Regarding trading--Not a debate but just interested in knowing who agrees

[Eckstein 4 Prez]

The only example I've seen of a terrible trade of a 1-for-1, 1-for-2 or 2-for-2 variety is the Campanella deal. While that one looks bad, I guess I can rationalize an occasional not-so-great deal - certainly there have been some bad deals in real life. Other than that, it looks like your real complaint is that the game needs to understand better how to tally up the value of multiple players. I agree with you on that point, but it hardly renders the game unplayable as is.

[mking55]

Originally Posted by Eckstein 4 Prez

Yes, exactly. There seems to be ample evidence from many different sources that the AI turns stupid as soon as there are more than 2 players on one side of a trade.

mking, I'd really like to know if you can take advantage of the AI as easily without using more than two players on either side of the deal. If not, then it's the problem that's already known (and which needs to be addressed) - the AI doesn't understand what to do with large multiplayer deals. If so, then might I suggest that mking be added to the OOTP12 beta team to do whatever he can to break the trading AI? I think everyone agrees it needs work, and who better to diagnose the problems than the biggest critic?

Sorry, you probably were posting this while I posted just above to your previous point that contains an example of what you're asking about.

I traded Ned Garver for ROY CAMPANELLA and Lou Brissie. That's I gave up 1 good pitcher, and they gave up a 3 time MVP Hall of Famer (when he was young) and a ballplayer who served in WWII and wrote about it. (The Pitcher Was A Corporal).

[mking55]

Eckstein 4 Prez;

I think we need to coordinate our posting. That's twice.

Martin

[robc]

Originally Posted by SteveP

So, presumably the operating theory here is that we keep beating on Markus until he achieves enlightenment on the secrets of programming an AI.

I'm happy that he's designed a pretty goo sim of real life baseball -- which I think is all he really set out to do. If lightning strikes and he achieves excellence as an AI programming, I will consider myself to be doubly blessed.

When I got my degree I took an AI programming class. Hmm, I barely remember what was even covered. Anyways...

Considering what Markus has achieved with OOTP, I don't think it is too much of a stretch to think he could tighten up the trade AI some if he is able to devote the time to it. I imagine with everyting he probably wants and needs to do with OOTP the time may be the big problem. Unless of course you have inside information and know it isn't possible.

[Eckstein 4 Prez]

Originally Posted by mking55

Eckstein 4 Prez;

I think we need to coordinate our posting. That's twice.

Martin

Yeah, true enough.

The Campanella deal is really lousy. I'm assuming their ratings and potential at the time of the deal were about what we'd think based on their historical output? If so, I don't know what to say about that one.

But I have a tough time thinking that you're going to be able to create 130-win teams using just one or two player trades on each side. If so, you are either much better at trading than I am or you're found some new exploit I don't know about.

[mking55]

Originally Posted by Eckstein 4 Prez

Yeah, true enough.

The Campanella deal is really lousy. I'm assuming their ratings and potential at the time of the deal were about what we'd think based on their historical output? If so, I don't know what to say about that one.

But I have a tough time thinking that you're going to be able to create 130-win teams using just one or two player trades on each side. If so, you are either much better at trading than I am or you're found some new exploit I don't know about.

If I stick to 1 for 1 or 2 for 2 or 3 for 3 and so on, or 1 from me and 2 or more from them and so on, I won't have 130 win seasons. I'll probably be a reasonably successful team but not the God of baseball simulations.

But as I've said, I'm too lazy to make up my own rules (even if it would be more challenging). I'm more interested in having a competitive AI which I don't believe is some sort of pipe dream.

[Jeffy25]

I just don't do any trade in any game unless I believe it would actually happen in real life.

Just don't take advantage of the AI.

I also made a trade approval thread on this forum for people to post trade opportunities for other forum members to say they approve or not of said trades.

Police yourself, I play on the hardest trading difficulty as well, but when a player is placed on the trading block, it is all too easy to land them.


Also, as the previous poster said, no 4 for 1 trades etc.

2 for 2, 2 for 3. 1 for 2 etc are okay.

[SteveP]

Originally Posted by robc

When I got my degree I took an AI programming class. Hmm, I barely remember what was even covered. Anyways...

Considering what Markus has achieved with OOTP, I don't think it is too much of a stretch to think he could tighten up the trade AI some if he is able to devote the time to it. I imagine with everyting he probably wants and needs to do with OOTP the time may be the big problem. Unless of course you have inside information and know it isn't possible.

Well, I am in real danger here is just repeating myself, so I will try to summarize instead:

1. The knowledge required to program a specific computer simulation of a real world activity is not the same knowledge required to program an AI.

2. I am sure that people have been complaining about the AI since this game was first introduced. I wasn't around for that event, but it's as certain as the sun coming up in the East. Another complaint thread is just another complaint thread.

3. Markus is really the only guy working on most of this stuff (Andreas primarily does the deep technical stuff). And he also has a business to run. Even if he wanted to hire a guy with more AI knowledge than he has, he can't afford to.

4. I have no reason to think that Markus regards the AI as low priority from a business perspective (see item #2).

5. The AI works as well as it does because a baseball sim doesn't ask too much of an AI. The simulation engine is more significant. Given this, I have no reason to think that Markus has an untapped wealth of AI programming knowledge on hand.

I hope that covers everything. I am sure that enough people will continue to beat this dead horse as long as there is an OOTP game to play. It's probably genetic.

[robc]

Originally Posted by SteveP

I am sure that enough people will continue to beat this dead horse as long as there is an OOTP game to play. It's probably genetic.

Perhaps we should just have one trade AI thread that we bump from time to time just to make sure Markus knows we all still want improved trade AI .

On a more serious note, if it could handle more players in a trade that would be a big step up. I usually do try to keep it at 2 players or less per side, 3 tops.

[William Satterwhite]

Originally Posted by mking55

You may be right and you may be wrong about teams offering trades for potential greats. I tend to believe you're wrong (but we'll never know). Why wouldn't I try to ask for Ted Williams? Maybe I'll learn the other GM is an idiot. There is no harm in asking so I tend to believe these players were asked for in trades.

Because the AI cannot be an idiot, it can only process the information available to it (numbers and values) and use that information to make decisions. Only another human can be an idiot.

Now sure, I imagine someone at some point may have inquired about Williams but I also imagine the conversations were probably very short and not worth having to begin with- in the case of the A's there probably was no conversation, if the Yankees called the conversation probably ended up with the Red Sox asking for Joe DiMaggio in return and the Yankees saying "Okay, just asking".

Look throughout the entire history of baseball and see how many times in-prime stars or "untouchable" future stars have been traded- the few times its happened there's been a story behind it. Jeff Bagwell didn't become a Houston Astro because the Astros wanted him so badly they "just asked" for him- the Red Sox needed a pitcher for a playoff push and had an expendable 3B prospect due to their in-prime star 3B and also had multiple 1B prospects. Even a straight up star for star trade like Frankie Frisch for Rogers Hornsby only happened because the two wore out the welcome with their respective teams. There's always a method behind the madness that only a human could reason through.

Your other point that maybe the AI team needs this or that isn't correct and you'll have to trust me on that. I see too many of the players I've traded to the AI become free agents after the current season. If they were really needed the AI team should move heaven and earth to resign them.

I think you're underestimating the AI's desire to always want to improve its team (which I think is the cause of most of the game's AI quirks). Basically i see it like this- the AI accepts your offered players because it helps the team today and in the near future until a time comes when the AI believes it can do better. The AI lets those players walk because it "thinks" it can get better players for those postions/roles. As a human you're able to think the situation through and understand that if the AI was willing to take those players, it should take care to try to keep them. The AI can't "think" the situation through like that though, its only concern is whatever decision makes the team better overall. The AI doesn't seem to have 5-year plans or anything like that, it has today plans.[/QUOTE]

It's like this- the AI realizes it has a need for a shortstop and a starting pitcher. You come along and tell the AI you'll give it a solid shortstop and a solid starting pitcher in exchange for a good-to-great leftfielder and you'll throw in a solid outfielder to replace him, the AI crunches all its numbers and says the total value of your package fits its needs. The end of the season rolls around and the AI crunches its numbers again and now decides, for whatever reason, those players aren't worth keeping right now. Its the same reason why top prospects get waived in the game seemingly at random- there is too much involved in the running of a true-to-life baseball organization to expect a computer program to handle all these nuances just as well as human beings.

[mking55]

Flogging the dead horse!

I took my trades from the "best trade AI in any baseball simulation" and tried the same trades in a different simulation program. It's hard to get them to match because of AI trading against each other, free agency, so on and so forth. Also, there will be a variation between player ratings in the 2 games but since we're talking about some really good players here, I don't think the variation is so much.

I looked at the 7 major trades my A's made in 1939 (4) and 1940 (3). These trades changed my team from the dogs I got to the contenders they were in 1940 before winning every pennant from 1941 to 1963. I made sure the teams had the players involved in the trade and I put the trades to the AI in the other baseball simulation. I know this is not a "scientific test" and isn't meant to be. I was just curious to do a small comparison and get a "feeling" for how they compare.

This is not a comparison of games. I'm not interested in that. Some people like Game A and some people like Game B and they're both right. This is just a small trade comparison.

First before I list the trades, I want to say that the other game didn't come close to accepting any of these trades. Does this mean it's a better game? No, not really. That's individual preference. Does it mean the trading module in this other game is tougher? I'm confident in saying it is.

As Rosie used to say, "Weigh it for yourself honey".

1939
Trade 1: NY Giants
The Giants give up 3b Billy Jurges, SP Bill Lohrman, C Harry Danning and 1b Zeke Bonura for C Hal Wagner and RF Eddie Collins.
Trade 2: Chicago Cubs
The Cubs give up SP Dizzy Dean, RF Jim Gleeson, SS Dick Bartell, and MR Ray Harrell for C Harry Danning, SS Al Brancato, MR Dave Smith and MR Roy Parmalee
Trade 3: NY Giants
The Giants give up 3b Mel Ott, SS George Myatt, C Ken O'Dea, and SP Cliff Melton for 1b Zeke Bonura (see trade 1), SS Billy Jurges (see trade 1), SP Buck Ross, SS Bill Lillard, and MR Jim Reninger
Trade 4: Boston Red Sox
The Red Sox give up 2b Bobby Doerr and SP Charlie Wagner for CF Lou Finney, SS Fred Chapman, 2b Dario Lodigiani, and MR Jim Schelle

1940
Trade 1: Cleveland Indians
The Indians give up SP Bob Feller for RF Wally Moses, C Billy Sullivan, MR Bud Thomas, SP Lynn Nelson, and SP Nels Potter
Trade 2: St. Louis Browns
The Browns give up CF Pete Reiser, 1b George McQuinn, MR Loy Hanning, and SS Sig Gryska for SP George Caster, 1b Dick Seibert, SP Eddie Smith and MR Chubby Dean
Trade 3: Boston Red Sox
The Red Sox give up LF Ted Williams, and C Bill Baker for SP Dizzy Dean, RF Dee Miles, and SS Skeeter Newsome

Let's break that down a little to see who gave up more players.
I give up 2 and get 4 in return.
I give up 4 and get 4 in return.
I give up 5 and get 4 in return.
I give up 5 and get 1 in return. This was the only totally lopsided trade in terms of number of players. But it was Bob Feller.
I give up 4 and get 4 in return.
I give up 3 and get 2 in return. (The Ted Williams trade.)

I honestly thought I threw more at the AI in terms of sheer number of players but it looks like I didn't.

So there you have it, whatever it is. I think Product B has a tougher trade AI than Product A (OOTP). What that translates to for me is that OOTP could, if so desired, be a tougher trade opponent.

[Le Grande Orange]

Originally Posted by mking55

Sorry, I forgot to comment on this.

I don't use scouting. I'm the scout. So I acquired Roy Campanella without him ever playing a game.

Do you think Branch Rickey would have signed a 2nd-rate black catcher back then? Don't you think when they acquired Campanella he was untouchable before he ever played a game in MLB?

Answer this question: had the player not been named Roy Campanella would you have made that trade?

You saw the name, then went after him. You didn't have to scout the player, you didn't have to examine his statistics and make a judgement about his potential value. No, all you needed to do was look for his name and make a deal. Indeed, all you ever need to do is look for names and make a deal.

Given this, I guess I don't see what the attraction for trading is for you. If it was a fictional league, or even a league with randomized historical names, you'd have to work a lot harder. You'd actually have to scout and examine and judge the players rather than simply seeing the name and know instantly you have to have that player.

[SteveP]

Originally Posted by robc

Perhaps we should just have one trade AI thread that we bump from time to time just to make sure Markus knows we all still want improved trade AI

I suggest taking up a collection and see if it's possible to buy one.

[mking55]

Originally Posted by Le Grande Orange

Answer this question: had the player not been named Roy Campanella would you have made that trade?

You saw the name, then went after him. You didn't have to scout the player, you didn't have to examine his statistics and make a judgement about his potential value. No, all you needed to do was look for his name and make a deal. Indeed, all you ever need to do is look for names and make a deal.

Given this, I guess I don't see what the attraction for trading is for you. If it was a fictional league, or even a league with randomized historical names, you'd have to work a lot harder. You'd actually have to scout and examine and judge the players rather than simply seeing the name and know instantly you have to have that player.

Let me try again. I promise this is the last time I will explain myself, although I'm puzzled why I have to explain myself.

In a way, I'm the GM and history is my scout. This isn't real life so I can't meet Roy Campanella or know anything about him or see him catch or hit for myself. I can look at numbers in the game and say, "oh, looks good". But because of a combination of laziness and the enjoyment I derive from seeing these names in front of me, I play it my way. So if I had just used the ratings in the game, I would have traded for Roy. But I used Branch Rickey's scouting staff to help affirm my choice.

I have no idea why this has become a referendum on how I play the game. What does that have to do with the price of tea in China? I would never in a million years put anyone down for how they play a game. It's nothing to me, and if someone wants to play a game a certain way, that's great. It's personal preference. Why am I some kind of moron for playing my way?

I commented in an earlier thread on my 3 year fictional league and how I cleaned up winning the pennant each year by between 15 and 35 games. Maybe you didn't see it. However, I can't do 30 years for a full test out because it would be worse than work to me.

All I'm saying is the trade AI could be much better. That's all. I play this game for about 2 to 4 weeks a year when it comes out. I buy it because that's the only way I can support this genre and if some fool feels he can write a sports text sim because there are people out there buying and playing the game, then that's great.

Every year I see I'm not satisfied with the trade AI and I move on. Some years I comment on it and some years I don't. I commented this year because I've seen better and because I've seen better, I was interested in offering whatever assistance I could (which has been turned down but I'm not offended by that).

A couple of posts up you'll see (if you haven't) an unscientific study comparing the trade AI between 2 games. I don't claim it's conclusive, but it does lead me to believe that there are better trade AI's out there, and therefore, this one could be better.

It's all about asking for a better game. It's not about how anyone plays the game. I admit fictional is or should be more challenging. I admit house rules would make more of a challenge. However, fictional bores me and house rules seem like work to me. Does that make me an idiot? I hope not. It's just about hoping and asking for a better trade AI and to me, that's a better game.

[PhillieFever]

Originally Posted by Chase Davis

Yeah, like in NFL Sports Talk Football where I could take it to the house every time with Kenneth Davis on the Bills.

I ran the Flip Weak Side run every play and every time he'd take it home, spin here, stiff arm there and I had six...

...of course I was 10 then so I was commonly making dumb decisions.

HaHa, don't forget the HB toss to Neal Anderson in one of the early Madden's, or the deke move in NHL that worked every time.

[William Satterwhite]

I think just about everyone agrees that the trade AI (if not the AI in general) is flawed to some degree. I think most of the discussion here revolves around the idea that Markus isn't doing anything more than he can to address the issue and the idea that due to the flawed AI, a human player shouldn't be burdened to come up with ways to make the game more challenging to himself.

An example was pointed out where a previous version did feature a tougher trading engine to the point where it was then almost impossible to make any trade for a star player with an AI club and that turned people off- certainly you can understand why folks who can find ways to fully enjoy the game as is would be leery of calls of dramatically changing things when there are clear workarounds.

[Hooray Saturday]

wow 8 pages and we haven't gotten an "OOTP only knows numbers, not names" response?

[Afino]

Originally Posted by mking55

But they're Microsoft so who are you going to complain to? Bill Gates? Steve Ballmer? Alexander Ovechkin?

I lol'd at that.

That and the post about Back to the Future that was thanked by someone named Doc Brown.

[CD1083]

Originally Posted by Hooray Saturday

wow 8 pages and we haven't gotten an "OOTP only knows numbers, not names" response?

It's funny you say that. I was going to respond with that earlier today; but than I said to myself, "surely someone has already responded with such an obvious retort to this asinine thread topic" and than I said "don't call me Shirley."

[knockahoma]

In an earlier thread, I had mentioned that I'd like to see the News section detail why trades are made. I wouldn't expect great writing, but the AI does have its reasons for making trades. I think that would go a long way toward helping the AI.

The community has given invaluable imput on a number of game-play issues over the years, which probably has lifted OOTP to the outstanding product it is today.

Stating the reasons in the news section would help us all evaluate how smart the AI is at trading in any given situation. Take for example the Campanella trade. If the community took umbrage at such a trade, we should define what made that trade such an abomination-- beyond time-travel reasons. Then, Markus could tweak the AI thinking based on those particular objections. Sure, the AI reasoning might be embarrassing in the light of day on some trades.

In fact, I'd like to see a detailed thread defining what factors the AI considers for a trade as the game now comes. Then, I'd like to see the community explore various scenarios for Markus. Perhaps he could then implement these ideas, or tweak the percentages based on comments. For example, does the AI look for a trade when the Yankees slugging 3B goes down for the season in May? Why, or why not?

Putting all the AI's reasons for a trade in the news parrots real life. The commentators dig deeply to understand trades. That's what makes baseball such an interesting conversation. Really, Atlanta? You're trading a young Dusty Baker for the Dodgers declining Jimmy Wynn? What was the thinking that went into that?

Putting the reasons for a trade into the news makes the AI reasoning accountable. Far more importantly, it helps the community make the trade AI better. To borrow a phrase from a sports movie, Markus, when it comes to the AI trading, "Let Meeeeeeee help yoooooooou."