Lineup Substitutions for Injured Starters Need to be Improved

jslmsca · 02-24-2018, 09:10 AM

I've noticed that the AI does not readjust lineups when substituting for an injured starter. For instance, my starting LF bats cleanup but gets injured for about a week. The AI will substitute a backup but then still plays him at #4 in the lineup when he should never be batting that high in the lineup.

There needs to be logic added to reevaluate a lineup when replacing a starter, not just a simple swap into the current position.

bwburke94 · 02-24-2018, 10:35 AM

If you allow other players to move up to #4, we're right back into the "bad player moved into leadoff spot" conundrum. You specifically set the other players in the lineup to bat #5/#6/#7/#8 (and #9 if DH), and the AI won't change that.

NoOne · 02-24-2018, 04:10 PM

That's not true. it'd be a very simple qualification for that sort of robustness from moving a #2+ to #1 in lineup. that info is readily available for use in any sort of logic you want to create.

this isn't rocket science... i bet i could come up with some 'simple' nested lop/dowhile logic to easily handle all substitutions of any kind.

compartmentalize the individual actions and it's even easier in a modular way.

heck, it could simple go through the exact same process that exists for making a lineup and exclude the hurt player... i would hope that's how the AI handles it. it wouldn't even require new code.

if you don't delegate that stuff, then it's your responsibility to act.

either way, there is nothing holding them back on this issue in regards to "barriers to entry" or being "too hard to do" excuse. they simply haven't done anything about it...

that's a neutral statement in the big picture, btw. lots of things can be worked on and limited time to work on them. but, it's defintieyl not toohard or impossible to do... in fact very little heavy lifting involved for a well-thought out process to replace current one - lots of functions can likely be re-used if they were written properly the first time.

along the smae lines it's absolutely pitiful the amount of work you have to do to use a catcher as a dh on off days and primary catcher otherwise. although the game's fatigue for catchers isn't conducive to that, if you use them less as a catcher and more as a dh they don't get tired... you can play them 150-162 games if smart about it.

with a great offensive catchers, that's a big deal. the difference between ~average and ~great for a catcher is much larger than other positions... ie greater return on improving it, ceterus paribus. ootp18 has defintieyl cut back on quality of catchers... the highest power catcher in 50years has been ~70/100 (140/200 ish). 1 other guy was 130's (excludes players with 'power' that aren't viable mlb hitters, ie low contact, no defense at catcher etc)

Brad K · 02-24-2018, 04:44 PM

It seems to me there are too many ifs involved for a human to cover them all in advance through preferences or to set a lineup for each possible situation.

If a spot seven or eight batter is injured it can't be very wrong with the substitute in his spot. If the injured batter is in 1 through 6, then moving x through 7 up one and putting the replacement at 7 is a reasonable solution.

NoOne · 02-25-2018, 01:03 PM

if the ai can create a lineup with 13 available players, it can create a lineup with 12 and a replacement (or just 12). same exact logic - as it already exists.

plus, it's actually a finite amount of "ifs" and most defintieyl can be covered ahead of time in an efficient and logical manner. just as there are a finite # of baseball situations to plan for.

you even showed just one simple qualification could eliminate a large portion of the AI's failure in this realm. i'm not just suggestion adding a few ifs as needed. there are ways to limit the # of ifs with a well-thought out approach to the problem.

e.g. if someon asks you to guess 1-100 and they will reply with 'lower' or 'higher', it should never take more than 7 guesses with the proper approach. -- 50, 75, 88, 94, 97, 99, answer must be known by now. by doing things in a particular order, it won't require many "ifs"

but, you can't expect it to do anything beyond a swap if you don't delegate linups to the bench coach.

if it is delegated and this is how it handles injuries thats %$#@ing pitiful -- not holding back on that one. i never give control of lineups, so i'm not 100% sure, but i doubt it would be that way if you delegate lineups... if it is go bomb the future suggestions forum everyday with a new thread, because that's ridiculous if true. i doubt it, because when i use "ask manager for lineups" it doesn't put the new guy onthe 25-man roster in the injured guy's spot (mil stuff, not mlb for this 'click').

3outs * 12 possible counts * 8 on-base configurations = 288 possible situations in baseball.. 288 is far from unmanageable for a program or even a human with a simple reference card that is easily made in advance with a ruler, paper and pen. in many contexts the same startegies would apply, so it's far fewer than 288 in application.

the game of baseball is simple and quantifiable in a babytown-frolics kind of way relative to the complexities of things we already know and understand. being predictable or understanding it thoroughly is completely different idea from knowing what will happen while enjoying a game of baseball... you have no idea what will happen next from that perspective of entertainment.

before somethign occurs, all possibilities are true. the actual result of one data point is meaningless on its own.

02-24-2018, 04:10 PM	#3
NoOne Banned Join Date: Apr 2015 Posts: 7,273 Infractions: 0/1 (3)	That's not true. it'd be a very simple qualification for that sort of robustness from moving a #2+ to #1 in lineup. that info is readily available for use in any sort of logic you want to create. this isn't rocket science... i bet i could come up with some 'simple' nested lop/dowhile logic to easily handle all substitutions of any kind. compartmentalize the individual actions and it's even easier in a modular way. heck, it could simple go through the exact same process that exists for making a lineup and exclude the hurt player... i would hope that's how the AI handles it. it wouldn't even require new code. if you don't delegate that stuff, then it's your responsibility to act. either way, there is nothing holding them back on this issue in regards to "barriers to entry" or being "too hard to do" excuse. they simply haven't done anything about it... that's a neutral statement in the big picture, btw. lots of things can be worked on and limited time to work on them. but, it's defintieyl not toohard or impossible to do... in fact very little heavy lifting involved for a well-thought out process to replace current one - lots of functions can likely be re-used if they were written properly the first time. along the smae lines it's absolutely pitiful the amount of work you have to do to use a catcher as a dh on off days and primary catcher otherwise. although the game's fatigue for catchers isn't conducive to that, if you use them less as a catcher and more as a dh they don't get tired... you can play them 150-162 games if smart about it. with a great offensive catchers, that's a big deal. the difference between ~average and ~great for a catcher is much larger than other positions... ie greater return on improving it, ceterus paribus. ootp18 has defintieyl cut back on quality of catchers... the highest power catcher in 50years has been ~70/100 (140/200 ish). 1 other guy was 130's (excludes players with 'power' that aren't viable mlb hitters, ie low contact, no defense at catcher etc) Last edited by NoOne; 02-24-2018 at 04:16 PM.

02-24-2018, 09:10 AM	#1
jslmsca All Star Reserve Join Date: Nov 2013 Location: Calgary Posts: 605	Lineup Substitutions for Injured Starters Need to be Improved I've noticed that the AI does not readjust lineups when substituting for an injured starter. For instance, my starting LF bats cleanup but gets injured for about a week. The AI will substitute a backup but then still plays him at #4 in the lineup when he should never be batting that high in the lineup. There needs to be logic added to reevaluate a lineup when replacing a starter, not just a simple swap into the current position.

02-24-2018, 10:35 AM	#2
bwburke94 Hall Of Famer Join Date: Jun 2008 Location: Belchertown, MA, USA Posts: 4,525	If you allow other players to move up to #4, we're right back into the "bad player moved into leadoff spot" conundrum. You specifically set the other players in the lineup to bat #5/#6/#7/#8 (and #9 if DH), and the AI won't change that.

02-24-2018, 04:44 PM	#4
Brad K Banned Join Date: May 2016 Location: St Petersburg Florida USA Posts: 6,693 Infractions: 0/2 (4)	It seems to me there are too many ifs involved for a human to cover them all in advance through preferences or to set a lineup for each possible situation. If a spot seven or eight batter is injured it can't be very wrong with the substitute in his spot. If the injured batter is in 1 through 6, then moving x through 7 up one and putting the replacement at 7 is a reasonable solution.

02-25-2018, 01:03 PM	#5
NoOne Banned Join Date: Apr 2015 Posts: 7,273 Infractions: 0/1 (3)	if the ai can create a lineup with 13 available players, it can create a lineup with 12 and a replacement (or just 12). same exact logic - as it already exists. plus, it's actually a finite amount of "ifs" and most defintieyl can be covered ahead of time in an efficient and logical manner. just as there are a finite # of baseball situations to plan for. you even showed just one simple qualification could eliminate a large portion of the AI's failure in this realm. i'm not just suggestion adding a few ifs as needed. there are ways to limit the # of ifs with a well-thought out approach to the problem. e.g. if someon asks you to guess 1-100 and they will reply with 'lower' or 'higher', it should never take more than 7 guesses with the proper approach. -- 50, 75, 88, 94, 97, 99, answer must be known by now. by doing things in a particular order, it won't require many "ifs" but, you can't expect it to do anything beyond a swap if you don't delegate linups to the bench coach. if it is delegated and this is how it handles injuries thats %$#@ing pitiful -- not holding back on that one. i never give control of lineups, so i'm not 100% sure, but i doubt it would be that way if you delegate lineups... if it is go bomb the future suggestions forum everyday with a new thread, because that's ridiculous if true. i doubt it, because when i use "ask manager for lineups" it doesn't put the new guy onthe 25-man roster in the injured guy's spot (mil stuff, not mlb for this 'click'). 3outs * 12 possible counts * 8 on-base configurations = 288 possible situations in baseball.. 288 is far from unmanageable for a program or even a human with a simple reference card that is easily made in advance with a ruler, paper and pen. in many contexts the same startegies would apply, so it's far fewer than 288 in application. the game of baseball is simple and quantifiable in a babytown-frolics kind of way relative to the complexities of things we already know and understand. being predictable or understanding it thoroughly is completely different idea from knowing what will happen while enjoying a game of baseball... you have no idea what will happen next from that perspective of entertainment. before somethign occurs, all possibilities are true. the actual result of one data point is meaningless on its own.