Matrix Games Forums

Started in 1951. Took 3 years before winning world series. Now finished 1958 season and won 5 WS in 8 seasons. Have assemble unbeatable team thru trades and free agency. Usually out bid for free agents, then make 2 for 1 trades to unload high price free agents for league all stars. 2 for 1 trades lowers team salary under $2M to allow higher bidding on free agents and restart the cycle over again. Going after Willie McCovey in 1959 season.

1951 76-86... (computer does good job drafting)
1952 88-74
1953 102-60... (champs)
1954 119-43... (champs)
1955 119-43
1956 120-42... (champs)
1957 122-40... (champs)
1958 134-28... (champs)


1958 team

134-28 for .827 win pct. avergaes 6.3 runs per game, allow 2.7 runs per game. Team batting 0.293, 333 HRs, 443 SBs, pitching with team ERA of 2.50 and 45 complete games. All starters have at least 27 HRs for season. 5 man pitching rotation.

1B FRANK ROBINSON........ 41-112-.360....103-SB
CF WILLIE MAYS ...............40-121-.320.... 63 SB
RF ROCKY CALAVITO.........56-146-.319
3B EDDIE MATHEWS..........48-147-.321
SS ERNIE BANKS...............33-99-.298..... 77 SB
LF ORLAND CEPEDA...........27-93-.317 ... 55 SB
C DEL CRANDELL...............27-105-.275
2B DON ZIMMER................29-82-.237

JACK SANFORD...... 27-1-2.22
MIKE FORNIELES..... 24-3-1.80
HERB SCORE...........22-3-2.54
RUBEN GOMEZ.........24-6-2.94
DON NEWCOMBE.....18-6-2.98

CARL WILLEY........11-1-1.55
HARVEY HADDIX...4-4-3.31
JACK HARSHMAN...3-3-4.89

TOM ACKER...... 1-1-1.41..41 SAVES

Players on Bench... Roberto Clemente-OF, Gus Bell-OF, Harry Chiti -C.

I could restart and play without trades, but got to have trading to be baseball.

Though that's a bit too out-of-balance for me, there's something to be said when you can BENCH Roberto Clemente of all people and still do that well.

Impressive nonetheless, T-bolts.

[;)]

Yeah, I've been saying for a while that the AI needs a bit of help to reach competitive-with-a-human status. My last long-career I won 20 straight league titles, and I think I went 14-6 in the World Series over that time? Never had quite the 134-28 season - that IS impressive! - but I was winning my 8-team league by 18-25 games pretty regularly.

See URett's thread Improvements for GM's for some very specific ideas for things which would help improve the AI of the game. In particular, I think having the AI GM's carry their 'status' through the offseason would help: a 'rebuilding' team shouldn't be willing to trade you their 26-year-old All-Star for two high-priced veterans... but the Yankees might be willing to.

make 2 for 1 trades to unload high price free agents for league all stars.

I think this is one of the key weaknesses of the AI. The AI seems to be reasonably good at evaluating one-for-one trades with the human player, but goes all weak-kneed when we start offering multiple players. It will almost always accept 'two good players for one great player', likewise '5 decent prospects for 1 great prospect' seems to work out pretty well.

In the meantime, to get more enjoyment out of the game via more challenge, you might try

... playing with a 'small budget' team and trying to compete with 'big budget' teams - e.g., play with a team from the Dakotas and try to compete with New York/Chicago/L.A./ et. al.

... filtering trades through an 'is it fair to both sides' algorithm in your own mind, e.g., trading players you don't need to a team which needs them for players they don't need is reasonable, but just 'cherry picking' their best is not (even if the AI would agree to it).

... forcing yourself to let a few players' contracts expire as free agents every season, maybe 1-3 starters plus any 'quality' player (over 50/50/50) who isn't starting. (This prevents you from stockpiling too much, especially in the minor leagues, and will make your offseasons more interesting.)

... forcing yourself to keep a few stars through retirement, playing them regularly even after they are no longer the best player you could field - e.g., Tony Gwynn, Cal Ripkin, etc. (You'll love watching them work towards career records, your fans 'know and love' the player, and it'll make taking the title just a little bit harder.)

... building your superstar team, signing the 'core' of the team to long-term contracts, and then changing to a different team in the same division: try to overcome the juggernaut you've just built.

How to improve the game.

1) Teams should not give up star players in trades, unless they are at end of career. But how can a team turn down a trade involving 2 all-stars for 1 all-star.

2) I usually max out my team salary at 11M (1958), but all other league teams have 3 to 6 milion dollars under salary cap for their region. They are not using 50% of their money in free agent bidding. I am spending twice the money 11M to 6M on player salaries.

3) Other teams usually give up best player in 2 for 1 or 3 for 1 trades. Trades are fair but I have more depth in players because I am carrying twice the team salary, 11M to 6M.

4) I can stock pile players in minor league aquired in free agency and then do 2 for 1 trades to get any player in league.

5) I usually max out my team salary and then dump salary in trades around all star break and before trading deadline of 31 July. Any player that doesn't extend his contract is traded immediately. I transfer these high salary players to other teams thru trades, and in a way raise and inflate the team salary of other teams.

6) I have played puresim baseball since the begining. There was one verision of puresim several years ago which was hard to dominate the league, Shaun Suillivan flooded the league with high quality players, all the teams had great players, all the league teams max out team salaries, several great players were un-used because all the teams were out of money. This league was a difficult league to dominate because there wasn't a shortage of great players, but there was a shortage of money. There were many great players available in to sign as free agents but these players out price themselves in the market and teams spend all their money and couldn't sign all these guys. All-Star players didn't play for several years because their salary demands were too high for their skills.

7) Can change puresim to allow different levels of difficulty. Hard level would have more great players than TOTAL money in league and teams would be evenly balance due to glut (over-supply) of good players. Glut of good players would lower high & low statistical extremes in league. All batting averages would migrate to closer mean value of league. The reason nobody hits .400 today because all the bad pitchers don't exist anymore in the leagues. The higher level of competition eliminates extreme records (minus the steriods). Great players hits .625 in high school because level of competition is uneven but same player hits .280 in pros because the competition gets higher across the board.

Funny enough this is the exact same way the ai does in OOTPBB as well when it comes to trades. And I'm prone to unload my players that are going to be free agents and are old like 33 to 35 on it pretty easily for some star prospect. The one thing in OOTPBB that sort of keeps the game inline and fair for the ai is the minor league free agency feature, it's very hard to keep the pack of star prospects I have until I'm ready for them with that 6 year minor free agency feature.

Something else one can do in OOTPBB, don't know about Pure Sim is lower the salary cap to around 80 million and cash onhand to around 5 or 10 million. Makes it a lot harder to stack up a bunch of expensive free agents at least. The ai in OOTPBB during free agency will make one pay a premium price for some players, although it does itself for some that shouldn't be paid half of what they are getting. I've seen em pay milllions for a 38 year old whos stats just die during the season and they've signed him for several years. STUPID! lol

If I could wish for a feature it would be to be able to independently set the salary cap for each team. So I could help the AI to make the NYY and Boston and some of the more prominent teams of today better challenges and then limit myself to 80 million or so. I can do it abstractly for myself, but, would like more control over more teams besides my own.

Others have suggested some trade restrictions for themselves like picking a team you want to trade with, making the offer and not going through the list of other teams just to find the best you can get for your player, AND if the AI says no deal you have to wait 24 hours to a week before making another offer. Just sort of ties your hands somewhat from exploiting the AI when it comes to trading.

It is kinda funny reading what others do in Pure Sim that I've done in OOTPBB as well like spending my free agency money and getting the best I can, then unloading them for star prospects. It's really cheap doing that you know though? Or another one of my old time favorites unloading a semi old player who's having a bad year that's costing me oodles of salary and not producing on some other team for any garbage that doesn't cost anything for the rest of the season. Somehow they need to program in "sharing salary for the remaining year" at least when trades are made for good players gone bad. I mean any human would see right through this trade in a heartbeat and I doubt any fool would be ignorant enough to accept it.

ravinhood -

The fact that the 2-for-1 problem exists in OOTP is both funny, and unsurprising when one considers how AI has to work - you basically have an 'evaluation function' for any player's trade value, and to figure out if its a good trade, you add the 'evaluation function' results for the players offered against the results for the players requested: e.g., "Wow, two players who evaluate at '50' for a player who evaluates at '95'? Well, 100 is more than 95, so its a good deal."

I'd love to see the 6-year minor-league free agency added to PureSim, I definitely think its absence makes the game easier.

although it does itself for some that shouldn't be paid half of what they are getting. I've seen em pay milllions for a 38 year old whos stats just die during the season and they've signed him for several years.

Same in PureSim. Same in major-league baseball, actually, which is where we stumble a bit: does the game try to model the choices real GM's make, or try to provide a challenge for us?

If I could wish for a feature it would be to be able to independently set the salary cap for each team.

Can do, in PureSim.

if the AI says no deal you have to wait 24 hours to a week before making another offer.

That's something that we might build into the game UI, actually - FM'05 does this, for transfers. You issue a query, and the request comes back within a day or two, and if you want to haggle, that consumes time while potentially saving you money in the long-term.

However, in both OOTP and PureSim, the trade screens have an 'instant feedback' functionality, which lets a patient player take three hours calling every team in the league to negotiate without paying any cost. If it took them a month of game time to go through the same process, then we'd be less able to 'exploit' the AI over it.

I also like Front Office Football's "the price goes up if you take a deal off the table" mechanic: if you find an acceptable deal with Green Bay for Brett Favre, for example, and you say 'not now' and go off to talk to the Chargers about Brees and Rivers, when you come back to Green Bay you find that their price for Favre has gone up. Its a little artificial, but it works.

Teams should not give up star players in trades, unless they are at end of career.

Agreed.

How can a team turn down a trade involving 2 all-stars for 1 all-star?

That deal needs to depend on the team's needs. If they are a 'young team' rebuilding, and I'm offering two 35-year-old All-Stars for their 26-year-old All-Star, the team needs to say 'No'. If they are a veteran team in contention for their league title, which is weak at the two positions of the offered 2 all-stars and strong at the position of the requested 1 all-star, then they need to say 'Yes'. (It feels like at the moment, they accept for the logic I detailed above.)

6) .. flooded the league with high quality players, all the teams had great players..

I don't think that's the answer, but it might work similarly to flood the league with middling players - those of the 50/50/50 style who are competent but unspectacular at their position. Teams wouldn't trade for them because there would be equally competent players available via free agency, and it wouldn't help me to stockpile them.

I can stock pile players in minor league aquired in free agency

I'd also like to see the free agent consider 'playing time' when thinking about who to sign for. So, if you have three quality outfielders, it should be very tough to sign a fourth 'starting quality' outfielder via free agency - he'd rather go somewhere to start than come sit on your bench for extra pay.

Amaroq,

Ive done your suggestion of having fun creating a juggernaut team winning a few seasons then switching teams to try and compete the ebast ivecreated. very fun indeed, and i will probably do this again in my current asc.

eck

[:)] Its always a challenge, at least!

ORIGINAL: Amaroq

The fact that the 2-for-1 problem exists in OOTP is both funny, and unsurprising when one considers how AI has to work - you basically have an 'evaluation function' for any player's trade value, and to figure out if its a good trade, you add the 'evaluation function' results for the players offered against the results for the players requested: e.g., "Wow, two players who evaluate at '50' for a player who evaluates at '95'? Well, 100 is more than 95, so its a good deal."

Maybe if the trade evaluation worked ona more exponential scale? That is, a a 25 player would equal a '1', a 25-50 = '2', 50-75 = '3-5', 75-100 = 6-10? So two 50-ish players would barely have you in the same allpark as a 75.

Its a good start, and in fact I think this is what Shaun has tried to do.

Unfortunately, the problem still stands, because the problem rooted in the idea that 'two players are worth the sum of their evaluation', or 'that the sum of unequal numbers of players is comparable'.

Accepting your 'exponential' scale, let's say that, going into Spring Training, I have 57 players on a 50-player roster. My worst 8 are 25's (1). I want to trade them, rather than cut the worst seven. Say I can make four 2-for-1 trades to upgrade those 8 to 4 40's (2). Then I can make two 2-for-1 trades to upgrade those to 2 60's (4), and I'm still over my roster limit, so I trade those two 60's for an 85 (8).

Sure, I've had to spend some time doing it, but functionally there is no way I should have been able to trade eight dross players who will never make it in the league for an All-Star!

So, something else needs to be done. One other thing Shaun has done is set a 'minimum difference', e.g., the team won't make a trade that appears to be break even (so my example would break down), but wants to find a distinct improvement. You'll notice this most if you try to trade players who you think of as equal in your mind, that play the same position.

There's also the idea that some players should 'evaluate' as negative - for example, if any of my example 25's had a high salary, they should evaluate as negative, and the ones that have no real potential should evaluate as 'zero' to 'mildly negative', leaving only the ones with potential having a positive evaluation. That helps - and you can observe that Shaun's already done this as well, as you can set up trades which are barely acceptable to the opposition, then add your worst player and see the opposition change to rejecting the trade (the guy has a negative evaluation).

Perhaps we should have something more like a 'law of diminishing returns' to a trade, where the third- or fourth- best player in the trade doesn't add as much as the best player? Then we might be making it too easy to cherry-pick a large number of players in a single trade.

... so, its very tough to balance, as you can see!

I agree, values should be also given for AGE and SALARY when making trades with the AI. Surely some mathematical equation could be formed that would advantage the AI when it comes to trading away star young players for old failing stats players who just happened to have had a good year before they turn into mud.

What about franchise players? Say 3 players that can't be traded no matter what, you either have to pay them what they want or dump them into free agency and in some cases suffer the salary loss??

OOTPBB goes one step futher in the evalutaion that I can see, there's a setup screen for how the ai looks at players when it comes to trading and negotiation for salary. It had four settings that must equal 100%

1) Attributes atm
2) Last years performance
3) previous years performance
4) previous to the previous years performance

I think AGE and SALARY should have been in that evaluation perhaps instead of previous years performance or add two more options for percentages. I just feel AGE and SALARY are a big part of trades, of course these games do not really model the aging factors of some famous baseball players, it picks and chooses at random I'm pretty sure. But, if it would just recognize it's getting screwed for a trade of old high salaried players (and in some cases like when a player is past 35 it does in OOTPBB) then it wouldn't make some of those silly and stupid trades.

Since I know the game looks at 35+ year old players in trades for value, I've resorted to trading anything and everything from 33 to 35, continue to get good new star players for players that only have maybe 2 good years left at the most and their stats are already failing which is pretty easy to see.

Now, here's an idea, human players cannot see attributes/talent ratings of players, they must rely on trial and error and scout reports and performance records only?? It would be an "option" cause some people would scream bloody murder cause they couldn't see the attributes or talent ratings and be able to exploit the ai. lol

I like the idea of negotiation time like 2 to 3 days for the trade to go through or not idea though. That would help a lot to keep us from exploiting the AI with cycle through all the teams and see who gives us the best in one day. lol

If PURE SIM could pull these things off I'd make the move over. As it stands OOTPBB has just a weee bit more that keeps the player from exploiting the ai, though not enough either, but, sufficient not to get teams with 138 wins and 24 losses.

PureSim does include a 'youth and potential' factor and a 'salary' factor in its evaluation function.

Unfortunately, it doesn't seem to include a negative for age, so, as you've noted, that 33-35-year-old player, whose value is dropping in my eyes, is still worth trading for in the AI's mind.

As I noted in __Improvements for GM's__, I'd like to see the offseason AI split into multiple parts, depending on the team's current states, and 'age' plus 'opportunity cost' would be major factors.

Looking at real life MLB one could make an argument that salary and years left under contract should be the most important factors for calculating trade value. Not so long ago Manny Ramirez went through waivers and nobody would take him "for free" despite him beeing one of the best players in baseball. That means that his contract more than made up for his baseball value and gave him a negative trade value.

That's a good point URett and these baseball sims should model that if it's possible.

I don't think PureSim is including a negative for 'short contract'. Sure, add it to the list.

The two improvements I'd really like to see are 'situational' improvements, so that I can't poach the best young player off of a rebuilding team no matter what decrepit veterans I'm offering [:)], and some form of balancing so that a multi-player deal doesn't automatically 'go through' just because the sum of the player evaluations is larger on one side than the other.

The things I want to see are outlined in the other thread, but I'll repeat here:

A 'core' team - 8 fielders, 4 starting pitchers, 1 reliever, 1 closer.
A '25-man' team, comprising the current 25-man roster.
A 'full' roster, comprising the entire roster, whether its 35, 60, or 200+.

A 'strategy', which teams use to evaluate all player moves; I've outlined in the other thread definitions for 'complacent', 'a top contender', 'close but not quite', 'on the rise', 'rebuilding', and 'dumping', and a progression I'd like to see used to move teams through those stategies over time based on results.

A different 'evaluation function' for players depending on which of the three 'rosters' they are being considered for. E.g., for players who will be in the minors, youth and potential are much more important than current ability; for players in the 'core team', the opposite is true.

A 'summed team evaluation function' which can identify the strength of the 'core team', '25-man team', and 'full roster'.

We would wind up with a matrix of evaluation functions - a 'rebuilding' team might valuate a slightly worse 27-year-old as 'more valuable' than a 'better' 38-year-old; a 'top contender' would look at them the other way around.

Finally, teams would consider the impact to the team of each move by considering all three aspects of the team "before" and "after" the move. I think that might help stop some of the unbalanced multi-player trading. The team could look at that and say, "You know, though getting your three hitters helps my 25-man team, it actually hurts my 'core' team lots more to give up my best pitcher, as none of them start, and the #5 pitcher who promotes into his spot is significantly worse."

Trades which genuinely help both teams would still go through, but it might help to reduce poaching by human players. It would also reduce some of the 'I trade my only quality shortstop away, forcing me to promote a useless minor leaguer into the bigs' deals which you can observe in AI-to-AI trades.