Player AI Evaluation Settings

Chomps · 03-26-2018, 06:19 PM

What are the recommended settings for OOTP 19?

PSUColonel · 03-27-2018, 06:40 PM

personally, I prefer either the default 65/20/10/5 or 55/25/15/5

The default ensures the perceived best players will play. It is also a good setting since the user sees overall AI evaluation in the players' overall rating...the option to see this based solely on ratings was removed. It will also help the AI with trades, waivers and other transactions.

The benefit of 55/25/15/5 is that actual performance is rewarded a bit more, and that ratings increase/decline aren't punished or rewarded quite as immediately. So for example the AI might be less likely to release a player just because he took a ratings hit. Contract negotiations may reward actual performance a bit more here also.

Now, if you want settings which reward performance much more, you could move to 50/30/15/5, or 40/30/20/10.

EDIT: These are the ones most people use, and it's really preference, although I feel the AI performs a bit better with ratings weighted more heavily. (Greater than 50 percent)

RubeBaker · 03-27-2018, 07:07 PM

Default.

PSUColonel · 03-27-2018, 07:16 PM

Quote:

Originally Posted by RubeBaker

Default.

I think the biggest reason this debate has raged on for years' is the fact the fictional MLB setup puts the AI evaluation at 30/50/15/5...which is obviously very stats heavy. Couple this with the fact the developers have admitted the MLB startup rosters are defaulted at 65/20/10/5 simply so the league will force certain players into starting roles, so that the league can mirror the actual MLB as much as possible....so in other words, the higher ratings "massage" certain players' into certain roles. Matt also commented that being too stats heavy will cause the AI to POTENTIALLY overreact to rating changes, while being to stats heavy will cause it to POTENTIALLY reward performance based on a small sample size.

PSUColonel · 04-02-2018, 12:20 AM

Are people using the default? How is it going?

NoOne · 04-02-2018, 01:48 AM

there'll only be fringe differences unless you make it drastically different. like 90ratings vs 10 ratings... or maybe 30-60? big enough? you'd have to compare the two and see who the ai gms change their lineups and depth charts etc.. longer term study - who they sign and pay differently in FA etc.

from what i see, most people jsut asy "this works" type stuff.. and really have no idea the effect it is having.. and neither do i, if that isn't obvious. never looked into it

my suggestion -- tie it to your choices of scouting accuracy.. the more accurate scouting is the more you should base it on ratings. the less accurate the ratings, the more you should place weight on stats. i would think this would minimize bad ai choices, but of course would never eliminate them.

the 2018 mlb league may start off with a differnt set, and it's for the specific reason that real players will keep their jobs at a higher %. well, at least the statistically successful ones.

Bears5122 · 04-02-2018, 03:38 AM

Isn't one of the problems that scouting plays a bigger role if you set ratings higher? So if a team has bad scouting, they'll continue to play worse players over those who might be better and are outperforming them?

Another reason why I don't have ratings too high is because it leads to unrealistic changes by managers. For instance you have a 32-year old who plays at an All-Star level for the past 5 years. Now his ratings drop in the offseason because he's starting getting older. Magically he's a bench guy to start next season. This just wouldn't happen that quickly. He'd likely be able to play until his stats showed that decline.

I usually keep ratings at 45 because it allows the AI to make good decisions but also keeps things realistic. But I also have ratings set at a scale of 2 to 8 for myself so I don't have as much data to use and have to rely on stats myself.

NoOne · 04-02-2018, 09:49 AM

results are stongly ... very stongly correlated to ratings.. they are what cause the results, eh?

i think ratings lage a bit to changs, but not anything like stats, of course.

more stats may make that particular situation or erealistic but it is a multi-edged sword, so to speak.

it will also cause that would-be ASG to lose his position to a lesser quality player due to a bad year etc... which would not happen in RL, either.

stats are volatile for most players, so i'd bet the latter happens more than the previous example you used.

there's no right answer here.. don't fret over the choices... try one method for a few years, try a new one... and i bet you cannot tell the difference without taking extensive notes and data mining the results in some way.

i've tweaked it anywhere from 40-65 and i cannot tell a difference. i typically shift a small amount from "3 years ago" to the 2 more recent years. usually 5-7% used for 3 years ago. then whatever is leftover for ratings, that's what i use. i keep it close to 50/50 nowadays.. give or take 5%.

HarryTMMyt · 02-11-2021, 11:20 AM

I may just be being a complete madman here but I use 10/30/30/30

03-26-2018, 06:19 PM	#1
Chomps Major Leagues Join Date: Apr 2002 Location: Taiwan Posts: 454	Player AI Evaluation Settings What are the recommended settings for OOTP 19? Attached Images

03-27-2018, 06:40 PM	#2
PSUColonel Hall Of Famer Join Date: Feb 2002 Posts: 13,089 Blog Entries: 37	personally, I prefer either the default 65/20/10/5 or 55/25/15/5 The default ensures the perceived best players will play. It is also a good setting since the user sees overall AI evaluation in the players' overall rating...the option to see this based solely on ratings was removed. It will also help the AI with trades, waivers and other transactions. The benefit of 55/25/15/5 is that actual performance is rewarded a bit more, and that ratings increase/decline aren't punished or rewarded quite as immediately. So for example the AI might be less likely to release a player just because he took a ratings hit. Contract negotiations may reward actual performance a bit more here also. Now, if you want settings which reward performance much more, you could move to 50/30/15/5, or 40/30/20/10. EDIT: These are the ones most people use, and it's really preference, although I feel the AI performs a bit better with ratings weighted more heavily. (Greater than 50 percent) Last edited by PSUColonel; 04-02-2018 at 06:41 AM.

03-27-2018, 07:07 PM	#3
RubeBaker All Star Reserve Join Date: Mar 2015 Location: USA Posts: 642	Default.

04-02-2018, 12:20 AM	#5
PSUColonel Hall Of Famer Join Date: Feb 2002 Posts: 13,089 Blog Entries: 37	Are people using the default? How is it going?

04-02-2018, 01:48 AM	#6
NoOne Hall Of Famer Join Date: Apr 2015 Posts: 7,228	there'll only be fringe differences unless you make it drastically different. like 90ratings vs 10 ratings... or maybe 30-60? big enough? you'd have to compare the two and see who the ai gms change their lineups and depth charts etc.. longer term study - who they sign and pay differently in FA etc. from what i see, most people jsut asy "this works" type stuff.. and really have no idea the effect it is having.. and neither do i, if that isn't obvious. never looked into it my suggestion -- tie it to your choices of scouting accuracy.. the more accurate scouting is the more you should base it on ratings. the less accurate the ratings, the more you should place weight on stats. i would think this would minimize bad ai choices, but of course would never eliminate them. the 2018 mlb league may start off with a differnt set, and it's for the specific reason that real players will keep their jobs at a higher %. well, at least the statistically successful ones.

04-02-2018, 03:38 AM	#7
Bears5122 All Star Reserve Join Date: Mar 2003 Posts: 847	Isn't one of the problems that scouting plays a bigger role if you set ratings higher? So if a team has bad scouting, they'll continue to play worse players over those who might be better and are outperforming them? Another reason why I don't have ratings too high is because it leads to unrealistic changes by managers. For instance you have a 32-year old who plays at an All-Star level for the past 5 years. Now his ratings drop in the offseason because he's starting getting older. Magically he's a bench guy to start next season. This just wouldn't happen that quickly. He'd likely be able to play until his stats showed that decline. I usually keep ratings at 45 because it allows the AI to make good decisions but also keeps things realistic. But I also have ratings set at a scale of 2 to 8 for myself so I don't have as much data to use and have to rely on stats myself.

04-02-2018, 09:49 AM	#8
NoOne Hall Of Famer Join Date: Apr 2015 Posts: 7,228	results are stongly ... very stongly correlated to ratings.. they are what cause the results, eh? i think ratings lage a bit to changs, but not anything like stats, of course. more stats may make that particular situation or erealistic but it is a multi-edged sword, so to speak. it will also cause that would-be ASG to lose his position to a lesser quality player due to a bad year etc... which would not happen in RL, either. stats are volatile for most players, so i'd bet the latter happens more than the previous example you used. there's no right answer here.. don't fret over the choices... try one method for a few years, try a new one... and i bet you cannot tell the difference without taking extensive notes and data mining the results in some way. i've tweaked it anywhere from 40-65 and i cannot tell a difference. i typically shift a small amount from "3 years ago" to the 2 more recent years. usually 5-7% used for 3 years ago. then whatever is leftover for ratings, that's what i use. i keep it close to 50/50 nowadays.. give or take 5%.

02-11-2021, 11:20 AM	#9
HarryTMMyt Bat Boy Join Date: Aug 2020 Location: Liverpool Posts: 9	I may just be being a complete madman here but I use 10/30/30/30