Improvements for GMs

[URett]

When I play PureSim (and I do almost every day) I act as the GM of my team. I never manage any games and most of the time I don't even watch World Series games. I enjoy building a winning team. Unfortunately this is much too easy.

Here are some of the main problems:

1. Player Salaries:

- Extensions aren't really extensions but renegotiations. Older players will accept extensions at significantly lower salaries, so you can free up cash by extending contracts. In PureSim it makes sense to give your 38 year old starter with 2 years left a 5 year extension, once he is no longer good enough you can just trade him or stick him into the minors and wait for him to retire. The AI doesn't (and shouldn't) do that.

- Teams can go over their spending limit at the end of the season. This might be a bug, but at the end of the season all expiring contracts are removed from the payroll, but if you extend a contract PureSim still thinks that the player has a current salary and only checks if the raise fits under your spending limit. So if you are at the limit and a 5m contract runs out you can resign the player for up to 10m. You will run into trouble down the line, but it can be done. In one association I had half the AI teams were overspending and so weren't able to do anything all year.

- Salary demands drop too early too far during free agency. It seems that after 6 weeks player demand less than half of what they originally wanted. I you are lucky and the free agent period goes until the 6th or 8th round you can pick up very good players at very low salaries.

- I think if inflation is enabled only the available finances increase, but not the salary demands.

As I said before, I think the financial system needs a complete overhaul to make it more realistic and more challenging for the human GM.


2. Trading

- The fact that AI trades are always 1-for-1 makes it difficult for the AI to execute the different strategies (dump salaries, get younger, etc). The human GM has a huge advantage here. It's probably a huge task to program AI multiplayer deals, but it would be a huge step towards making the AI more competetive.

- There are very few players that can't be traded away and very few players that can't be traded for. As a human GM you can change half your team every year just by trading and still improve. Here is how I did it in some associations: 1. Look for a young star with a good contract and open trade talks, 2. chose some more young prospects from that team, 3. put together a package of players with expiring contracts or other players you can afford to lose. Do this several times each season and before long, even as a small market team, you will have a deep team that will run away with its division every year.

- The AI GMs have no concept of salaries and contract lengths'. If there are teams with availabe cash you can dump your declining veterans or overepaid busts without any problem. In real life nobody claims great players like Manny or Griffey Jr. from waivers because they are too expensive. The AI also gives up too much for expiring contracts and too often doesn't even resign those players.


3. Player generation

- I might be wrong here but I think in PureSim there are too many good/very good players but not enough superstars who will dominate year after year. While the AI is paying huge salaries in the first few weeks of the free agent period the human GM can easily wait until later or even until after the free agent period and still pick up similar players at much lower costs.

[waltwa]

i agree strongly particularly with item #2. until there is a complete minor league system trading in this league will never be right. trading is most realistic when you have multiple options when trading such as minor league players, cash and draft choices.

[Amaroq]

Excellent post, my friend. I play the same style - never managing, quick-simming everything, although I do watch the World Series games.

There are some things I might add to that.

1. Player Salaries

I don't know if this is aimed at 'modelling real life', but it feels to me like the player salary demands are a bit too age- and time-of-service dependant and not based enough on 'current ability'. Its often cheaper to sign a top player at age 27 than it is to sign the same player at age 34 as he's on the decline. I look at their 'value' the other way around: the 27 year old has maybe six to eight good seasons left in him, while the 34 year old has at most one or two. This is especially noticeable with extensions.

The salary demands for young players seem off - I think young players should demand MORE for longer contracts, not LESS. The typical thought there should be "I have potential, and I want to maximize my earnings. To keep me off the free agent market, you need to pay me a premium - more than you think I'm worth now, because I'll BE worth more down the line."

Extensions have a different problem, as well. It seems like, if a player has 2 years left on his contract, offering a 3 year extension uses the salary demand for a 3 year contract rather than for a 5 year contract. This is subtle, but one of the things I wanted to do was modify the XML in a way that addressed the previous item, which worked for the amateur draft or free agency, but once you had a young player on the team, you could continually keep him at a 5-year contract by offering 2- or 3- year contract extensions every two to three years.

In addition, I seem to always be able to get my players to agree to extensions. I haven't had a star declare that he's opting for free agency - usually if I offer him a 3-year extension and he turns it down, offering him a 2-year extension or 4-year extension works.

In general, the 'discount for length' feels very overwhelming as a feature of the salary landscape: I find that I can keep myself out of serious financial trouble by having everybody tied up to long-term contracts pretty much all the time. Since the AI teams play with more short-term contracts, they're always in more financial trouble than I am.

2. Trading

Agree with you completely. It feels like the routine for AI trading is, "For each team, each day, throw a random number. If it indicates that the team should make a trade, then - cycle through all players on my team, comparing them to all players on other teams, until you find a trade which works financially, which I would agree to and the other team would agree to. If so, make the trade."

Likewise, human-AI trades are "Do I rate the sum of the incoming players higher than the sum of the outgoing players?"

Both of these are missing an element of 'squad sense'. A team is willing to trade its superstar shortstop despite not having any other big-league-competent shortstops. Or to acquire another good shortstop despite having a superstar, and give away another starter.

I'd like to see the team think of its 'core 14', e.g., the 8 position players, 4 starting pitchers, 1 reliever, and closer as a unit, and then also to have its 'big league 25', e.g., the other 11 who flush out the senior team. See section 4, but teams would regard those differently, and consider the 'opportunity cost' when making trades - so a team which has two very good shortstops would be much more willing to trade one, while a team with no big-league-quality shortstop would be very unwilling to trade their starting shortstop without getting one in return.

I'd love for 'player evaluation' to drop to 0, or even to negative, based on the player's abilities. I personally would regard five never-going-to-make-it players as a negative, not a positive, but I have found that if I'm over 50 players, I can often trade away the players I am going to cut to other teams and have the other team cut the player instead.

Finally, the AI for trading might want to walk through it in a different way, as follows: determine a 'Need', e.g., a hole in the 'core 14', or an underperforming player in the 'core 14', or 'we need young talent', or 'we need to dump salary'. Look around the league for teams which can help you address that need - e.g., teams with two players at your 'hole', or with 'young talent', or with room to take salary. Try building some one-for-one trades with those teams. If a trade would fail for one team but not the other, try adding a minor-league player or two to it, and see if you can juggle it into a trade which 'passes' for both teams. If not, move on to the next plausible one-for-one.

3. Player generation.

It feels to me like the combination of 'the way players are generated' and 'the way players improve over time' tends to make the distribution of values a bell curve, centered around 45 out of 100, and tailing upwards very slightly.

Statistical research of athletic performance shows that athletic performance at the top end of the spectrum (which we're modelling here) is more likely to follow a pyramidal scheme in which a 10% increase in the 'bar' knocks out 90% of the athletes. So, for example, if we had a linear scale from 1-10, we would have 1 player at a 10, 9 players at a 9, 90 players at an 8, 900 players at a 7, etc, finishing up with 900,000,000 players at a 1. Obviously that's not what we have herem and nobody wants a database of 1 billion putative baseball players, most of whom are utterly awful. But we could change the modelling a bit in a way that more accurately reflects the pyramidal structure - and we'd wind up with the same sort of 'failure rate' that real minor-leagues have.



I'd also add the following notes.

4. Off-season algortihm.

I love the way that the AI adjusts its evaluation of players during the season, based on results. If a team is 25 games out of first place as the trade deadline approaches, they consider themselves in a 'rebuilding' phase, and are willing to unload veterans for young talent. Great.

However, all the teams use the same algorithm in the off-season.

I've seen a team come in last, 40+ games out, clearing plenty of salary room by unloading their veterans and acquiring some young talent... only to blow it all by, in free agency the following year, picking up two 40+ year-olds who have no chance of helping the squad for outrageously high prices.

What I'd like to see is the different AI evaluations used throughout the season and off-season, to give them more of a sense of 'strategy' than of 'tactics', and to give them a sense of continuity.

a. Complacent - a team in first with a large lead. In the offseason, the only route to this logic would be a runaway division title followed by easy series en route to a World Series title. These teams make no mid-season changes in terms of minor-league callups or starting youngsters. They are, for the most part, unwilling to make trades, although they will trade aging players outside of the 'core 14' for youth players if they feel there is a good young player at the same position in the minors so that it doesn't hurt the 'big league 25'. They're willing to accept an overpaid veteran with 1 year left on his contract if he improves the 'big league 25', but wouldn't accept an overpaid veteran with a long-term contract. In free agency, they are most likely to be very conservative, unwilling to overpay, as they think the side is 'perfect' as it is; they would only seek to fill any 'holes' in the 'core 14' caused by retirement, aging, or free agency. They are likely to make extensions to everybody in their 'core 14', and to all of their young, talented players. In the draft, they are looking for potential over talent, but obviously both is best. They are willing to take a flyer on a long-term, all-potential project player.

b. A top contender - a team in first, the defending divisions champions, or any team in a close fight for the wildcard. These teams are very unlikely to try sending 'underperforming' players to the minors, and unlikely to give youngsters a break. They are willing to trade young talent for improvements to the 'core 14' as described earlier, or trade marginal players for improvements to the 'big league 25'. They're willing to accept veterans with 1 year left on their contract if they improves the 'big league 25', and are really the only team which would accept an overpaid veteran with a long-term contract. They are the type of team that will package multiple minor-leaguers for one player to help their current lineup. In free agency, they are most likely to be the most aggressive, considering only the 'does it help the core 14' aspect, and ignoring finances. They are likely to make extensions to veteran players, trying to get 'one more year' out of the stars who helped them so much last year. In the draft, they are willing to take older players who can help immediately, and players who are closer to reaching the majors over long-term 'potential' projects.

c. Close but not quite - a team which is 'one or two pieces away', usually a team which is within shouting distance of the playoffs but not getting there at the moment, or a team above .500 the previous year. These teams prefer not to send 'underperforming' players to the minors, and if they bring a youngster up, he's likely to sit on the bench. They are willing to make radical moves to improve the 'core 14', or to upgrade the 'big league 25'. They're willing to take players in the last year of their contract, but not overpaid veterans with long-term contracts remaining. In free agency, they can't be the most aggressive, since they probably feel like they need 2 or more players to reach the tops; they don't want to blow it all on a single superstar. Still, they'll be fairly aggressive, acquiring any player who improves the 'core 14' or the 'big league 25' without regards to age. They make extensions to players in the core 14, and to their young talent - 'big league 25' players under 33, and minor leaguers with real potential. In the draft, they are willing to take older players who can help immediately, and players who are closer to reaching the majors over long-term 'potential' projects.

d. On the rise - a team which is trying to build for 2-3 years down the line, usually moved into from e. after an improving season, teams not too far under .500 and trending upwards. These teams will send 'underperforming' veterans to the minors, and are willing to play younger players over better veterans in the theory that the younger player will improve. They are willing to trade veteran players for younger players, but are always looking to improve the 'core 14'. They're highly likely to trade veterans from the 'big league 25' if it improves the core, or to trade players over 35 out of the 'core 14' looking for younger players. They are unwilling to accept players in the last year of their contract, and in fact are looking to trade their players whom they cannot/will not agree extensions with. They will hoard talented minor-leaguers in the 23-29 age range, but might be willing to trade talented high-potential players too young (18-21) to help in their 2-3 year plan for players who will help. In free agency, they are likely to be the most aggressive about players in that 23-29 range who fit their 'core 14', and eschewing all players over 34. Their extension policy is primarily aimed at renewing the most talented players under the age of 34, desperately trying to prevent anybody from leaving via free agency before they can become a top team. In the draft, they are looking for a blend of current talent with potential, not willing to invest in an 18-year-old who won't be ready for five or six years, but not discounting potential either.

e. Rebuilding - A team which is (or finished) under .500 and a long way out, with no hope of victory, but an average age which doesn't qualify them for f. These teams are aggressively looking to make major improvements. They're defeinitely willing to send 'underperforming' veterans to the minors, and are willing to play younger players over better veterans to improve the youngsters. They are definitely willing to trade veteran players for younger players, but are looking for young players who will improve their 'big league 25'. They're willing to trade out of the 'core 14' if its an older player and they're acquiring two or more quality young players in return. They are unwilling to accept players in the last year of their contract, and in fact are looking to trade their players whom they cannot/will not agree extensions with. They're looking to hoard all of their talent underneath the age of 28, and it should be very hard to prise a talented youngster out of their hands. In free agency, they aren't likely to be very aggressive: the team has too many holes to justify making a large expenditure on a single player; they're looking for anybody under 30 who can help improve the 'big league 25', and are willing to pick up other team's minor-league rejects on a flyer. Their extension policy is to renew players under the age of 30, but to let older players leave via free agency. In the draft, they are looking for potential over talent, but obviously both is best.

f. Dumping - usually a team which started the season in a., b., or c. mode, but is falling well short of expectations. They are under .500, a long way out of contact with the division leaders, and feel that their average age is too high or their payroll is too close to the limit. A team in this stage at season's end will remain in this stage if their average age is too high. These teams have a very short hook for their 'underperforming' veterans, and are absolutely willing to bring up any promising minor leaguers, and even give them playing time. They are willing to dump players over 28 from their 'core 14' for youth and potential, and are looking to make 2-for-1 trades or better, as they feel that they need lots of talent. They factor 'cost' in very heavily if they are anywhere near the edge of their budget. They may be willing to trade veterans with long contracts for veterans in their last year of contract just to free up salary, and they . In free agency, they avoid all players over the age of 30, but are willing to sign younger players who would improve their 'big league 25'. Their extension policy is letting all players over 30 expire, and being sure to aggressively renew the contracts of all of their youth and potential players. In the draft, they are willing to take a flyer on a long-term, all-potential project player, and are definitely looking for potential over talent.

Teams would occupy one of these niches at all times, and their decisions made in the off-season would be governed by their current 'state of mind', and it should be clear from their actions which state they're in.

This would help prevent teams which finished 40 games out making an expensive veteran push to climb to 25 games out.

5. Anti-hoarding

One 'anti-hoarding' measure would be for the AI to be less likely to trade with successful teams. This isn't entirely 'historically correct', but if one of the problems URett and I are experiencing is a 'runaway leader' effect which lets us get to a point that we can always cherry-pick in trading, a solution to that is to make teams more 'aggressive' in their demands for trades with teams with a history of success, and less 'aggressive' in their demands during trades with teams which don't have a history of success. This is probably especially relevant within the division - teams should be very unlikely to make a trade to their defending division champ, and more likely to make trades into the other league.

Personally, I'd love to get the entire, correct MLB player transaction and movement rules, including a Rule 5 draft and other intricacies: I think they're well designed for encouraging that players rise to the big leagues, and that there is competitive balance between the teams, while PureSim is subject to player hoarding.

In particular, the rules that would trounce my typical PureSim player-hoarding strategy are

The Rule 5 draft. The best players in my minor-league system are definitely better than some of the *starters* on the worst teams in my association. These players are being 'held down' artificially, and those teams need the talent. If they had the Rule 5 draft at their disposal, they could agree to take these players, contracting to keep them on their 25-man roster for the entire season, and thus improving their team as well as the players' careers.

The 6-year minor-league free agency. Again, the best players in my minor-league system aren't getting big league playing time. If they could leave voluntarily after their 6th season, and test the waters via free agency, they might well find a place in somebody's starting lineup. (This would be simple to do artificially with the current PS engine: players who aren't in the majors don't agree to a contract extension beyond their sixth year.)

Both of those would require the '40 man roster' concept to implement the exact MLB rules.

[DonBraswell]

Well it looks like you two have covered just about every thing. [:)]
One thing I would like to see fixed is AI teams sending talented vets to the minors and bringing up poor minor leaguers to play in the Championship Games! [X(] Sending last years allstars to the minors and bringing up minor leaguers for the next season. [&:] And when I trade a Star Pitcher or Fielder (in his prime 27 to 30) to a AI team and they send him to the minors or severly limit his playing time. Why ask for him or trade for him? [:(] I have traded star players, just to try to even up the talent in the league, only to see them spend the rest of their carrer in the minors. [:@] I have won all but 8 seasons, playing from 1949 thru 1991. I like to win, but not so easily. I want to have to work for each Championship.
Don Braswell

[Steely Glint]

IMO this game needs more in-game managerial strategy options before it needs any more GM options. As it stands it is almost a GM-only game.

[DonBraswell]

IMO this game needs more in-game managerial strategy options before it needs any more GM options. As it stands it is almost a GM-only game.

I totally disagree with your statement. The game will change as Shaun sees fit. But, this game started as GM only. Most, if not all of the changes over the past two years has been in game management or multiplayer. I'm looking for AI improvements.

Don Braswell

[Steely Glint]

ORIGINAL: DonBraswell
I totally disagree with your statement. The game will change as Shaun sees fit. But, this game started as GM only.

And I disagree totally with yours. Almost every change I have seen in PureSim has been made at the direct request of and specifically for the GM-only crowd. On the other hand, the managerial aspects of PureSim are scarce and have been relatively neglected in favor of the GM aspects. The managerial strategy choices need to be increased dramatically.

PureSim has to date been pretty much a GM-only game. Until that changes, I expect it to remain something a niche product. OOTP and Diamond Mind are both more popular than PureSim, and a big reason for that is because they each offer both a good GM experience AND a good manager experience.

I am by no means saying Shaun should make PureSim a managerial-oriented game. What I am saying is that before another thousand GM-only features are added that Shaun needs to sit down and flesh out the managerial strategy options sufficiently to make PureSim a game with a good managerial component to it, and only then go back and give the GM-only crowd the next thousand items that they want.

For a long time now the GM tail has been wagging the PureSim baseball game dog, and that needs to change - at least for a while - so that the game can be multi-dimensional.

[Amaroq]

Can you guys keep the debate about 'whether this needs to be done' to a different thread, and keep this one on-topic to its thread title and initial post?

Steely Glint, I appreciate where you're coming from - I just don't think this is the thread for that discussion. Thanks!

(In fact, I'd love it if you'd start a thread "Improvements for Managers", and run with it - exactly what you'd like, in the same way that URett has gone here. I felt like there was such a thread, maybe started by KG Erwin? but I was unable to find it - it might have been on the old forum. If I don't see such a thread in the next day or two, I'll start one myself.)

[ravinhood]

After reading this thread this is why I enjoy OOTPBB so much. Many of the issues I've read here are not in OOTPBB and everything from trading to winning the world series is a challenge and the rpg manager mode is unmatched by any other, to be able to get fired and have to start all over again with another crummy team gives OOTPBB the grand prize for best strat-sim baseball game on the market.

Now back to yalls arguing over what improvements this game needs most, GM or managerial hehe

[Amaroq]

It is? Not last time I played OOTP... [;)]

... but I should really give you the same answer I gave Steely Glint - mind keeping this thread 'on topic'?

[Steely Glint]

Ravinhood, sure, OOTP is a good game, but it has some AI issues and some bugs that are old enough to vote by now. PureSim, while not as mature or as popular, can be customized by the user and has a superior interface. With the Sienna skin installed Puresim is the best looking baseball strategy game out there and the easiest to use and master.

[Steely Glint]

And to go back on topic, IMO more GM enhancements are a good thing once Shaun gets the managerial options up to speed.

[lynchjm24]

One easy fix that would be a huge improvement.

Stop having players that are unsigned for 2 or three years end up with huge potential ratings.

Players will languish somewhere in the minors until they are 21, then in free agency they will show up at age 24 after being free for 2 seasons with big potential ratings. Then you just snap them up at low salaries for 5 years and you have a never ending supply of good to great hitters that are always in their primes. When they hit 30 or 31 you just move on to the next wave.

So in short, if you are an unsigned free agent your potential rating cannot go up - that might work just fine.


ONE OTHER IMPROVEMENT

Limit the number of offensive catchers. There are more catchers that can hit then any other position on the field. I've usually got 2 extra catchers in the minors who could put up above average offensive numbers in the majors.

[Steely Glint]

I am shocked, shocked that Lynchjm24 showed up to make even more GM-only suggestions here while disparaging managerial improvements in the Improvements for Managers thread.

PureSim needs to get out of the iron grip of the "PureSim is a GM-only game!" crowd.

[Amaroq]

2. Trading

A different idea for a trading AI occurred to me - for those of you who are keeping this thread on-topic [;)], anybody care to weigh in on which algorithm they would prefer, this or the one I posted earlier?

Most multi-player leagues run by having teams post players who are "on the trading block", and post what their "needs" are. FM'05 runs by having teams post players "on the transfer list", or "available for loan", which winds up filling the same role.

What if the AI for trading were to function by maintaining a list of team's "available for trade" players and their "publicly posted needs"? If a team decides to 'make a trade', it runs through the 'available players' comparing them to its players at the 'need' positions, and considering whether its willing to make a trade.

Let's consider four teams.

The Albatross have two closers, one very-good and one excellent. They badly need a shortstop (best is minor-league).
The Bears have five good outfielders but no real strengths. Their catchers are poor, and they need another starting pitcher.
The Canaries have three very-good outfielders, five good starting pitchers, and one excellent minor-league starting pitcher.
The Devils have two excellent shortstops, a good second baseman, and badly need a closer. They currently have one SS playing 2B.
The Eagles have an excellent outfielder, their #2 outfielder is very good, and their #3 outfielder is poor. They have two good catchers.

- The Albatross are the first team to post needs and trade players.
The needs list is empty, so they can't make offers.
They post "Need = SS, On the block = Very good Closer"

- A week into the season, the Bears are the next team to post a need.
They don't need a closer, so they post a new need.
They post "Needs = P, C, On the block = Good OF"

- A couple days later, the Canaries think about it.
They don't have need a closer or an outfielder (the two things on the list at the moment)
They don't think they need to post a need at the moment. They're happy.

- The Devils are next up.
They look at the Albatross' needs and find a match - they have an excellent SS, and need a closer. However, they don't make an offer, because they feel like they'd be getting ripped off (giving up 'excellent' SS for a 'very good' closer).
They post "Need = Closer, On the block = Excellent shortstop."

- The Eagles are next to consider trades (and this might be three weeks into the year)
They look at the Bears' post, and find a match: the Bears are offering an outfielder, and need a catcher.
A 'good' C for a 'good' OF looks like a fine deal for the Eagles - they would start the latter, and aren't starting the former. They make the offer.
The Bears consider it, and have the same evaluation - we'd start the one, we wouldn't start the other. The trade is made.
The Eagles consider needs, and don't post a new need.
The Bears reconsider their needs. They still have four good outfielders, and they need a pitcher. They change their posting to "Needs = P, On the block = Good OF"

(The list is now: Albatross need SS, offer very good closer; Bears need P, offer good OF; Devils need closer, offer excellent SS)

- At this point, one of the Canaries' outfielders suffers a season-ending injury.
They get a free pass to 'consider trades' as part of juggling their lineup to deal with the injury.
They need an OF, and they have six starting pitchers. They Bears need a pitcher, and are offering a good OF.
The Canaries consider the situation, and decide to offer a good P for the good OF (and promote their excellent pitching prospect to the bigs).
The Bears consider the offer (good OF for good P) and decide that they come out ahead. The trade is made.
The Canaries consider needs, and don't post a new need.
The Bears reconsider their needs. They don't have any needs, so they change their posting to 'nothing'

- The Albatross then consider their needs.
They spy the 'Excellent SS' the Devils are offering, which matches their needs.
They look at their closers (the Devils needs) and see that they are over-stocked.
Comparing the drop-off (Excellent closer to Very-good) for the gain (Poor SS to Excellent) they decide to offer the trade.
The Devils look at the opportunity cost (Excellent SS to Good 2B) compared to the gain (Very Poor closer to Excellent) and decide to accept.
Both teams look at their remaining needs, and change their posting to 'nothing'

.....

We've saved the 'cycle through all combinations' things, and the 'cycle through all possible combinations between these two teams' part of the first algorithm. We've also come up with a slower-moving way that the human player can 'offer to teams' (the "I get an instant response from every team" model feels a bit artificial) and a way to specify what my 'needs' are to the AI, greatly increasing the likelihood that I will get a trade offer which is meaningful.

Combine that with "strategy" ideas in my earlier post (section 4) and we would have an AI making the right tactical trades for the right strategic reasons - which would I think be a first in the genre!

[lynchjm24]

It's funny, I was going to post that I do think Shaun should upgrade the in game management first. I didn't bother because I didn't have anything else to add.

Your suggestions in the other thread just aren't ever going to work - there is no way that a text sim can successfully pull off that level of detail. There are plenty of things that can be added.

You'll also note that my two huge suggestions for the GM experience are both simple and don't exactly hurt someone who plays out their games, unless they feel like having a never ending supply of 25-30 year olds at below market value and want to dominate even further.

[URett]

Amaroq, I like your idea, but I think it should work slightly differently.

Instead of posting a list of available players the AI should make up a list of players who will not be traded. That list should include current superstars in their prime, top prospects and for small market teams good players with long-term cheap contracts. If another GM decides he really really wants one of those players he would have to overpay massively. This should make it harder for the human GM to just pick out the best players and then put together a package of several average players that the AI will accept.

All other players should be available, but trades should only be made if they benefit both teams (seems obvious, but I'm not sure it happens that way for AI-AI trades, and it certainly doesn't apply for AI-Human trades).

As I said before, AI-AI multiplayer trades would be a huge step towards making the system work, but it's probably extremely hard to code properly.

[Amaroq]

Absolutely, to your idea of 'have a list of players who will not be traded'. I definitely have between 10 to 15 players whom I consider the 'core' of my team whom I am by default unwilling to trade; if I'm taking over a poor team, that number might be down to 4 to 6 players. I'll also pick some prospects who are my 'future', that I'm unwilling to trade.

Brian Sabean (the Giants manager) thinks in the same terms, I think, judging from the moves he's made - he's refused to make some trades in the past three years because the other team was asking for prospects that Sabean had on his 'no-trade' list.

However, you'd want to be flexible with it, as I noted in my trading example - playing by a simplistic mindset, the Albatross might have had their 'Excellent' closer on a 'no-trade' list, and the Devils might have put both of their 'Excellent' shortstops on a 'no-trade' list, which would have stymied that final block-buster deal, which seems to have clearly helped both sides.

[DonBraswell]

I hope some thing useful comes from all of these post. We all know Shaun's time constraints. PureSim 2005 will be a success and PureSim 2006 will be even better.
Thanks Shaun,
Don Braswell