 |
View Old Forum Thread
Old Forum Index » Sim Forums » Baseball Game & Site Questions » A possible help for the bot invasion
Michael1
A possible help for the bot invasion
February 07, 2016 at 03:13PM View BBCode
Require owners to meet a threshold a minimum talent level to win at least 50 games or at least potentially win 50 in the Majors. That can be a average number from all 25 players in the majors or no more than X numbers of C or below on the 25 man. IF a team falls below that level ABE will turn there team name RED in the standings so everyone knows this owner is below standard. Then encourage the league to make trades with this owner or at least make available to him better talent to bolster his average grade. A lot of these bot teams are that way because past owners either couldn't or wouldn't get what was needed to stay competitive.
Frunobulax
I posted this a while ago in support, privately. I also offered help with the coding, but didn't even get a reply on this one.
snipp
----------------------------------------------------------------------
The problem is that it fails horribly at developing players. I have seen 21-year old pitchers promoted and put in the #5 starter spot and end up with 30 ICs. Top talents will sit on the bench because at OS23 they are worse than the 8 best players in absolute value. And, if this goes on for some seasons, the teams are in so bad shape that nobody wants them anymore.
What the bot should do: Predict the player skills at OS30, and rank all players according to these projections. Players over OS30 get their current skills as projection. With a linear weight system they end up with a numerical player value. (A very simple way would be to use 1/2*vel+con+1/4*end for pitchers, CvR+PvR + 1/3(BvL + CvL + Spd + Rng + Arm) for position players, both normalized (so an player with all 80 values will have a total value of 80).
So every player has a numerical value, and for players younger than OS30 the value is a projection at OS30. (That's what I do for my leagues.) This is the absolute value.
A tweak for the bot would be to introduce a mild penalty for players over 30. That way, youngsters are favored if they project to be better than the vets on the team. This is the lineup value. Say one point for every year above OS29.
Now, there would be a very simple way to program a bot. My first assumption that the Bot always goes for best development, because if there is a competing team out there then it will be bought sooner or later.
Then I propose the following:
1. Player ranking is done as described above. (Again, young players are slightly favored.)
Minors:
2. In a CP league, promote position players at OS22 and pitchers at OS23 if their value is above a certain threshold. Say we promote players that project to 76 or better in this system.
3. Put CPs on the best X remaining talents, X depending on the league (5 in system 2, 15 in system 5, 4 in system 6).
Majors:
4. Find the 3 best position players, 2 best starters and 2 best relievers with at least B+ mentoring. Put them in the lineup. If there are not enough B+ mentors, pick the best mentors (starters over relievers).
5. Complete the lineup by adding the best players according to the system described above. The penalty on the older players makes sure that young players are favored, so if there are talented youngsters on the team then they will play.
So the bot will usually create lineups with 3 mentors and a lot of young players. Drafting happens according to the projection.
Trimming the roster:
6. In the majors, the aforementioned mentors are untouchable. From the remaining players, demote the ones with the worst absolute value.
7. In the minors, simply cut the worst players.
Draft
8. Rank all players as described above. Assign a bonus to older players in CP leagues, because they require less CPs.
You'll agree that this is a very simple and robust system. I'd bet that a bot created this way will be on par with most owners who don't know how to rebuild teams :) But then, these teams would be bought, so we don't have to worry about that. And since the focus is rebuilding, it's unlikely that the botted teams would compete - they would just keep rebuilding until the team has enough young talent to find a good human owner.
.f
February 09, 2016 at 10:17PM View BBCode
That's the wrong approach. You can't keep owners from leaving burnt out teams. The problem is that the bot keeps them running into the ground further, so nobody wants to buy a team that was under bot control for a while.I posted this a while ago in support, privately. I also offered help with the coding, but didn't even get a reply on this one.
snipp
----------------------------------------------------------------------
The problem is that it fails horribly at developing players. I have seen 21-year old pitchers promoted and put in the #5 starter spot and end up with 30 ICs. Top talents will sit on the bench because at OS23 they are worse than the 8 best players in absolute value. And, if this goes on for some seasons, the teams are in so bad shape that nobody wants them anymore.
What the bot should do: Predict the player skills at OS30, and rank all players according to these projections. Players over OS30 get their current skills as projection. With a linear weight system they end up with a numerical player value. (A very simple way would be to use 1/2*vel+con+1/4*end for pitchers, CvR+PvR + 1/3(BvL + CvL + Spd + Rng + Arm) for position players, both normalized (so an player with all 80 values will have a total value of 80).
So every player has a numerical value, and for players younger than OS30 the value is a projection at OS30. (That's what I do for my leagues.) This is the absolute value.
A tweak for the bot would be to introduce a mild penalty for players over 30. That way, youngsters are favored if they project to be better than the vets on the team. This is the lineup value. Say one point for every year above OS29.
Now, there would be a very simple way to program a bot. My first assumption that the Bot always goes for best development, because if there is a competing team out there then it will be bought sooner or later.
Then I propose the following:
1. Player ranking is done as described above. (Again, young players are slightly favored.)
Minors:
2. In a CP league, promote position players at OS22 and pitchers at OS23 if their value is above a certain threshold. Say we promote players that project to 76 or better in this system.
3. Put CPs on the best X remaining talents, X depending on the league (5 in system 2, 15 in system 5, 4 in system 6).
Majors:
4. Find the 3 best position players, 2 best starters and 2 best relievers with at least B+ mentoring. Put them in the lineup. If there are not enough B+ mentors, pick the best mentors (starters over relievers).
5. Complete the lineup by adding the best players according to the system described above. The penalty on the older players makes sure that young players are favored, so if there are talented youngsters on the team then they will play.
So the bot will usually create lineups with 3 mentors and a lot of young players. Drafting happens according to the projection.
Trimming the roster:
6. In the majors, the aforementioned mentors are untouchable. From the remaining players, demote the ones with the worst absolute value.
7. In the minors, simply cut the worst players.
Draft
8. Rank all players as described above. Assign a bonus to older players in CP leagues, because they require less CPs.
You'll agree that this is a very simple and robust system. I'd bet that a bot created this way will be on par with most owners who don't know how to rebuild teams :) But then, these teams would be bought, so we don't have to worry about that. And since the focus is rebuilding, it's unlikely that the botted teams would compete - they would just keep rebuilding until the team has enough young talent to find a good human owner.
.f
MNRebel
http://forum.simdynasty.com/viewthread.php?tid=423680
February 09, 2016 at 11:03PM View formatted
You are viewing the raw post code; this allows you to copy a message with BBCode formatting intact.
Fru...to be fair, Chris did acknowledge some of your ideas at the end of this Support thread.
http://forum.simdynasty.com/viewthread.php?tid=423680
paulcaraccio
1) Injuries to SPs, when they heal they need to be put back in the rotation
2) In leagues that play 3 or 6 GPD, team needs to be logged into once every so often during season (3 weeks?) to avoid entering tank-assist mode. (or that dumb rule could just be removed site-wide)
other big problem is the bot is terrible at drafting. thats a stupidly simple fix, just need to change the draft settings.
alternately, fix up the draft settings, have the bot run the draft, and then step aside. Give a few vet owners a couple bucks/free seasons to set the lineup/rotation/CPs on opening day for unowned teams, this will take an experienced owner about 5 minutes per team, but same 2 issues.
(what i would actually do, is hire someone who can give us more than 1 or 2 hours of work per week, start making some fixes/enhancements, and get people interested in buying teams.)
February 10, 2016 at 04:29AM View BBCode
one big problem with the bot is that it runs every day, always shuttling players around. I'd program it to just set it and forget it on opening day. the only problems with that are1) Injuries to SPs, when they heal they need to be put back in the rotation
2) In leagues that play 3 or 6 GPD, team needs to be logged into once every so often during season (3 weeks?) to avoid entering tank-assist mode. (or that dumb rule could just be removed site-wide)
other big problem is the bot is terrible at drafting. thats a stupidly simple fix, just need to change the draft settings.
alternately, fix up the draft settings, have the bot run the draft, and then step aside. Give a few vet owners a couple bucks/free seasons to set the lineup/rotation/CPs on opening day for unowned teams, this will take an experienced owner about 5 minutes per team, but same 2 issues.
(what i would actually do, is hire someone who can give us more than 1 or 2 hours of work per week, start making some fixes/enhancements, and get people interested in buying teams.)
Frunobulax
You have to take care of the rotation. If a mentor gets injured, big problem. And an injured SP will be moved out of the rotation, so he won't pitch a lot and he won't start once he comes back, losing stamina.
The system I proposed is rather robust in that way. Player values in the majors don't change, players in the minors are promoted only at the start of a season. So the Bot will just shuffle guys around if there are injuries. It could even put guys on DL (which may be necessary in max injury leagues, or if a catcher gets injured) and just find a replacement from the waiver wire.
@MNRebel, yes, Chris did. (Thanks for pointing this out, I don't follow the board anymore that closely.) What bugs me - I offered help with the coding, in a way that I wouldn't need access to the source code of SimD. Never got a reply to that one. The window of opportunity is closed now, I did have some spare time around christmas.
.f
February 10, 2016 at 10:07AM View BBCode
Originally posted by paulcaraccio
alternately, fix up the draft settings, have the bot run the draft, and then step aside. Give a few vet owners a couple bucks/free seasons to set the lineup/rotation/CPs on opening day for unowned teams, this will take an experienced owner about 5 minutes per team, but same 2 issues.
You have to take care of the rotation. If a mentor gets injured, big problem. And an injured SP will be moved out of the rotation, so he won't pitch a lot and he won't start once he comes back, losing stamina.
The system I proposed is rather robust in that way. Player values in the majors don't change, players in the minors are promoted only at the start of a season. So the Bot will just shuffle guys around if there are injuries. It could even put guys on DL (which may be necessary in max injury leagues, or if a catcher gets injured) and just find a replacement from the waiver wire.
@MNRebel, yes, Chris did. (Thanks for pointing this out, I don't follow the board anymore that closely.) What bugs me - I offered help with the coding, in a way that I wouldn't need access to the source code of SimD. Never got a reply to that one. The window of opportunity is closed now, I did have some spare time around christmas.
.f
paulcaraccio
despite your assurance that we will all agree your system is simple, it isn't, and your assertion that these teams would all suddenly get bought if only you could code the bot is definitely hilarious, i'll give you that.
i'm just as bewildered as you that your message to admin along the lines of "you suck at coding, let me do it" went unanswered.
people aren't buying as many teams as they used to because they don't wanna play anymore. it starts at the top, Tyson is long gone, Chris doesn't have time, and they won't hire anyone else. They're content to let the site die slowly while collecting a good deal of money for almost no work in the interim. It just doesn't make sense or feel right to spend more time and money at a shop like this. making the bot better wouldn't make much difference.
February 11, 2016 at 11:54PM View BBCode
yeah...i said that dude. you write well enough to be able to read, ha.despite your assurance that we will all agree your system is simple, it isn't, and your assertion that these teams would all suddenly get bought if only you could code the bot is definitely hilarious, i'll give you that.
i'm just as bewildered as you that your message to admin along the lines of "you suck at coding, let me do it" went unanswered.
people aren't buying as many teams as they used to because they don't wanna play anymore. it starts at the top, Tyson is long gone, Chris doesn't have time, and they won't hire anyone else. They're content to let the site die slowly while collecting a good deal of money for almost no work in the interim. It just doesn't make sense or feel right to spend more time and money at a shop like this. making the bot better wouldn't make much difference.
Frunobulax
You're right about one thing: The site is dying, and if Chris doesn't fight it, then the only way to react to this is to leave. Which is what I'm doing, team by team.
[Edited on 2-12-2016 by Frunobulax]
February 12, 2016 at 11:52AM View BBCode
You know, I certainly won't miss you when I leave SimD, which I will rather sooner than later. You know, I was trying to _help_, you [insert expletive here].You're right about one thing: The site is dying, and if Chris doesn't fight it, then the only way to react to this is to leave. Which is what I'm doing, team by team.
[Edited on 2-12-2016 by Frunobulax]
Pages: 1