Matrix Games Forums

..

Just purchased PureSim. Was enjoying it a bit. Getting in to it. Nice look, good interface.

Then .. just bizarre .. broken ... things.

Tell me, honestly, if this is the way the game is. I don't want to invest anymore time going further in to the program if this is the current state so you can be blunt. I paid the money so nobody is going to lose any jack or anything. I just don't know what to think.

Started by using the 2005 "in-game" option. No particular psa, just standard in game with the Lehman built-in.

All sorts of bizarre .. just wrong .. stuff happening.

Even with things set to average in tendencies, pitchers going 1.2 and getting pulled after giving up four runs. Then, a few days later, a pitcher going the full nine and giving up 13 runs. ? WHOAH !

Show-stopper there.

Except then I run in to .. I can trade Bob Wickman and Jody Gerut for Manny Ramirez.

WHOAH

Then ... to top it off .. I do trade Jody Gerut and Scott Sauerback on April 20th and on May 13th of a quick simmed game .. THEY BOTH SHOW UP IN THE BOX SCORE FOR MY TEAM. They had been traded for three weeks. I triple checked it. And checked it again.

So, all a text sim really has is the integrity of the statistical outcome. Anomalies do happen. It's baseball. But, things like these start happening and everything that the game performs becomes .. a joke, kinda. Not valid. Void.

And that's all a text sim is is the integrity so .. I'm new here and you can tear in to me if you want but .. I'll take the screenshots and show them to you. I'm not trying to start anything. I'm just asking .. Are these known issues, is this just the way PureSim is?

In this day and age there's so many things to occupy your time, I'll just eat the price and move on. I just figured I'd ask before I put it away.

JB

Even with things set to average in tendencies, pitchers going 1.2 and getting pulled after giving up four runs. Then, a few days later, a pitcher going the full nine and giving up 13 runs. ? WHOAH !

 
Ever hear of Willie Randolph?
 
 
I have played PureSim since 99 or 00. Many many many incarnations ago, like when there were no playoffs. I have never had a head scratching moment or WHOAH moments. I dont play with real players, so maybe different thing happen.
 
Trade AI in any game is a challenge. If theres any holes in it a human can fool the AI 9 out of 10 times. Feel free to create an AI who knows who Manny Ramirez is.
 
PS has a customizable XML, which needs no adjustments right out of the box.
 
If you dont like the game you dont like it. But it works.

ORIGINAL: JB44


Even with things set to average in tendencies, pitchers going 1.2 and getting pulled after giving up four runs. Then, a few days later, a pitcher going the full nine and giving up 13 runs. ? WHOAH !

Show-stopper there.

Except then I run in to .. I can trade Bob Wickman and Jody Gerut for Manny Ramirez.

WHOAH

Then ... to top it off .. I do trade Jody Gerut and Scott Sauerback on April 20th and on May 13th of a quick simmed game .. THEY BOTH SHOW UP IN THE BOX SCORE FOR MY TEAM. They had been traded for three weeks. I triple checked it. And checked it again.

So, all a text sim really has is the integrity of the statistical outcome. Anomalies do happen. It's baseball. But, things like these start happening and everything that the game performs becomes .. a joke, kinda. Not valid. Void.

And that's all a text sim is is the integrity so .. I'm new here and you can tear in to me if you want but .. I'll take the screenshots and show them to you. I'm not trying to start anything. I'm just asking .. Are these known issues, is this just the way PureSim is?

In this day and age there's so many things to occupy your time, I'll just eat the price and move on. I just figured I'd ask before I put it away.

JB

With regard to the handling of pitchers -- you haven't played enough games. I've seen pitchers removed after 1 inning and I've seen some pitchers go 10 or more innings before. Lots of factors in work here (other pitchers tired?). Anyway, PS is handles pitchers fairly well right out of the box. However, you can tweak the XML file and make them even better (in my estimation).

But -- you have to play more than a few games!

Your trade problem is a "known bug" that has been very hard to replicate or track down. Personally, I thought it was squashed a while back. It is an extremely rare occurance.

The game does not preform as a joke. There are too many people at this forum who know what they are doing and we know that PureSim 2007 is one great BB text sim. Some of us have spent a lot of time with it the past 3 years or more, and we wouldn't waste that time if was a joke.

BTW, I've been playing BB sims since 1963 so I think I know what I'm talking about.

Hey, JB, I've been playing PS'07 since the beta period, and I haven't had the problem(s) you're describing.

Which version are you playing? I would imagine, if you just purchased it, that you're using PureSim 2007 v1.10?

ORIGINAL: JB44
Even with things set to average in tendencies, pitchers going 1.2 and getting pulled after giving up four runs. Then, a few days later, a pitcher going the full nine and giving up 13 runs.

I'd want a bit more information about this before deciding if I was unhappy with it.

A pitcher I didn't trust, who'd given up 4 runs and loaded the bases in the second? If I had a rested healthy bullpen, I might well yank him for my long reliever; if its a guy I trusted, the bases were empty, or I had a tired bullpen, I might well leave him in.

Now, the guy going nine and giving up 13 runs wouldn't happen if I'd managed it myself. That's a bit odd, but again, what was the state of the bullpen, and what was the score of the game? The AI tends to leave the starter *IN* if they have a big lead, so as not to consume the bullpen - if the score was 20-13, that choice wouldn't surprise me. Also, were the 13 runs scattered around, or in the ninth inning, or in the first?

At any rate, if you can give more information, Shaun's usually pretty good about tracking down bugs; its a rewarding feeling when you see him crush one you reported in a later beta patch.

Except then I run in to .. I can trade Bob Wickman and Jody Gerut for Manny Ramirez.

'Name value' is meaningless to an AI: its looking at its evalution of the three players, and deciding that the two together are better than the one *for the teams current needs*. I can't tell you if that's 'right' or 'wrong' without seeing the stat lines for all three players, e.g., what their Contact, Power, Eye, and fielding stats are, or what the pitchers Stuff, Velocity, and Control is.

I do think that the AI has trouble in N-for-1 trades, as it seems to amount to reducing the players to a numeric value, and summing the numeric values to see if they're equivalent. This means you can *always* trade, say, 8 AAA players for 4 halfway decent major leaguers, 4 decent major leaguers for 2 good major leaguers, and 2 good major leaguers for 1 superstar. It also tends to overvalue relief pitching - two relievers might well tempt it away from a star hitter.

The AI is very good at 1-for-1 trades, however.

Sitting Duck has been working on improvements to the trade AI, which can be made via the .XML file; see his thread down on page 2: Success: remaking the Trade AI Primarily, this addresses the overvaluing relief pitching; I'm not sure if its helped or hurt in the N-for-1 trades category.

Then ... to top it off .. I do trade Jody Gerut and Scott Sauerback on April 20th and on May 13th of a quick simmed game .. THEY BOTH SHOW UP IN THE BOX SCORE FOR MY TEAM. They had been traded for three weeks. I triple checked it. And checked it again.

I've never seen that one; I know Shaun would love to track it down. You might be able to get ahold of him - email shaun at puresim dot com, and offer to send him your saved game. He might be able to find it with the debugger.

So, all a text sim really has is the integrity of the statistical outcome. Anomalies do happen. It's baseball. But, things like these start happening and everything that the game performs becomes .. a joke, kinda. Not valid. Void.

And that's all a text sim is is the integrity so .. I'm new here and you can tear in to me if you want but .. I'll take the screenshots and show them to you. I'm not trying to start anything. I'm just asking .. Are these known issues, is this just the way PureSim is?

I definitely agree with those statements: statistical integrity is key to the experience, and I hope you'll find that these are anomalies, not 'just the way PureSim is'.

I've had few enough issues with it that I'm still happily playing the association I started during the pre-release beta program.. there are some issues, but on the whole I've been very happy with it.

I hope you'll give it a bit more of a chance, and, even if you do decide to shelve it, will send Shaun the info he needs to help track those down.

I have found 1 for 1 trades are more than fair.  It does need some work after that. 

That trade evaluation logic that compares incremental values seems to be popular in these games, and just doesn't work well with unbalanced numbers of players.

The AI needs to realize that the vast number of players on the market have 0 incremental value to the CPU team. If a guy isn't going to make the starting lineup/rotation, or play an important role in the bullpen, he shouldn't have any trade value to the CPU. So the AI needs to measure the incremental upgrade a player represents, and stop giving credit for throw-in players. None of us would accept trades from the CPU that are constructed that way.

As for prospects, a similar question should be asked - does this guy go to the top of my propsect charts for that position, and is he likely to start for my major league team in the near future.

That's only half of the issue, BleacherBum.

If the AI assumes that the vast number of players have 0 value, then it is too easy to 'steal' them from the AI - the opposite of the current problem.

I've posted at length, previously, about what I think an AI needs to do to do the job 'right', but in general you want to look at several factors:
1. 'Absolute' value of the player
2. 'Replacement' value - the value of the player replaced. E.g., if I trade away my best SS, this value would be the difference between that SS and my second-best SS, who would be my new starter
3. 'Cost' of the player - $$$ and roster space
4. 'Long-term' value

Compare and contrast. [:)]

If the AI assumes that the vast number of players have 0 value, then it is too easy to 'steal' them from the AI - the opposite of the current problem.

Yes, definitely don't want that. I was focused on the other side of the ledger because that's what seems to be the problem with the many-for-1 trades. It seems the CPU is already decent at putting value on its own players, so it can probably stick with that. So again, the typical player on the market has 0 incremental value to the CPU, but an equivalent player currently on the CPU roster is not replaceable for "zero".

If the Trade AI at least could do a before an after roster evaluation, some of the bad trades would be eliminated. Looking at simple stuff like does one of these multi-million $ players I just picked up end up sitting on my bench, or worse getting demoted to my minors because I already have 2 1st basemen just as good?

ORIGINAL: BleacherBum
Yes, definitely don't want that. I was focused on the other side of the ledger because that's what seems to be the problem with the many-for-1 trades. It seems the CPU is already decent at putting value on its own players, so it can probably stick with that. So again, the typical player on the market has 0 incremental value to the CPU, but an equivalent player currently on the CPU roster is not replaceable for "zero".

If the Trade AI at least could do a before an after roster evaluation, some of the bad trades would be eliminated. Looking at simple stuff like does one of these multi-million $ players I just picked up end up sitting on my bench, or worse getting demoted to my minors because I already have 2 1st basemen just as good?

Yeah, now you're talking. I think both of those would add a lot to the trade AI. Your first paragraph, giving a different 'in' versus 'out' value for most players makes good sense - we might almost want it to be contextual: the AI might be willing to trade AAA player for AAA player, for example, but wouldn't be willing to trade a major-leaguer for a AAA player (normally).

The second paragraph would also go a long way, I think, to figuring out what the actual "delta" caused by the trade is.

Interesting thread. Trade AI is a real kicker, especially when folks start projecting their opionions of real-life players.

This area has seen some improvements lately (with the dialing back of the over-valued missle relievers)

I actually do re-evaluate the AI's per-team valuation of players after each deal, but as with any AI this is an area that could do with continued improvement.

Usually if I wait around long enough Amoraq comes up with a good algorithm