11.19 Computer Trades [Archive] - Sports Mogul Forums

HoustonGM

04-10-2008, 01:33 PM

I still think that some work can be done on the computer to computer trades.

One thing it needs to stop doing is making huge deals involving multiple star players and starters....such as Dan Uggla, Josh Willingham, Jorge Cantu, Hanley Ramirez, Justin Miller, Paul Hoover, and Tagg Bozied for Matt Cain, Billy Sadler, Ray Durham, and Randy Messenger. The Marlins traded their starting 2B, LF, 3B, SS, a reliever, their backup catcher, and a minor league for a starter, 2B, and two relievers. That's unreal. Such a trade would never happen, especially in-season. In the offseason, it wouldn't ever happen like this. The Marlins might trade off that many starters (they have before), but it would always happen in a series of deals with multiple teams, not one gigantic trade.

(By the way, Florida ends up dealing off all its starters nearly every sim, often finishing with 110 losses or so as a result. These players aren't making any money, and they're all pretty young. Something should probably be adjusted so such doesn't happen, but I'm not sure what exactly.)

Also, in-season, I think starters are traded too often. (Starters as in, starting players, not just pitchers) Nearly all the trades involve starters, but in real life, those deals pretty much only happen around the deadline. Most real-life trades besides deadline ones involve minor leaguers and bench players. I'm running season sims right now, and in the one I just did, 22 trades were made in-season, and they all involved either starting position players, starting pitchers, or prominent relievers. That's unreal. Such trades do happen more often in the offseason. As it stands, they occur way too often during the April-June area.

I think these issues should be pretty high on the priority list.

ohms_law

04-10-2008, 02:06 PM

One thing it needs to stop doing is making huge deals involving multiple star players and starters...
Definitely

but it would always happen in a series of deals with multiple teams
Exactly

Something should probably be adjusted so such doesn't happen, but I'm not sure what exactly.
I feel the same way. The small market teams really need a differentiated strategy to their AI. The biggest thing is that they need to be aggressive about keeping their arbitration eligable players, and looking to acquire more of them. Simply put, they need to over value them, and "under value" any free agency eligable players. They shouldn't exclude free agent eligable players of course, but they should under value them.

Also, in-season, I think starters are traded too often. (Starters as in, starting players, not just pitchers)
Absolutely. Again, this comes down to real valuation. The AI is extremely good at valuing individual players in isolation, but the one huge hole in the AI is it's lack of awareness in terms of putting together an actual team. The same concept which I outlined above could be used here: Over value the players who will be season(s) long 25 man roster players, and slightly devalue other players. If that one change were made (and fixed, and fixed again... yea, we know the drill) then that would likely solve the majority of the AI's weaknesses in one fell swoop.

I think these issues should be pretty high on the priority list.
Their #1, in my book.

HoustonGM

04-10-2008, 02:37 PM

Great ideas on the over-valuing and under-valuing that the AI should do. I agree.

HoustonGM

04-11-2008, 07:07 PM

So, it seems like 11.08 introduces offseason trades. With that, I think it's necessary to change the way teams value players depending on the time of the season. As you can see by looking at the past season in real life, swaps of starting players are much more common in the offseason/

HoustonGM

04-11-2008, 07:20 PM

Throughout my sims, if I see any whacky deals, I'm going to add them here, because more information and examples can never hurt. These happen in my roster set (due out tonight), and I'm not going to bother posting ratings as they're value is basically the same as real life and half a season of stats isn't going to affect it that much.

---

Within two days at the end of April (21 and 22), the Nationals revamped their roster. First they sent two of their top relievers, closer Chad Cordero and Saul Rivera, to the Rangers for RP Joaquin Benoit, LF Frank Catalanatto, P Chris Michalak, and C Adam Melhuse. That's an okay trade that I could see maybe at the deadline or in the offseason, and I wouldn't think anything of it any other time, except the next day it was followed by this monstrosity:

Washington receives: RF Nick Markakis, RP Dennis Sarfate, C Omir Santos, 3B Mike Costanzo, RP George Sherrill, and RP Greg Aquino
Baltimroe receives: 3B Ryan Zimmerman, SS Felipe Lopez, SP John Lannan, RP Joaquin Benoit, RP Luis Ayala, LF Garrett Guzman

The value is near even on both sides, but the trade makes little sense for either team - another example of proper player evaluation, but improper implementation of that evaluation.

----

As an example of all the "name" players being dealt in-season:

Rafael Soriano, Rafael Betancourt, Chris Duncan, Josh Hamilton, Zack Greinke, Aaron Harang, Jeremy Guthrie, Daniel Cabrera, Milton Bradley, Jimmy Rollins, Adam Dunn, Jack Cust, Aubrey Huff, Bronson Arroyo, Brian Roberts, Delmon Young, Adam Jones, Kevin Millwood

Now, of course, a handful of those players I can see being traded, and a handful I think have a snowball's chance in **** of being traded, but all together, that amount of player movement with players of that status is unreal.

The Reds shipping off Harang and Arroyo is an example of a team deconstructing its roster midseason which would never happen. The A's in real life this past offseason had a chance to ship off their top 2 starters of similar quality, but only dealt one, and again, it happened in the OFFseason. For what it's worth, the Reds return for Harang was Zack Greinke, and they got Brandon Moss and Jed Lowrie for Arroyo.

----

Here's a trade that deconstructs the Pirates starting lineup while only gaining one player. They picked up Hanley Ramirez for starters Ronny Paulino, Jason Bay, and Adam LaRoche, plus one of their top relievers in John Grabow.

----

The lower market teams trade much more often than other teams. This, I suppose, is partially true. However, again, it happens in the offseason, and I think it still is currently too much. In one sim, the Marlins made 12 trades between April 20 and July 20. The players they picked up in trades, they shortly after flipped to other teams. By July 20, they had traded away ALL 4 of the Pirates they had gotten for Ramirez.

ohms_law

04-12-2008, 03:40 AM

With 11.08:

14640

14641

:eek:

This goes back to something I proposed the middle of last year: The total trade value according to the AI should be progressively reduced as more players are added to the deal. It's as simple as taking off 10% per player from the value evaluation. Probably... I'll have to try to mess around with it and see what happens.

HoustonGM

04-12-2008, 03:57 AM

Wow. Those are more extreme than anything I've seen recently. Holy smokes.

ohms_law

04-12-2008, 04:19 AM

A large part of the problem is that the player values on either side of a trade proposal are totaled, so you can "make up" the value of one great player with several average or worse players. Of course, we know that doesn't work out, which is why these deals stick out so much (aside from the fact that they're just huge deals).

SirKodiak

04-12-2008, 04:41 AM

Perhaps using a logarithmic scale for trades would be better?

ohms_law

04-12-2008, 04:47 AM

Probably... That's not as simple, but it would probably work better.

CatKnight

04-12-2008, 11:42 PM

Here's another trade that looks....wrong: (All ratings commissioner mode)

January 1972

SD to CLE:
1B Bill White (70 overall/71 peak) [38 years old] (assigned to Bench)
SS Tommy Dean (65/66) [26] (AA)
RP Frank Reberger (80) [27] (MR)
RP Dick Mills (67/68) [27] (AA)
RF Mickey Stanley (85) [29] (Starter)
C Ron Slocum (71/72) [26] (AA)

CLE to SD:
2B Vern Fuller (79/80) [28] (Bench)

So...the AI has learned the "pile players on" strategy...and maybe doesn't do it that well. :)

ohms_law

04-13-2008, 01:31 AM

So...the AI has re-learned the "pile players on" strategy...and maybe doesn't do it that well. :)

Fixed
:rolleyes:

SirKodiak

04-13-2008, 01:48 AM

Just keep in mind that trades don't just involve talent, they involve contracts as well. So the finances of each team plus the talent/contract ratio must be considered as well.

ohms_law

04-13-2008, 01:51 AM

oh btw, doing this:
(All ratings commissioner mode) doesn't make sense and isn't fair to the AI. The AI makes trades using it's own team's scouting differential, so evaluating trades with perfect information is an apples to oranges comparison.

HoustonGM

04-13-2008, 03:07 AM

Giving up 5 players for a bench player shouldn't happen, regardless of ratings. :p

ohms_law

04-13-2008, 06:44 AM

True, I was just making a point is all.

HoustonGM

04-13-2008, 06:44 AM

I know :)

grasshopper

04-13-2008, 08:10 AM

I've had the same thing, especially with Florida. Hanley Ramirez will probably be a guy we tell our grandkids about, but they keep trading him by 2010 almost every time. Plus, he makes the minimum salary. Also, Nick Markakis on Baltimore. He is the the guy they are building their team around for the future. He ends up on another team almost every time. What kills me is that they always trade way up in salary.

I hate to say this, but maybe there should be a franchise setting on certain young players.

CatKnight

04-13-2008, 06:02 PM

True, I was just making a point is all.

Hm...and what's this point?

True, the AI wouldn't have used commissioner ratings when it made its trade...but it wouldn't have used my team's either. I suppose I could have gone in under CLE and SD to see what they 'thought' of the trade...but commissioner mode DOES tell us whether, objectively, it was fair.

It might be worth noting that CLE's scouting is a B, SD's a C in that particular example.

If B and C scouting is so far off that it makes the AI think that trade is balanced, then I would suggest we have a bigger problem than the computer "piling players on" and that FRS is absolutely right, expenses need to be fixed and/or the scope between A and D scouting/farming narrowed SUBSTANTIALLY.

ohms_law

04-13-2008, 08:40 PM

Define "balanced". In terms of perceived value, I guarantee that the trade is balanced, since that's the way the program is written. In terms of how useful that perceived value actually is though... That's a whole different story.

CatKnight

04-13-2008, 08:50 PM

Hm. Maybe you can define how the computer evaluates players then.

Completely ignoring where the players wound up (as we know the AI doesn't look at the entire team,) I'm looking at a 70/71, 65/66, 80, 67/68, 85 and 71/72, only one of these players (well) past their prime, balanced against a 79/80.

Now, SD *is* substantially over budget, so I could buy a payroll dump strategy as this would pretty much bring them into balance. However, the two minor leaguers (65/66, 67/68) don't help that strategy and (shouldn't) be an incentive to CLE.

How bad does their scouting have to be to convince SD that giving up an 80 AND an 85 (along with assorted others) is worth a 79/80? What are these parameters for balancing trades that I'm not seeing?

Zeth

04-14-2008, 11:45 PM

There are some pretty simple ground rules we should be able to tell the AI to go by in trades (with other AI teams or with a human team):

1. Players with Overall below about 78 and Peak below 80 have no trade value unless the team acquiring one is (a) in contention and (b) plugging a gaping hole in their roster. (If the top 2B in the organization is a 66/72 guy, they might be interested in acquiring somebody's 77/77 guy.) Such players should be ignored in a trade.

2. Cap trades at 6 players per side. I can't remember the last time I saw a trade with more than 6 players going in either direction.

3. Increase the trade value assigned to a player's Peak rating a little. A young 65/95 player should be more valuable in a trade than the AI currently believes he is.

4. Increase small market teams' urgency to trade star players who are about to go FA at the deadline, if they're not in contention. Similarly, increase any non-contending team's urgency to trade medicore-to-good veterans for whatever they can get around the deadline. This just keeps things interesting.

HoustonGM

04-15-2008, 04:33 PM

Some of these trades are making less and less sense. For example, the Marlins sent Hanley Ramirez to the Mets for:

Carlos Delgado, Moises Alou, Ryan Church, Pedro Martinez, Ruddy Lugo, Raul Casanova, and Brant Rustich.

This trade is total nonsense for both sides. The Mets already have Jose Reyes. The Marlins just gave up their top shrotstop making the league minimum for 3 old injury-prone expensive guys (Delgado, Alou, Martinez) and others. Hanley Ramirez ended up playing a pinch-hitter role for the Mets.

Oh, and then, just for good measure, the Marlins later sent Delgado back to the Mets (in the same season) for a minor league pitcher. They also shipped off Pedro and Church to other teams later on.

I think this should be an urgent fix :(

ohms_law

04-15-2008, 05:37 PM

The frequency needs to be dialed back down some. That's the immediate problem, anyway. As I've been saying though, the larger problem is simply awareness. The value of these deals, considering the contracts and the player's skills, may be OK, but they don't consider the impact on the teams at all. The AI lacks team awareness.

HoustonGM

04-15-2008, 05:41 PM

I can't figure out how that Mets-Marlins trade has equal value...but nonetheless, it seems like it doesn't even take into account its own financial status, except "we're small-market, trade!" and then it makes trades without regard to finances. Trading a league-minimum star player for 3 high-priced guys in their late-30's and early-40's is ridiculous.

ohms_law

04-15-2008, 05:43 PM

ummm... yea, I see your point.

GreenDiamond2

04-15-2008, 06:04 PM

Rate players primary/secondary position. Auto sort lineup and rotation (including minors). Calculate team value. Create new auto sort using proposed trade. Compare team rating. Calculate proposed salary differential in terms of budget/current standings. Accept/Reject/Counter-offer.

gosensgo101

04-15-2008, 06:18 PM

I think that having the AI look for holes in the team, where certain guys fit in, etc., this would solve a lot of the bad trades. As the AI would primarily trade for needs with places where they have extra talent.

200tang

04-17-2008, 05:31 AM

I just had one with the Marlins using HGM roster (not sure if it makes any difference) v11.09

TO STL
Dan Uggla (MLB) 390k
Josh Willingham (MLB) 390k

TO FLA
Yadier Molina (MLB) 3.87M
Todd Wellemeyer (MLB) 1M
Hugo Castellanos (72/75) (AAA) 30k
Troy Glaus (MLB) 11.78M
D'Angelo Jimenez (BENCH) 390k
Anthony Reyes (MLB #3 STARTER) 390k
Josh Phelps (AAA) 30k
Cliff Polite (AAA) 30k
Adam Kennedy (MLB) 3.75M

I wouldn't say this deal is HORRIBLE, but the Cardinals obviously shed a bunch of payroll and the Marlins now have a higher payroll than their budget mainly due to an aging Troy Glaus

3RunHomer

04-17-2008, 09:18 AM

FRENCHREDSOX

04-17-2008, 10:29 AM

Are the AIs of some teams "dumber" than other teams? The AIs of the Marlins, Orioles and Reds seem to be mentally handicapped, while there are other teams (Red Sox) who consistently steal big talent from other teams. My examples (http://forum.sportsmogul.com/showthread.php?t=174661) from a general discussion thread.

If some teams are programmed to be stupid, making them more like the smart teams would go a long way toward fixing the imbalanced trade problem.

No the AI is simply DUMB full stop - the difference is that "large markets" hide this through being better scouters & developers of talent (which has been discussed ad infinitum in 120 win dynasty threads).Simply put after 3 or so years the same teams are A+ (NYY/NYM/BOS/LAD etc) & the same are D+ (FLA/PITT etc) - thus when FLA trades they are at +/- 9 & at that level will ALWAYS get "screwed" in a deal & also they TRADE way too late.

IRL the Marlins trade their players contract year (or even PRE contract year)* whereas in BM they play them & let them walk to FA....needs to be addressed in Patch 9000+

* great example was the Beckett + Lowell deal.

ohms_law

04-17-2008, 11:02 AM

The AI is the same for all teams.

HoustonGM

04-17-2008, 11:29 AM

I wouldn't say this deal is HORRIBLE, but the Cardinals obviously shed a bunch of payroll and the Marlins now have a higher payroll than their budget mainly due to an aging Troy Glaus
It is horrible in the sense that it would absolutley never ever happen in real life. Teams don't just swap half their starting players with each other. Often times, it just seems like teams in Mogul make trades just to make trades. As I said above, it seems that a small market team recognizes that it's small market, and then decides "We better start doing something. Let's make trades" like the Marlins, and then just makes boneheaded nonsensical trades.

ohms_law

04-17-2008, 12:10 PM

That's been basically my point all along. The trades are only "nonsensical" because we realize that teams are moving large portions of their starting lineups. The trade portion of the AI just doesn't recognize that.

As for that St. Louis - Florida trade, I find that extremely interesting. One thing that the AI pays very close attention to is the team's payroll budget... Have the Marlins been "on the cusp" in that game?

HoustonGM

04-17-2008, 12:13 PM

As for that St. Louis - Florida trade, I find that extremely interesting. One thing that the AI pays very close attention to is the team's payroll budget... Have the Marlins been "on the cusp" in that game?
That trade makes more sense financially than the Ramirez for Pedro/Delgado/Alou trade I mentioned ealrier. :p

ohms_law

04-17-2008, 12:16 PM

True, but it's fairly unusual to see a typical Marlins type team picking up $10+ million/season players at all. Their valuation tends to fall below the acceptable threshold.

HoustonGM

04-17-2008, 12:21 PM

True, but it's fairly unusual to see a typical Marlins type team picking up $10+ million/season players at all. Their valuation tends to fall below the acceptable threshold.
The thing is, their payroll is ridiculously low, and their payroll budget is generally 10-20 million bucks higher. And that's realistic I'd say, but spending that money in the way they do in Mogul, by shipping off young cheap guys for old expensive guys, isn't logical. In real life, the Marlins are getting tons of money from the league and what not in real life, but they're being ridiculously stingy on player contracts. They could certainly support a $40 mil payroll or so, as they usually get up to in Mogul. (In my rosters, they start out with a payroll below $20 million and a budget about $20 million greater than that).

ohms_law

04-17-2008, 12:25 PM

Yea...

FYI, this:

In my rosters, they start out with a payroll below $20 million and a budget about $20 million greater than that
is making them quite a bit more trade happy then they should be. The AI sees that and basically says "I need some veteran talent so we can win now, since I can afford it". The AI isn't at all into "Tanking". You might want to bump down their fan loyalty a few grades, for balance reasons.

HoustonGM

04-17-2008, 12:31 PM

They're already at D-...and apparantly, I can't set it to F. It just refreshes at D-.

Either way, these are still fundamental problems with the AI, regardless of financial situation.

HoustonGM

04-17-2008, 12:36 PM

Also, they usually sign a boatload of free agents in the first few days, which brings their payroll up to around $30 million, so when they're actually doing the trading, their payroll isn't that much lower (~$5-7 mil) than their budget.

200tang

04-17-2008, 05:01 PM

Same game as the FLA/STL trade.

to CIN
Eddie Kunz (68/82)(AA) 20k
Brant Rustich (67/81)(AA) 20k

to NYM
Bronson Arroyo (84)(MLB) 8M

<Personally I don't see anything really wrong with this trade since Arroyo is their #3 starter and he has a 3.6 ERA>

Here's the trade that seems really weird.

To CIN
Justin Upton (85/95) (A)

To ARI
Adam Dunn (90) (MLB)

I don't see a need for this trade to happen and I can't see Arizona ever getting rid of Upton any time soon and then Cin trades Dunn away for Upton and stick him in single A.

CatKnight

04-19-2008, 05:18 PM

I had four recent AI trades in winter that are leaving me a little puzzled. Rather than give you specifics, the situation's nearly identical in all cases so I can generalize.

In all cases, a very weak team is trading away one good player for 1-2 prospects. The good player is in his 80s overall. The prospects usually hang around 68/87 or thereabouts.

On the surface this is the AI entering 'rebuild' mode. Well enough. However, with these being winter trades the 'rebuild' tendency is accelerated. One of the weak teams is nearing the point where it's almost expansion quality - and wasn't all that good to begin with.

Further, from a suggestions standpoint, do we want teams giving up on the season in JANUARY? These rebuild trades look solid (or at least credible) for July....not Christmas.

Any thoughts?

ohms_law

04-19-2008, 06:11 PM

Further, from a suggestions standpoint, do we want teams giving up on the season in JANUARY? These rebuild trades look solid (or at least credible) for July....not Christmas.
Good point. With off-season trading enabled, it's probably a good idea to adjust the AI for that anyway.

HoustonGM

04-19-2008, 07:35 PM

I'm putting this thread in with the Computer Trades thread. It deals specifically with similar things raised there (the difference between offseason/in-season trading, for example).

ohms_law

04-19-2008, 08:53 PM

Yea, I was thinking about doing that myself...

HoustonGM

04-23-2008, 06:32 PM

HoustonGM

04-23-2008, 10:51 PM

What's that have to do with trades?

HoustonGM

04-23-2008, 11:26 PM

They wouldn't have to trade as much if they had decent players nor would be as
'desperate' and make some 5 for 1 trade because they 'need' something
I don't think that's whats going on. After all, they ARE trading away their decent players...their stars even. I honestly have no idea why they're making these trades.

HoustonGM

04-23-2008, 11:38 PM

I think its because they have 'needs' to address.
The AI doesn't really identify it's own needs.

Also, judging by the trades, if its because they have needs, they are not identifying thier needs at all. Trading Hanley Ramirez for: Carlos Delgado, Moises Alou, Ryan Church, Pedro Martinez, Ruddy Lugo, Raul Casanova, and Brant Rustich.

The Marlins have two perfectly fine outfielders - Josh Willingham and Jeremy Hermida - who were similarly rated to Alou and Church. Their first basemen Mike Jacobs has similar ratings to Delgado. I suppose they have a need in the rotation and bullpen for Pedro and Lugo. Rustich is just a filler minor leaguer, and Casanova is a backup catcher, of which the Marlins have one already of similar ratings.

Meanwhile, making that trade created a huge need at shortstop!

HoustonGM

04-23-2008, 11:43 PM

Maybe the issue is they don't know there own needs then.
I'm fairly certain that we've established that the AI does not identify team needs or what it currently has.

I think just limiting trades is a stopgap. We need identify a core issue then everything down the line will be easily adjusted.
Nobody's suggested limiting trades. The AI needs improvement.

FRENCHREDSOX

04-24-2008, 04:32 AM

It's just a hunch, but fixing this might subtlely help that whole "super team" problem. The Marlins routinely lose 100-115 games in my 2008 sims because they trade away their good players, and then trade away whatever decent players they got for their good players. Now, of course, that's FAR from the only thing that needs to be done for the competitive balance issue, but it's still one step in the right direction.

Just for fun I altered the CASH,Average City Y (Income for those not used to the abbreviation) & Fan base for FLA & "made them" a 120 million revenue team & THEY STILL traded away Hanley,Olsen,Uggla before 2011!

Ok,these were using my rosters (not GM's*) & their 08 season average was 80.2 but the point remains,why ??

* but the 2 sets are pretty close in their ratings of who is & who is not a FLA star ;)

SirKodiak

04-24-2008, 04:47 AM

Just for fun I altered the CASH,Average City Y (Income for those not used to the abbreviation) & Fan base for FLA & "made them" a 120 million revenue team & THEY STILL traded away Hanley,Olsen,Uggla before 2011!

Proof that Loria owns the Marlins, even in BM :p How much more realistic can you get than that? :rolleyes:

HoustonGM

04-24-2008, 11:33 AM

Just for fun I altered the CASH,Average City Y (Income for those not used to the abbreviation) & Fan base for FLA & "made them" a 120 million revenue team & THEY STILL traded away Hanley,Olsen,Uggla before 2011!

Ok,these were using my rosters (not GM's*) & their 08 season average was 80.2 but the point remains,why ??

* but the 2 sets are pretty close in their ratings of who is & who is not a FLA star ;)
That's troublesome...

SoulMan

04-24-2008, 12:26 PM

Like the title says...players are being traded more than once in the offseason. For some reason, Cincinnati seems to be the primary culprit - 14 offseason deals in 1902/1903, which included trading for Bobby Wallace twice and trading him away twice.

SoulMan

04-24-2008, 12:27 PM

Btw this was with trade frequency set at +0%

HoustonGM

04-24-2008, 03:34 PM

Man, wish there was a "Commisioner approves all trades" option until this gets worked out
Simple...after a trade, use Commish Mode to force the players back if you don't like the trade. That's what i do.

And actually, for my current game, I went through and gave the serial offender players (most of the Marlins, Harang, Arroyo, etc.) no-trade clauses. They sometimes do get traded since they can approve trades to certain teams, but rarely.

HoustonGM

04-25-2008, 03:24 AM

Just had these two whacky unrealistic trades:

ohms_law

04-25-2008, 03:25 AM

lol
nice

HoustonGM

04-25-2008, 03:32 AM

Let's just say the Yankees and Red Sox really got the good end of the stick there. Getting Martin allowed the Yankees to DH Posada, and then they acquired Paul Konerko when a decently performed JasonGiambi went down for the season. And what'd they give up for Martin? A gaggle of relievers.

The Red Sox improved their already strong bullpen with one of the best relievers in the game in Betancourt, and they gave up one frustrating prospect in Hansen, an aging medocre pitcher in Wakefield, and a bunch of mediocre bench players.

:( It's really starting to frustrate me

FRENCHREDSOX

04-25-2008, 06:32 AM

That's troublesome...

Seems that AI has a "memory" lag even when you have altered finances/city data etc (thus the FLA trades in the 120 million sim) or simply BAD AI (again!)

dizzyorange

04-26-2008, 04:08 PM

I'm a newbie but I agree the computer trade AI is a bit easily fooled. For example, started a new season with Tampa Bay and immediately made this trade:

Troy Percival
Al Reyes
Dan Wheeler
Carlos Pena (making 8.5 million)
Trever Miller
Gary Glover
Rocco Baldelli (out for the season)
Cliff Floyd
Jeremy Hellickson
Kurt Birkins (injured for 50 days)
Chad Orvella (injured for 40 days)

FOR

Joba Chamberlain
Ian Kennedy
$8,500,000

GreenDiamond2

04-28-2008, 03:34 PM

Seems like the AI doesn't consider the impact of injured players as much as it should, in weighing a trade.

In 2008, I think it is too ready to part with large sums of cash for players ... or take on large salaries in a trade deal. I think the AI should add some more weight when taking on these liabilities, when acquiring them. (not necessarily when trading them away).

Sure a hot prospect may be worth 15-20 million dollars, or there may be salary room for an 8 million dollar bench contract, but the AI should be deciding whether that money would be better spent in scouting or farm development, or as a reserve for a future stadium improvement.

MizzouRah

04-28-2008, 09:09 PM

I'm not seeing any of this.. although I've been using Houston's 2007 rosters and transferred my franchise from BM 08.

ohms_law

04-28-2008, 11:31 PM

I'm pretty sure AI teams can't nor need to save cash to build stadiums. I think they just pop up in the correct year.
correct

GreenDiamond2

04-29-2008, 01:06 AM

I'm pretty sure AI teams can't nor need to save cash to build stadiums. I think they just pop up in the correct year. I'm a proponent of having the stadiums upkeep as a tax and money sink for big market teams.
I haven't really paid much attention to when the new stadiums pop up, but in long sims starting in the current year, I doubt long term data is available beyond 2012 or so for stadiums under construction.

Using the stadiums as a money sink is good. This idea might be applied to stadium maintenance, future stadium funds (possibly as a budget slider which sets the funds aside and doesn't penalize the team's attendance for reserving cash), or for minor stadium improvements.

Smaller market teams would tend to build multi-use facilities (soccer/football/concert events) and might get some financial assistance when looking to build/upgrade a stadium.

ohms_law

04-29-2008, 01:37 AM

There is no data going into the future. The only stadium changes taht happen are those that already have.

GreenDiamond2

04-29-2008, 01:40 AM

Especially if the price of beer is cheap :)

Any mechanism which creates a money sink or redistribution for high attendance teams is good. Also good if the lower level teams have a means of improving their stadium or marketability over a period of time. Possibly a gradual increase in the amount that cities are willing to finance stadium renovations/upgrades as they decay. When a stadium gets too old the city should either pony up the cash (if the team doesn't have financial reserves) or a new city should provide incentives to move.

I also liked the idea of variable city growth on an annual basis. Which would keep markets relatively stable. While population decline is a factor in some areas, these are often temporary and over time population definitely trends towards increasing. While this isn't always in the cities, it probably tends to affect suburban areas significantly.

(Us population by year) http://www.npg.org/facts/us_historical_pops.htm

HoustonGM

05-20-2008, 07:51 PM

Just had another gigantic Hanley Ramirez trade, this time coming in the offseason after 2008...sending Ramirez, Mark Hendrickson, and Lee Gradner to the Giants for 10 players....

Arctic Blast

05-21-2008, 05:20 AM

Really, the biggest issue with trades seems to be that the AI simply equates quantity with quality, like they're one and the same. Throw enough chunks of tin foil at it for that diamond it's hording, and soon it just sees you offering a whole lot of shiny stuff for its' one piece.

ohms_law

05-21-2008, 05:57 AM

Yup

GreenDiamond2

05-21-2008, 10:11 AM

There is no data going into the future. The only stadium changes taht happen are those that already have.

There is available data (outside of the game) on stadiums that are in construction or planned.

Really, the biggest issue with trades seems to be that the AI simply equates quantity with quality, like they're one and the same. Throw enough chunks of tin foil at it for that diamond it's hording, and soon it just sees you offering a whole lot of shiny stuff for its' one piece
Generally true, although a lot of players in the farm system are likely to yield an occaisional surprise benefit, particularly if the players are fairly young (thus the reports of trades involving "four players to be named later..." ).

The AI does protect itself by adding fairly significant value (in 08) when it is asked to part with several prospects from a human player, this should probably not be reciprical.

It may be applying that logic in its own trades as well, thus the presence and frequency of large multi-player deals. This may actually be heightened by the AI not using cash in its internal deals at all, as it will require more complex deals to create what it percieves as balanced trades.

Scouting accuracy may be a factor in some cases.

FRENCHREDSOX

05-21-2008, 11:18 AM

Generally true, although a lot of players in the farm system are likely to yield an occaisional surprise benefit, particularly if the players are fairly young (thus the reports of trades involving "four players to be named later..." ).

The AI does protect itself by adding fairly significant value (in 08) when it is asked to part with several prospects from a human player, this should probably not be reciprical.

It may be applying that logic in its own trades as well, thus the presence and frequency of large multi-player deals. This may actually be heightened by the AI not using cash in its internal deals at all, as it will require more complex deals to create what it percieves as balanced trades.

Scouting accuracy may be a factor in some cases.

The trading is extremely lop sided even amongst itself -----------many multiple player trades during the season in which BOTH teams "rip" up their Opening Day line ups & even then,they "flip off" the "prize" acquisition within 24 hours.

I am not saying that teams don't trade but in 90% of cases the 25 man roster on OD is the same as the end of season roster with 2 or 3 changes through trades.The rest of the changes are call ups or pick ups in FA.For example the only "big deals" during the 07 season were the 2 Texas deals with Boston & Atlanta........

GreenDiamond2

05-22-2008, 08:14 AM

There are the occaisional "three way deals" which is essentially what a quick flip represents. Major trades involving a number of starting players are fairly rare though.

Most of the bigger trades are a couple of big names and some minor league players/prospects as "currency" since the league limits cash deals.

The game only models the better minor league talent, some deals involving lesser talents are almost like trading farm system development points, strange idea, just popped into my head, so its not thought out at all.

Large power deals have probably occoured, but definitely not common.

FRENCHREDSOX

05-22-2008, 10:30 AM

Large power deals have probably occoured, but definitely not common.

That is the point BM has multiplicity in this domain,even Division leaders rip up their team which although logical in BM (get a 89 instead of your 87 position player) never happens IRL.

Teams that win IRL stick together (team spirit/cohesion) it is the 24 others WHO try either :

a) improve to get into the playoffs or
b) "sell off" veterans at their peak value (trade deadline deals to (a))

CatKnight

05-22-2008, 11:28 AM

(Referencing ohms' comment there's no stadium data going into the future)
There is available data (outside of the game) on stadiums that are in construction or planned.

I'm fine with no future stadium data. Stadiums that are under construction or planned...don't exist. We don't know if they'll be finished, or if they'll open on time, or if rabid woodchucks won't chew their way through the support pillars and it'll collapse after two weeks. They don't need to be in the game.

Regarding trades, the computer is MUCH better than it has been in the past. Yes, it'll still pull some really weird..stunts...but it seems to be doing better.

I've turned off winter trades (by reducing Trade Freq to -100% then re-upping it in April) though. There are teams that are trading like it's July: Dumping prospects to 'win now', or vice versa.

ohms_law

05-22-2008, 11:26 PM

Much better yea, but it still needs improvement. The primary problem is still a lack of overall team awareness.

Your correct in that offseason trading needs to be adjusted as well.

GreenDiamond2

05-23-2008, 07:35 PM

Much better yea, but it still needs improvement. The primary problem is still a lack of overall team awareness.

Your correct in that offseason trading needs to be adjusted as well.

Agree with both points. There is no reason why a team, especially a successful one, would desire to make massive roster changes. Even if they desire to do so from an ownership perspective, they have to find another team that is also in need of, and has an inclination to do so. Even when such moves are considered, it would be extremely rare that a workable transaction could be made. Roster awareness would go a long ways in that regard, as it would more accurately reflect the difficulties of putting together a blockbuster deal. Trades with personnel needs considerations taken into account would probably nix a lot of these in the bud, at least as far as inter-AI trades.

In terms of offers from the AI to player, most seem to happen near the deadline. Ideally they would have their motivations worked into the AI for moves at that point.

Interesting that the "friends" aspect has such an influence on happiness, yet the AI seems to maintain high levels even when doing massive trades. Depending on their prospects to be utilized, most player are not likely to be thrilled by the trade occouring. A temporary decline in happiness might be worth considering, especially if they are losing starting lineup priveleges as a result of the exchange. If the player seems likely to be elevated, and are unhappy in their current situation, they may actually respond positively. Roster awareness would probably help the game in making that determination.

I'm fine with no future stadium data. Stadiums that are under construction or planned...don't exist. We don't know if they'll be finished, or if they'll open on time, or if rabid woodchucks won't chew their way through the support pillars and it'll collapse after two weeks. They don't need to be in the game
No real preference on this, while possibly interesting it may well represent unnecessary work based on speculation, although these projects stand a reasonably high probability of completing.

However in long running future sims, teams should have a mechanism for moving, building, maintaining or upgrading stadiums, even if randomly generated. Again this also provides one of the possible money sinks, so is part of the greater AI issues, and a possible avenue of resolution.

ohms_law

05-23-2008, 09:04 PM

However in long running future sims, teams should have a mechanism for moving, building, maintaining or upgrading stadiums, even if randomly generated. Again this also provides one of the possible money sinks, so is part of the greater AI issues, and a possible avenue of resolution.
I agree, but this is something for another thread. We can do this sort of thing ourselves easily enough, so It's not that important anyway. Just something to actually pay attention to.

If the player seems likely to be elevated, and are unhappy in their current situation, they may actually respond positively. Roster awareness would probably help the game in making that determination.
This is actually the main thing that I've been harping about, when it comes to both computer trades and (re)signings for a couple of years now. In terms of signings, there's no way that a player would ink a deal with a team where they're guaranteed to receive less playing time. There's no way that, say, Hanley Ramirez would sign with the Yankees right now for example (unless they offered him something ridiculous like $20 million per for 10 years, or something). In terms of trades though, you're probably on the right track when it comes to happiness. For some of these deals it's bad enough that the player might even refuse to report to their new team...

GreenDiamond2

05-23-2008, 10:14 PM

I agree, but this is something for another thread. We can do this sort of thing ourselves easily enough, so It's not that important anyway. Just something to actually pay attention to.
Agreed. Without belaboring the point, it is something I would like to see automated, although far from my highest priority. I think it would be an interesting advance for the product as a whole.

This is actually the main thing that I've been harping about, when it comes to both computer trades and (re)signings for a couple of years now. In terms of signings, there's no way that a player would ink a deal with a team where they're guaranteed to receive less playing time. There's no way that, say, Hanley Ramirez would sign with the Yankees right now for example (unless they offered him something ridiculous like $20 million per for 10 years, or something). In terms of trades though, you're probably on the right track when it comes to happiness. For some of these deals it's bad enough that the player might even refuse to report to their new team...
Especially where the player is above league average for their position, particularly with proven experience. A player at the end of the aging spectrum may be more willing to accept a minor role on the team. Players with salaries in the top 2/3 of the league for their position would be resistant. Players in the top 1/2 of skill and salary would probably make an issue of it.

If such a player wasn't in the lineup by the time spring training started (barring injury), they would probably be raising all sorts of ****. I am not in favor of binding contractual playing time agreements (probably as a percentage of non-injured time) as they would cause a lot of micromanaging and other problems. Might be interesting to see a non-binding playing time negotiated in contracts at some point.

Basically the GM would have the option of ignoring that point (behavior/performance clauses or whatever was leveraged behind the scenes) but the result would be a penalty to the player's happiness (and consequently team chemistry). This would tend to haunt the team in terms of performance and/or negotiation next time around. At a certain level it might trigger a trade demand as well.

Something to think about.

GreenDiamond2

05-23-2008, 10:15 PM

COMPUTER TRADES
Since this is the thread topic, I am throwing another thing out there for consideration. This particularly weighs on mid-season trades, but is probably always an influence. This might be programmed in to some degree already, not sure.

These are more true if the team is contention, in terms of trade preference (trade value weighting). Essentially a list of common-sense priorities, especially when trading away starting players. You usually do not want to meet your own players in the post-season. You probably don't want to build up a team that you will be competing with next year either, unless you are absolutely sure you have the best of the deal.

1. Trade with non-contending teams in the other league
2. Trade with non-contenders in other divisions
3. Trade with non-contenders in same division (? always dicey)
4. Trade with challengers in other league (challenger = likely non-contender, in fairly distant running.)
5. Trade with challengers outside your division
6. Trade with contenders (in first or close) in the other league
7. Trade with contenders outside of your division
8. Trade with challengers inside your division
9. Trade with contenders inside your division

For purpouses of these rankings, I would consider any team in the wild card race in your league to be a contender as well.

One way to implement this would be to reduce a perceived (+.5 to +1) for each grade beyond the first to the acquisitions overall rating, from the trading teams perspective. This would simulate their reluctance to give any ground to an immediate rival or potential rival.

For example, You are offering a player rated 92, to a challenger inside your division in a trade deal. The AI would rate the player as an 88, (using half points), to determine what they would give up in exchange for that player.

This awareness would also be useful in helping the AI avoid some of the more dubious sabotage type deals.

Off season/early season deals would probably have this built in as well. Instead of taking the team's standing into account, it could determine contenders by basis of the teams overall rating, power ranking, or one of the other scores available in the standings chart at the time of the deal. This would help it decide which teams were safer to deal with.

Additionally, using this type of logic would probably help with team balance, as weaker teams would get more favorable trading terms, generally speaking. Since each team would rate from their own perspective, the worst teams would likely be the beneficiaries of the better deals.

ohms_law

05-23-2008, 10:27 PM

Pretty much all of that logic is actually already there... I think that the weightings are just all messed up, or something.

GreenDiamond2

05-23-2008, 10:43 PM

I had noticed that the "cash offers" list was prioritized to some degree, which gives some insight. Although this hits a ceiling by the amount of cash the team has available, which may have some carry-over effect.

Overall, the more dangerous teams seem to be willing to offer more, but I wasn't sure if that was situationally related or more a function of their bankrolls. If the weighting is in place, it doesn't seem to be all that apparent, although I haven't really made a deep study of the issue. Certainly not enough of an impact to help the weaker teams much, at this point.

metsguy234

08-05-2008, 03:16 AM

17662

The two I drew attention to I found to be ridiculously dumb moves by the AI.

filihok

08-17-2008, 08:21 AM

11.19
HGM's Rosters

BINGLEBOP

08-17-2008, 01:04 PM

Ouch... Atkins in the same trade. It's odd, and likely just a coincidence, but I have had three different sims where Garrett Atkins and Matt Holliday are traded to the White Sox in 2008, although not always in the same trade.

rockiesfan4ever

08-17-2008, 01:23 PM

It looks like the AI, for the Rockies, is just as smart as Dan O'Dowd.

filihok

08-18-2008, 07:04 PM

I think the White Sox GM is to blame

Now, they are $13,000,000 over budget but most of these moves don't seem to reduce payroll.

Obviously the White Sox are AI controlled and none of these trades involved a human team.
Trade Frequency was set at +/- 0

HoustonGM

08-18-2008, 07:05 PM

WTF. :confused:

Wow.

ohms_law

08-18-2008, 07:10 PM

Yea, there's definitely a bug there somewhere...

HoustonGM

08-18-2008, 07:11 PM

Filihok, what's your trade frequency set to?

filihok

08-18-2008, 07:12 PM

just updated...it was at +/- 0

filihok

08-18-2008, 07:13 PM

Now...this league is out of wack
(from the post I made about the cities data)
a lot of teams are over their budgets but only the White Sox were doing these kind of crazy trades.

ohms_law

08-18-2008, 07:15 PM

Yea, there's some variable that's become out of whack. It's not Trade Frequency since it's only one team that you're noticing the problem with. It's one of the more internal variables that's somehow become "corrupt". It's probably an overflow problem, based on something that I've already emailed Clay about.

filihok

08-18-2008, 08:32 PM

So, does this mean I have to spend the rest of this game's life reversing trades made by the White Sox? :)

YEAH DAAAAWG

08-18-2008, 09:08 PM

Is it safe to assume that since this is in Bug Reports now that the Trade AI (or overall AI?) is being worked on/improved for a future patch?

HoustonGM

08-18-2008, 09:31 PM

Is it safe to assume that since this is in Bug Reports now that the Trade AI (or overall AI?) is being worked on/improved for a future patch?
It was moved to Bug Reports as Clay has decided that AI problems should be considered bugs.

asianinvasion

08-19-2008, 02:42 AM

Willy Mo Pena for Victor Martinez? Seattle got a steal. Its not like them. Usually, its them trading a guy like Victor for Pena.

geniuslegume

09-05-2008, 01:08 AM

In 1936 the Washington Senators offer me Jimmie Foxx (93)--28 years old, $10k salary on contract for 6 years--and a 30 year old relief pitcher (77) on a $1.8k/4 year contract for 32 year old LF/1B (76) on $7.3k/4 year contract and a RP (80) on a $9.9k/1 year contract.

Maybe somewhere in the AI's logic it had some reason for doing this, but I don't quite get it. This is with -20% trade frequency. Upon checking a little deeper, Washington IS 7 games back, but then again, it's only the middle of May, too.

asianinvasion

09-05-2008, 01:48 AM

Its not that bad of a move compared to the ones the Sox and Marlins were making. I kinda can see the logic behind this as the reliever will pretty much cancel out Jimmie Foxx and he will be off the books at year's end. They may ship the LF out in the next coming days.

HoustonGM

09-15-2008, 02:02 AM

I just have to bump this because it's really annoying me.

Small/mid market teams that aren't currently good should be building for the future. That means NOT trading away their young, cheap, GOOD players.

In my game, I introduced 2 expansion teams. The one team from Vancouver got a gem in the expansion draft that I manually ran - centerfielder Drew Stubbs. He didn't do much in 300 or so at bats with his previous team over 2 years. His first year in Vancouver, the team's first year, he made the All Star Game, hitting well, stealing bases, and playing good defense. During July, Vancouver traded him to the Dodgers, a division rival. I, as I follow this league closely, forced the trade back. Vancouver traded him back again. And again. And again. Then the offseason comes, and AGAIN, they trade him to the Dodgers. Again, I force it back. Then they trade him to the Phillies for a 35 year old Pat Burrell.

He hit .285/.393/.483 with 19 HR, 57 RBI, 26 SB, 110 runs. He's not yet arb-eligible, just turned 27. They should be looking to keep him, not trade him. Teams seem to always been in "win now" mode, regardless of their situation, and it's detrimental to them AND the league's competitive balance.

OH, it's also worth noting that the Dodgers have 3 perfectly good starting outfielders, rated 93, 88, and 86. They have no need whatsoever for Stubbs.

BINGLEBOP

09-15-2008, 02:21 AM

I've noticed that even more... AI teams trading for a star player at a position where they already have a star player! It's not limited to the Yankees where they have star players in the minor leagues sometimes.

Not only do they not have the "smarts" to use a shortstop at second base (for any length of time) when have two star shortstops, but it ruins the player's career when they have a string of All-Star seasons and all of a sudden spend an entire year on the bench.

ohms_law

09-15-2008, 05:41 AM

agreed

filihok

10-06-2008, 10:14 AM

Or the AI Dodgers signing Rafael Furcal for 4 years 5.3 million and two months later signing Edgar Renteria for 2 years and $4.7 million.

or the AI Twins signing 3B Casey Blake for 2 years 2.5 million and two days later 3B Joe Crede for 2 years 3.5 million.

And...these kinds of trades.

actionjackson

10-23-2008, 03:55 PM

Small/mid market teams that aren't currently good should be building for the future. That means NOT trading away their young, cheap, GOOD players.

I guess this is another bump because this and other methods of sabotage by crappy teams is bothering me as well, such as oh, I don't know:

1) Increasing ticket/concession prices to further alienate their already fed up fans.

2) Slashing spending to nothing thus guaranteeing continued craptacularity in the future.

3) HoustonGM's complaint about trading away valuable chips or building blocks for garbage.

So, I've taken matters into my own hands by equalizing cities, re-jigging fan loyalty at season's end (sort of in reverse order of that season's records) and erasing the debts of bankrupt teams and giving them $1 to prevent bankruptcy languishing. That turned out to not be enough to prevent the sabotage (these crappy teams genuinely want to suck), so I went even further.

I now set ticket prices, concession prices and expenses for each team daily during the season (weekly in the offseason) in commish mode before the day's games are played. This seems to be working so far as the teams with the most cash (generally smaller payroll teams thanks to equalized cities, fan loyalty adjustments, and bankruptcy elimination) tend to put the most money into the farm system and scouting. Medical expenses are a different story. Teams that are having a successful season currently tend to spend more (as they should) unless they're running short on cash, which tends to happen to teams with ridiculously high payrolls.

Currently, I'm not running a team, as I prefer divisional play and more than one round of playoffs and that doesn't happen until 1969 (right now I'm in 1902), so that's when I'll take over the worst team in the league at the start of spring training (according to "Overall Rating" in the Standings screen). It's my own little experiment to see if I can improve the financial balance of the game, which I believe will solve most of the competitive balance issues that the game as it stands faces. I intend to maintain this system even when I take over a team, which shouldn't be problematic, since my payroll is always at or near the bottom. I really hope it gets rid of the stupid expensive free agents in AAA, AA, and A issues, but we'll see (I don't expect miracles).

Maybe I'll even post some Noll-Scully ratios and "extreme" team results as well as playoff contestants and World Series winners as the decades go by. Thanks to CatKnight for the original ideas on equalized cities, fan loyalty adjustments, and bankruptcy elimination. Just taking it a little further to save the moronic teams from themselves and thus make the league better. I'll be happier when the game can do this for itself, but until then clickety-clickety-click-click-click. The acid-test of course will come in the big money era. We'll see how things hold up then.