AI Trade - why even make this trade?

[SittingDuck]

Would you look at these two guys? Both back-up catchers, neither doing much, around the same age, ratings and salary. Verrrry close, indeed.

So why even make this trade? No rhyme or reason, if you ask me. I tend to think it is the AI just meeting the odds on trading. IOW, meeting the activity level that is set by the XML parameters. Just trading for the sake of trading.

So I've been toying with the idea of eliminating AI trading all together. Of course, I'd need to come up with some way to help these teams or the whole concept of team improvement via trading is d-e-a-d. Maybe it needs to be - I don't know...

[SittingDuck]

What a catch.

[SittingDuck]

Maybe it was one of those Shea Hillenbrand kind of deals?

I need to come up with some type of rationale.

[akcranker]

Looks to me like if the AI traded the first player for the second player.. then it was for defensive purposes and to get a few years younger...

[KG Erwin]

I don't know, man.  Does it have anything to do with the use of 1-10 ratings scale?  Perhaps if you returned to a higher level scale (I went back to 1-100), you could detect the trading logic easier. 
 
How are the teams' management tendencies set? 
 
Heck, in this case, I'd go with the younger guy with proven major league experience.  That may be the key factor.

[Amaroq]

Alternately, you might have had a team in playoff contention willing to take the marginally-better 28-year-old as its backup trading away the marginally-worse 23-year-old who doesn't look like he'll ever be a star anyways.

That strikes me as a situation where one team got marginally better in the short-term, and the other team got marginally better in the long-term, which is a fine trade - and doubly so because neither team 'crippled' itself by creating a situation where they were caught without a backup catcher.

[SittingDuck]

I've made some switches to my xml trade frequency now and I find it a bit more acceptable.  Seems they trade a bit more if they have to for team purposes.  In my mind, at least.

I also see some interesting trades from time to time.  As if once in a blue moon the system sees something that would accomplish something for each team.

That said, let's not forget that we have the benefit of looking and see what glory or gack a player might be historically.  The AI goes on values that it sees at the present.  And IRL, how many trades actually turn out beneficial for both parties in equal fashion.  Not that many.  So I can accept some trades, and many of them I can chalk up to 'hey, your GM just got suckered'.

I'd just be very happy to see the AI not give away Yaz or Boog or Mickey for a scrub or even some chump.  To me, that is the breaker.  If we can get that sorted out, then I can deal with it all.

[Amaroq]

Hmm, did I miss the 'frequency' changes, or are they in your trade AI thread?

One thing that's tough is, IRL *most* trades are M-for-N trades - IRL you're much more likely to see "Joe Bloggs and an undisclosed amount of cash for Tom Smith, AAA catcher Hank Diggs, A propsect Mort Evans, and a player to be named later" than "Joe Bloggs for Tom Smith".

I'd always assumed that 'frequency' code would impact 'how often does an AI team consider making trades' - but since you're saying that the trades look *better*, I wonder if the 'frequency' change impacts the 'how good does a deal have to look, to me' code? If that's the case, would it wind up creating the right frequency of trades if Shaun implemented AI M-for-N trades?

[KG Erwin]

SD, should I mention that baseball's history is chock-full of "sucker" transactions, with the most infamous being conducted on Jan. 3, 1920, between the Red Sox and the Yankees. Need I say more?  [;)]

[KG Erwin]

Amaroq, the ugly face of finances and players-for-cash makes sense IF it applied through ALL historical eras, when cash was really in short supply for some teams.  
 
How does one reconcile the reserve clause with the need of money, UNLESS a different method of finances was applied.  We're now talking about operating costs as opposed to player salaries, and this opens up a kettle of worms that could break the game as we know it.  
 
The accounting factor already exists, but to reasonably sim the pre-free agency era you'd HAVE to account for "other expenses". 
 
I'm dead set against going that route.   It would only lead to a downward spiral and a fate that has beset a game I won't mention. 

[rowech]

When you guys use these 1-10 rating scales, you aren't seeing the full picture like the AI might.  In addition, I agree that the younger guy brings some youth and a heck of an arm. 

[waltwa]

consider this

has any game ever had an AI that handled trades well. i play ps,bm and ootp and all have problems in that department. this is probably why i have gravitated towards online leagues.

i hate to think this way but i think we are looking at about 5 years of bad trades by the computer.

i have sometimes set up a house rule that i will not initiate any trades but the problem with that is that in all 3 games you never are offered a trade by the computer that is a fair trade. i just don't think this is an area that is going to be easily dealt with in the near future.

[KG Erwin]

That's  what it gets down to, waltwa.   Designing a competent AI is the toughest part of game design, and it's been the bugaboo of many wargame designers.

To Shaun's credit, the AI in-game manager isn't too shabby. I have a tough time against it.   It doesn't make many stupid decisions, and it DOESN'T cheat.   I have yet to find a way to manipulate it  --  you play as you would in real -life, and it's a good challenge to one's management expertise.  This is one of Pure Sim's strongest points, IMHO. 

 

[SittingDuck]

Waltwa, I'll agree with you there - I just wish there were more people who wanted to run/play online leagues.  There are several things that keep that at a minimum, IMO:

1) PS has a generally small community compared to others.  Back in the day of FPS, there were tons of leagues and tons of players.  So, smaller pool of players to work with
2) Hard to agree on an era to play - so many want to play in different eras
3) Many don't have "time" (for what, I might add.  do you waste time watching the idiot box instead?)
4) Wanna play head-to-head or not at all (ok, I really don't have that kind of time, folks, and neither do many productive people)
5) a few others,  but that's enough

So the ultimate experience with PSBB would be an online league where people could wheel and deal.  I would love that because as you initially say, it takes the AI right out of it for GM mode.  And that is great.

Yes, trade AI is one of the most difficult things to ever get running in a semi-functional way.  It would also be a killer component of a landmark game, should it ever be accomplished.

[SittingDuck]

Amaroq - not really sure they are all that much 'better', but hopefully what I am doing by decreasing the call for the AI to trade - "hey, trade if you feel you need to, not because the xml is making you find a trade", and thereby getting better trades, or at least ones that make a little sense.  As I said earlier, I realize I can't prompt it to make interesting trades.  If I can just eliminate the true boners that I've mentioned (superstars for scrubs), then I'll be quite happy. 

Anyhow, here is what I've done:

<GM_AI Activity="15" PromotionActivity="75">

<TRADEFREQUENCY Value="1" MinGames="20"/>

New ranges and values:
<INFLUENCE_GAMESPLAYED RangeMin="0" RangeMax="30">.05</INFLUENCE_GAMESPLAYED>
<INFLUENCE_GAMESPLAYED RangeMin="31" RangeMax="45">.1</INFLUENCE_GAMESPLAYED>
<INFLUENCE_GAMESPLAYED RangeMin="46" RangeMax="60">.2</INFLUENCE_GAMESPLAYED>
<INFLUENCE_GAMESPLAYED RangeMin="61" RangeMax="75">.3</INFLUENCE_GAMESPLAYED>
<INFLUENCE_GAMESPLAYED RangeMin="76" RangeMax="90">.45</INFLUENCE_GAMESPLAYED>
<INFLUENCE_GAMESPLAYED RangeMin="91" RangeMax="100">.6</INFLUENCE_GAMESPLAYED>
<INFLUENCE_GAMESPLAYED RangeMin="101" RangeMax="107">.75</INFLUENCE_GAMESPLAYED>
<INFLUENCE_GAMESPLAYED RangeMin="108" RangeMax="113">.9</INFLUENCE_GAMESPLAYED>

I'll see how it goes.&nbsp; At least the amount of trades is somewhat muted.&nbsp; This does allow me to analyze them further.&nbsp;&nbsp; I'll continue to comment on trades in this thread as my league progresses.