, and it inflates their value. So there is no way to get a great fielder with a good bat (Brooks) to go higher in a team's priority list without whacking out pitchers and making them gods among the league. Just can't happen. And I think that's a very bad thing at this point. Hopefully it will change, but I don't know how.
I am hoping that the tweak to the Endurance weight along with the historical reliever adjustment of v1.19 will really make historical leagues much better, as most v1.18 teams are always looking for closers and relief pitching. Here's hoping. I do know it at least lowers the RPs on the priority list.
So we'll see.
On power, I have noticed that in past versions that power was a real issue for a significantly high BA. This may not be the case now, wherein I would be tempted to draw the CH and PH levels even. I didn't think the CH5, PH3 factors were very good and caused the big sluggers to go unappreciated. This isn't the case now.
You'll note, and here's the big issue, that range, hands, and arm (as well as speed) really don't work well in the whole 'evaluation' thing. Because pitchers get rated for those areas, too, and it inflates their value. So there is no way to get a great fielder with a good bat (Brooks) to go higher in a team's priority list without whacking out pitchers and making them gods among the league. Just can't happen. And I think that's a very bad thing at this point. Hopefully it will change, but I don't know how.
RE: Success: remaking the Trade AI
Here is some feedback.
I'm about 20 games into a 1936 season. Most teams are looking to make minor improvements and they rank my Top 10 players when I open the trade window and ask for nobody as pitchers. I have Jimmie Foxx who is 10CH/9P/10E and he is 16th. He does have a huge salary, but every team has plenty of available funds. I will say that almost all of those pitchers have 10 Stuff ratings, however Foxx isn't the highest ranked position player.
To continue to change the rankings, I would have to raise contact/power/eye and decrease the pitching stats, right?
I'm about 20 games into a 1936 season. Most teams are looking to make minor improvements and they rank my Top 10 players when I open the trade window and ask for nobody as pitchers. I have Jimmie Foxx who is 10CH/9P/10E and he is 16th. He does have a huge salary, but every team has plenty of available funds. I will say that almost all of those pitchers have 10 Stuff ratings, however Foxx isn't the highest ranked position player.
To continue to change the rankings, I would have to raise contact/power/eye and decrease the pitching stats, right?
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
v1.18 or v1.19? That might be making a diff. And are these mostly RPs? If so, that is a program thing.
What is Foxx's salary in comparison to many of the others?
What are those teams looking for? Relief pitching and closers?
Eye has nothing to do with player evaluation for trading or talent evaluation, for whatever reason.
What is Foxx's salary in comparison to many of the others?
What are those teams looking for? Relief pitching and closers?
Eye has nothing to do with player evaluation for trading or talent evaluation, for whatever reason.
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
Yeah, you could try increasing the power/contact a tad. I wouldn't touch the pitching.
You could also try adjusting the salary level down to ONE and see if that helps. In fact, I would do that first and then you can rule out salary or no.
You could also try adjusting the salary level down to ONE and see if that helps. In fact, I would do that first and then you can rule out salary or no.
RE: Success: remaking the Trade AI
I am using 1.19.
Boston is looking for 1B, Closer, and CF and has 1MM in funds.
Six of the top 7 are starting pitchers, so that's good. Foxx's salary is significantly higher than everyone else. He makes 180K vs. 110K for the #1 player listed.
I did tweak the settings and raised contact to 10 and dropped salary and there was no impact at all. I wonder if it is 1.19 issue or where I am in the season, currently 24 games in. I'll sim a couple of more weeks and see if it changes at all.
If I tweak the XML with the game open and save the file will the game recognize the new settings?
I appreciate the hard work. Hopefully this feedback is helpful.
Boston is looking for 1B, Closer, and CF and has 1MM in funds.
Six of the top 7 are starting pitchers, so that's good. Foxx's salary is significantly higher than everyone else. He makes 180K vs. 110K for the #1 player listed.
I did tweak the settings and raised contact to 10 and dropped salary and there was no impact at all. I wonder if it is 1.19 issue or where I am in the season, currently 24 games in. I'll sim a couple of more weeks and see if it changes at all.
If I tweak the XML with the game open and save the file will the game recognize the new settings?
I appreciate the hard work. Hopefully this feedback is helpful.
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
No, I think you have to restart the game. It reloads the xml file.
I have just found something that you tipped me off on. This could be a significant thing. Give me a bit.
I have just found something that you tipped me off on. This could be a significant thing. Give me a bit.
RE: Success: remaking the Trade AI
I've been reloading the game so that isn't the cause.
Does the game take into account position? For example if the CPU has a super star 1B will it devalue other first baseman? I bet it does, but can't be tweaked in the XML. Any evidence of that in your research?
Thanks again for the hard work.
Does the game take into account position? For example if the CPU has a super star 1B will it devalue other first baseman? I bet it does, but can't be tweaked in the XML. Any evidence of that in your research?
Thanks again for the hard work.
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
Well, it would figure what it 'needs', I guess. But no, there is no ability to mod that, as you say. Positions are not taken into account as far as trade value weights.
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
Yup, ok found something important but I'll have to post up the info tomorrow. New values and new files. This will get it all in line.
- Claymore Cut
- Posts: 270
- Joined: Tue May 16, 2006 9:48 am
RE: Success: remaking the Trade AI
Sorry if I missed it, but did Shaun get back to you yet with the questions you asked?
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
No, haven't heard from him.
OK, here's the deal:
Because the trade evaluation considers speed, arm, range and hands, this means that both position players AND pitchers are getting values accrued from their ratings. What is bad about that is that pitchers are getting ratings for the fielding stuff, which inflates their value. This is not what we want. We'd really like to have position players only get rated for speed and fielding stuff, but currently the code won't permit it.
Therefore, the temporary answer is to eliminate all weight values for speed, arm, hands and range. Take it completely out of the equation. So now hitters get rated on CH and PH; pitchers get rated on Stuff, Control, Endurance. Unfortunately, because they also possess CH and PH ratings, those values get lumped into pitchers trade value, too. Again we don't want that, but the ratings for pitchers in the hitting categories are meager, whereas their fielding ratings are often higher than position players. As you can see, this all poses a problem.
I am posting the new changes to the parameters in the previous posts. I have highlighted what needs to be changed in xml files. As well, I am updating and sending the two xml files to Padresfan for posting on his site. This will be the second version.
What you will see:
At the beginning of the season, no one is interested in trading and no player evaluation has been done other than simply based on ratings. I'll use the 1966 SF Giants as an example throughout this discussion. At the beginning everyone wants Juan Marichal and Gaylord Perry. Premo starters. Mays and McCovey come next.
However, after a few weeks of simming and teams have scouted and evaluated players and their own situation, things start to change. McCovey and Mays will leap up the list while Marichal might be #2 or #3 (what a great pitcher). Each team's situation dictates what they are interested in. For teams needing starters, even if McCovey is hot he might be outranked by Marichal. We want that, don't we? We want teams looking for what they need, and not just the hot player.
I also checked in as the O's manager and found that teams after a few months were tripping over themselves over Frank Robinson and sometimes Boog Powell, having dropped McNally and Palmer a bit down, just like the Marichal/Perry thing stated above.
Now as time goes on, I'll tweak the talent evaluation parameters more so that teams are putting a proper emphasis on current stats vs. player reputation (ie., ratings). This is important. And finding the right mix of getting the teams to evaluate and when is also important.
And hopefully Shaun can work (when he has time) on the coding for the player evaluation weights so we can include fielding ratings properly.
OK, here's the deal:
Because the trade evaluation considers speed, arm, range and hands, this means that both position players AND pitchers are getting values accrued from their ratings. What is bad about that is that pitchers are getting ratings for the fielding stuff, which inflates their value. This is not what we want. We'd really like to have position players only get rated for speed and fielding stuff, but currently the code won't permit it.
Therefore, the temporary answer is to eliminate all weight values for speed, arm, hands and range. Take it completely out of the equation. So now hitters get rated on CH and PH; pitchers get rated on Stuff, Control, Endurance. Unfortunately, because they also possess CH and PH ratings, those values get lumped into pitchers trade value, too. Again we don't want that, but the ratings for pitchers in the hitting categories are meager, whereas their fielding ratings are often higher than position players. As you can see, this all poses a problem.
I am posting the new changes to the parameters in the previous posts. I have highlighted what needs to be changed in xml files. As well, I am updating and sending the two xml files to Padresfan for posting on his site. This will be the second version.
What you will see:
At the beginning of the season, no one is interested in trading and no player evaluation has been done other than simply based on ratings. I'll use the 1966 SF Giants as an example throughout this discussion. At the beginning everyone wants Juan Marichal and Gaylord Perry. Premo starters. Mays and McCovey come next.
However, after a few weeks of simming and teams have scouted and evaluated players and their own situation, things start to change. McCovey and Mays will leap up the list while Marichal might be #2 or #3 (what a great pitcher). Each team's situation dictates what they are interested in. For teams needing starters, even if McCovey is hot he might be outranked by Marichal. We want that, don't we? We want teams looking for what they need, and not just the hot player.
I also checked in as the O's manager and found that teams after a few months were tripping over themselves over Frank Robinson and sometimes Boog Powell, having dropped McNally and Palmer a bit down, just like the Marichal/Perry thing stated above.
Now as time goes on, I'll tweak the talent evaluation parameters more so that teams are putting a proper emphasis on current stats vs. player reputation (ie., ratings). This is important. And finding the right mix of getting the teams to evaluate and when is also important.
And hopefully Shaun can work (when he has time) on the coding for the player evaluation weights so we can include fielding ratings properly.
-
- Posts: 1186
- Joined: Sat Aug 31, 2002 9:08 pm
RE: Success: remaking the Trade AI
And files sent to Mr. Fan O' Padres...
- PadresFan104
- Posts: 1147
- Joined: Mon Aug 08, 2005 3:29 am
- Contact:
RE: Success: remaking the Trade AI
[:)]
I'll post 'em when I get home in a few hours....
I'll post 'em when I get home in a few hours....
PadresFan's Text Sim Mod Website: http://www.padresfanmods.net
Follow Me On Twitter: http://twitter.com/padresfanmods
Follow Me On Twitter: http://twitter.com/padresfanmods
RE: Success: remaking the Trade AI
So this file doesn't bring about any multiplayer trades? Only one for one players.