AI Minor League Evaluation Needs MAJOR Work

majesty95 · 01-31-2019, 02:45 PM

In our online league, CPU team Toronto is completely butchering their minor league system. Their GCL team (2018 start) has 35 players and 32 are pitchers. Their DE team has only 15 players and they are ALL pitchers.

We have the upper level minors (A-AAA) set to a max of 28 players and the rookie leagues set to 35. We also have incomplete minor leagues off. Their AA team has 8 OF while rookie has none. Several teams have 3 catchers and 3 1B while rookie has none.

This evaluation system needs some serious work.

BMW · 01-31-2019, 03:15 PM

Quote:

Originally Posted by majesty95

In our online league, CPU team Toronto is completely butchering their minor league system. Their GCL team (2018 start) has 35 players and 32 are pitchers. Their DE team has only 15 players and they are ALL pitchers.

We have the upper level minors (A-AAA) set to a max of 28 players and the rookie leagues set to 35. We also have incomplete minor leagues off. Their AA team has 8 OF while rookie has none. Several teams have 3 catchers and 3 1B while rookie has none.

This evaluation system needs some serious work.

This probably won't solve all the problems, but there is a setting in League Settings > Rules.

You can set Roster AI with incomplete minors to:

Match players to levels based on skill (that's probably what your league has)
Fill teams top to bottom first (this tries to make complete teams)

The second option will make a difference, although it won't help if, say, the entire organization has 2 catchers across 5 minor league teams.

NoOne · 01-31-2019, 03:17 PM

Edit -- wrote this while other post was posted too... so, may be redundant in some spots?

fwiw, his idea is FAR easier to do

, but if you want mil rules to work properly on their own, read on....

original post below----

reduce rules and restrictions or massively increase your # of draft rounds or international discoveries or iafa etc if you want to keep existing rules and make them work no matter what...

--

personal tip on roster limits:

don't set any # limit at rookie... even "35". i used to be adamantly against roster # limits at any level, but i've been working with them lately and can avoid the pitfalls with a few undertandings.

AAA -- use more than 28.. AAA must contain mlb depth as well as any additional prospects you care about on top of a functioning AAA team -- since many of those additional players likely won't be there next year, you have to cover your bases, figuratively and literally lol.... i use 34, but 32-34 is probably sufficient, if you use one... no roster restriction and just an age limit can do wonders too.

A & AA -- a couple extra is fine. this is mostly for asthetics on #'s. i use 28, but .. all sorts will work to the point you don't see 8 1b or 32 RP etc...

S-A -- bit on fence, but i typically set on lower than A/AA, but that's mostly how my age rules jive and how many players are typically available to play in SA each year etc etc. keep it similar to A/AA

R -no roster limits.. even 35 is a problem.

if rookies isn't getting enough, this is based on math you can predict. if you use age limits intead of service time at Rookie and Short A, it's a bit messier and harder to predict. in those cases you need more players per year created to make up for the greater disparity per team caused by age-based rules at rookie.

i'm referencing service time limits because it's cleaner math. you can translate to age-based rules and add more than what is needed below for similar contexts based on age, as i said.

# of players created per year / (# of years they can play at rookie * # of teams). without parnthesese, that's simply # of players / year and then that quotient further divided by number of teams.

does it come out to anything near ~30-35? if not, you are starving those rookie leagues, and that's why some teams are all RP.

(e.g. if "21" for rookie, that's ~3 years for high schoolers and less than 1 on average forcollege kids... that's ~1/2 the draft? if that? averages out to maybe 1.5 playeable years on average at rookie? if only 1 rookie team you need ~25-30 rounds at least. 2 rookie teams, and it's 50-60 rounds. you can see why you need "more.")
-----

if you use MiL rules, you better work them out and ensure they work long-term. doing the math ahead of time helps. the further out you go from rookie, the murkier it gets to predict # of players per year.

i have a post somewhere that maps out how you can calculate this stuff step by step, so i won't re-write the entirity here... at that time i probably didnt'like roster limits, go ahead and ignore that portion

they can work, but again, can cause more problems then they solve if not implemented with some sort of over-arching MiL plan.

strongly recommend using service time at rookie, if any restrictions, and age at A-AAA, if anything. s-a? whateer works and doesn't cause oddities. some differentiation between a/a+, obviously, but failed attempt to keep it simple.

-----

if their season hasn't started and the draft hasn't taken place, don't assess the current state of rookie or short A seasons, yet.

having too many duplicates of 1 position can happen, but some limited roster rules at A-AAA can reduce that. nothing you can do about rookie league without causing problems that are far worse than the asthetics of some redundancies on that roster.

1st base in minors is .. maleable, to say the least. lots of players that are future 1B, if anything, are listed as something else, amongst other reasons. take a peak at their actual defensive abilities... even re-position them in your head as you go... it's probably not as bad as it looks once they have the proper label. the experience bars in their profile can give a clue too as you scroll through... if it varies from "position" then you know they really aren't that position based on use in the minors.

in the minors, what they are labeled as is often not what they do.

8 OF isn't too bad on a mil team. bet that happens in RL too. not that i'd use that as a bar for ootp mil in all cases, but it isn't too bad relative to RL in this case.

fwiw, i have a MiL system of rules that functions and never displays the things you mention and allows for easy movement of younger/better prospects that develop faster, even in the AI systems. what i mention works, or can work if done in a way that makes sense. ther are far more vairations that cannot work than variations that can work. some aspects of how the game works may not allow for any one persons exact perception or wishes of how it should work, either.

koohead · 01-31-2019, 06:56 PM

I've seen this concern raised on multiple versions over the years but luckily have personally never experienced it. I do set roster limits, as well as service time limits. I wonder if there is enough available players in the FA pool so that teams have enough players to sign? there has to be some factor that is causing the AI to make these types of odd personnel decisions. My FA pool is always in the thousands...I think it's sitting at 1200 or something right now a month before Spring Training. so even mid season I have hundreds of people available. having enough of a pool of varying ages and talent levels (and perhaps generating enough rounds for your draft pool) may give the AI enough players to pull from so that it doesn't have to fill out the roster with anything it can get its hands on.

majesty95 · 01-31-2019, 11:54 PM

Thanks for the post NoOne. Sucks you have to be that intricate about it lol. This is the first season with a 1995 historical (I think I inadvertently said it was a 2018 post initially).

Are there enough position players overall? I dont know. I'm rolling with what the game gave us. I will say, I have use incomplete minors off which removes the ability to set fill rosters top to bottom.

Either way, the AI could do a WAY better job than this. I'll try setting age limits and see what happens

Mr. Marlin · 02-01-2019, 07:18 AM

I have seen a several annoying problem with minor league rosters. Even with age and service limits it's not uncommon to see rookie league teams with 70+ players on the roster. But turn roster limits on and you get a different annoyance: mass signings and releases. The AI controlled teams will sign a player and release them sometimes in the same day. a dozen or more sign and release transactions involving the same handful of players every day is common.

What might fix this whole thing is if the minor league players were actually paid like in real life. if the parent clubs had a MiL salary budget with each farm club player getting the same salary (for simplicity) then there will be a finite number of players without need for roster limits or otherwise tricking the AI from signing every single free agent prospect in the game at that time.

Besides, unpaid MiL players are practically slaves in OOTP and that is wrong! LOL

NoOne · 02-02-2019, 08:53 PM

Quote:

Originally Posted by majesty95

Thanks for the post NoOne. Sucks you have to be that intricate about it lol. This is the first season with a 1995 historical (I think I inadvertently said it was a 2018 post initially).

Are there enough position players overall? I dont know. I'm rolling with what the game gave us. I will say, I have use incomplete minors off which removes the ability to set fill rosters top to bottom.

Either way, the AI could do a WAY better job than this. I'll try setting age limits and see what happens

just make sure 2-3 years are "possible" at most levels, especially Rookie and SA. if using real-world MiL system, it's gigantic and uneven. you'll want to focus on the largest system at any particular level to make sure you provide enough players. probably need to allow ~4 years at rookie without running low.

it may not be so large in 1995. a 2018 may need ~4-5years at rookie, if you set rules... enough to fill 35+30+30+30... teams, give room... how many rounds? how many years allowed... does it add up. 40rounds might be enough to fill 4 such teams over 3 years. that's 120 players, and may cut it thin for some.

you don'tneed any simming for to know if you can fill rookie.

onen note about historical... i think running low on minors common if you don't allow fictional players... it's unavoidable in many cases with this setting.

there are settings that prevent fictional players from reaching MLB, and that's an option you can make use of to help fill MiL. allow fictional players along with the historical, but don't allow them into MLB. or use the ghost players etc... tbh, the ghost players may be better, since they won't get in the way of playing time like fictional players could occasionally. the teams with "more" ghost players willb e dead last... by huge margin often, lol.

majesty95 · 02-08-2019, 06:51 PM

I had ghost on originally but we had a guy sign 25 free agent pitchers. So I worried guys would just stock up without paying mine to balancing their rosters. It may be an option we have to explore. Would be hard to keep fictionals out of a 20+ person MLB I think.

Mr. Marlin · 02-08-2019, 08:07 PM

If it's any consolation all leagues, even historical, become fictional eventually if you keep playing them.

majesty95 · 02-08-2019, 10:55 PM

Well when it takes 6 weeks to play a season (online league) and you start in 1995, you’re looking at 2.5 years of playing before the issue resolves itself... probbaly not a great selling point.

majesty95 · 02-08-2019, 10:55 PM

Dupe post

NoOne · 02-09-2019, 05:57 PM

make a copy of the league and zoom out... if you are the commissioner.

it doesn't even have to be the exact same league, just hte same settings. you can make a template from existing game and erase the players in game creation and re-fill etc... or something easier built-in already. so, it won't even give a glimpse into near future as far as competition in your league... (further out the less you can glean, so it should be irrelevant or avoidable even with the same players and just a copied .lg directory.)

the results aren't even relevant, you can just sim to specified date while you go grocery shopping.

01-31-2019, 02:45 PM	#1
majesty95 All Star Starter Join Date: Apr 2005 Location: San Antonio, TX Posts: 1,789	AI Minor League Evaluation Needs MAJOR Work In our online league, CPU team Toronto is completely butchering their minor league system. Their GCL team (2018 start) has 35 players and 32 are pitchers. Their DE team has only 15 players and they are ALL pitchers. We have the upper level minors (A-AAA) set to a max of 28 players and the rookie leagues set to 35. We also have incomplete minor leagues off. Their AA team has 8 OF while rookie has none. Several teams have 3 catchers and 3 1B while rookie has none. This evaluation system needs some serious work. __________________ College Football Sim League

01-31-2019, 03:17 PM	#3
NoOne Banned Join Date: Apr 2015 Posts: 7,273 Infractions: 1/2 (4)	Edit -- wrote this while other post was posted too... so, may be redundant in some spots? fwiw, his idea is FAR easier to do , but if you want mil rules to work properly on their own, read on.... original post below---- reduce rules and restrictions or massively increase your # of draft rounds or international discoveries or iafa etc if you want to keep existing rules and make them work no matter what... -- personal tip on roster limits: don't set any # limit at rookie... even "35". i used to be adamantly against roster # limits at any level, but i've been working with them lately and can avoid the pitfalls with a few undertandings. AAA -- use more than 28.. AAA must contain mlb depth as well as any additional prospects you care about on top of a functioning AAA team -- since many of those additional players likely won't be there next year, you have to cover your bases, figuratively and literally lol.... i use 34, but 32-34 is probably sufficient, if you use one... no roster restriction and just an age limit can do wonders too. A & AA -- a couple extra is fine. this is mostly for asthetics on #'s. i use 28, but .. all sorts will work to the point you don't see 8 1b or 32 RP etc... S-A -- bit on fence, but i typically set on lower than A/AA, but that's mostly how my age rules jive and how many players are typically available to play in SA each year etc etc. keep it similar to A/AA R -no roster limits.. even 35 is a problem. if rookies isn't getting enough, this is based on math you can predict. if you use age limits intead of service time at Rookie and Short A, it's a bit messier and harder to predict. in those cases you need more players per year created to make up for the greater disparity per team caused by age-based rules at rookie. i'm referencing service time limits because it's cleaner math. you can translate to age-based rules and add more than what is needed below for similar contexts based on age, as i said. # of players created per year / (# of years they can play at rookie * # of teams). without parnthesese, that's simply # of players / year and then that quotient further divided by number of teams. does it come out to anything near ~30-35? if not, you are starving those rookie leagues, and that's why some teams are all RP. (e.g. if "21" for rookie, that's ~3 years for high schoolers and less than 1 on average forcollege kids... that's ~1/2 the draft? if that? averages out to maybe 1.5 playeable years on average at rookie? if only 1 rookie team you need ~25-30 rounds at least. 2 rookie teams, and it's 50-60 rounds. you can see why you need "more.") ----- if you use MiL rules, you better work them out and ensure they work long-term. doing the math ahead of time helps. the further out you go from rookie, the murkier it gets to predict # of players per year. i have a post somewhere that maps out how you can calculate this stuff step by step, so i won't re-write the entirity here... at that time i probably didnt'like roster limits, go ahead and ignore that portion they can work, but again, can cause more problems then they solve if not implemented with some sort of over-arching MiL plan. strongly recommend using service time at rookie, if any restrictions, and age at A-AAA, if anything. s-a? whateer works and doesn't cause oddities. some differentiation between a/a+, obviously, but failed attempt to keep it simple. ----- if their season hasn't started and the draft hasn't taken place, don't assess the current state of rookie or short A seasons, yet. having too many duplicates of 1 position can happen, but some limited roster rules at A-AAA can reduce that. nothing you can do about rookie league without causing problems that are far worse than the asthetics of some redundancies on that roster. 1st base in minors is .. maleable, to say the least. lots of players that are future 1B, if anything, are listed as something else, amongst other reasons. take a peak at their actual defensive abilities... even re-position them in your head as you go... it's probably not as bad as it looks once they have the proper label. the experience bars in their profile can give a clue too as you scroll through... if it varies from "position" then you know they really aren't that position based on use in the minors. in the minors, what they are labeled as is often not what they do. 8 OF isn't too bad on a mil team. bet that happens in RL too. not that i'd use that as a bar for ootp mil in all cases, but it isn't too bad relative to RL in this case. fwiw, i have a MiL system of rules that functions and never displays the things you mention and allows for easy movement of younger/better prospects that develop faster, even in the AI systems. what i mention works, or can work if done in a way that makes sense. ther are far more vairations that cannot work than variations that can work. some aspects of how the game works may not allow for any one persons exact perception or wishes of how it should work, either. Last edited by NoOne; 01-31-2019 at 03:30 PM.

01-31-2019, 06:56 PM	#4
koohead Hall Of Famer Join Date: Dec 2001 Location: Phoenix, AZ Posts: 3,200	I've seen this concern raised on multiple versions over the years but luckily have personally never experienced it. I do set roster limits, as well as service time limits. I wonder if there is enough available players in the FA pool so that teams have enough players to sign? there has to be some factor that is causing the AI to make these types of odd personnel decisions. My FA pool is always in the thousands...I think it's sitting at 1200 or something right now a month before Spring Training. so even mid season I have hundreds of people available. having enough of a pool of varying ages and talent levels (and perhaps generating enough rounds for your draft pool) may give the AI enough players to pull from so that it doesn't have to fill out the roster with anything it can get its hands on. __________________ GM - New Jersey Bears of the NPBL; Last edited by koohead; 01-31-2019 at 06:58 PM.

01-31-2019, 11:54 PM	#5
majesty95 All Star Starter Join Date: Apr 2005 Location: San Antonio, TX Posts: 1,789	Thanks for the post NoOne. Sucks you have to be that intricate about it lol. This is the first season with a 1995 historical (I think I inadvertently said it was a 2018 post initially). Are there enough position players overall? I dont know. I'm rolling with what the game gave us. I will say, I have use incomplete minors off which removes the ability to set fill rosters top to bottom. Either way, the AI could do a WAY better job than this. I'll try setting age limits and see what happens __________________ College Football Sim League

02-08-2019, 06:51 PM	#8
majesty95 All Star Starter Join Date: Apr 2005 Location: San Antonio, TX Posts: 1,789	I had ghost on originally but we had a guy sign 25 free agent pitchers. So I worried guys would just stock up without paying mine to balancing their rosters. It may be an option we have to explore. Would be hard to keep fictionals out of a 20+ person MLB I think. __________________ College Football Sim League

02-01-2019, 07:18 AM	#6
Mr. Marlin All Star Starter Join Date: Apr 2010 Posts: 1,274	I have seen a several annoying problem with minor league rosters. Even with age and service limits it's not uncommon to see rookie league teams with 70+ players on the roster. But turn roster limits on and you get a different annoyance: mass signings and releases. The AI controlled teams will sign a player and release them sometimes in the same day. a dozen or more sign and release transactions involving the same handful of players every day is common. What might fix this whole thing is if the minor league players were actually paid like in real life. if the parent clubs had a MiL salary budget with each farm club player getting the same salary (for simplicity) then there will be a finite number of players without need for roster limits or otherwise tricking the AI from signing every single free agent prospect in the game at that time. Besides, unpaid MiL players are practically slaves in OOTP and that is wrong! LOL

02-08-2019, 08:07 PM	#9
Mr. Marlin All Star Starter Join Date: Apr 2010 Posts: 1,274	If it's any consolation all leagues, even historical, become fictional eventually if you keep playing them.

02-08-2019, 10:55 PM	#10
majesty95 All Star Starter Join Date: Apr 2005 Location: San Antonio, TX Posts: 1,789	Well when it takes 6 weeks to play a season (online league) and you start in 1995, you’re looking at 2.5 years of playing before the issue resolves itself... probbaly not a great selling point. __________________ College Football Sim League

02-08-2019, 10:55 PM	#11
majesty95 All Star Starter Join Date: Apr 2005 Location: San Antonio, TX Posts: 1,789	Dupe post __________________ College Football Sim League

02-09-2019, 05:57 PM	#12
NoOne Banned Join Date: Apr 2015 Posts: 7,273 Infractions: 1/2 (4)	make a copy of the league and zoom out... if you are the commissioner. it doesn't even have to be the exact same league, just hte same settings. you can make a template from existing game and erase the players in game creation and re-fill etc... or something easier built-in already. so, it won't even give a glimpse into near future as far as competition in your league... (further out the less you can glean, so it should be irrelevant or avoidable even with the same players and just a copied .lg directory.) the results aren't even relevant, you can just sim to specified date while you go grocery shopping. Last edited by NoOne; 02-09-2019 at 05:59 PM.