Why would the AI release this player?

[SirMichaelJordan]

Easy, he's declining and must have refused a demotion and was released. Want other years of stats to matter more? raise them. I believe current year stats are prorated until a certain amount of appearances are reached in the current season, not sure though.

but 50% ratings would suggest they see him declining on the 1-250 scale.

[sprague]

I still get an odd situation, only with my organization, not the ai.
if i click, ai set up my complete organization...it will always tend to release a 4 or 5 star prospect in the minors.
then i have to go and manually pit them back

[MizzouRah]

Originally Posted by SirMichaelJordan

Easy, he's declining and must have refused a demotion and was released. Want other years of stats to matter more? raise them. I believe current year stats are prorated until a certain amount of appearances are reached in the current season, not sure though.

but 50% ratings would suggest they see him declining on the 1-250 scale.

He doesn't show any signs of declining..

I personally do not think the AI evaul settings matter as much as we think they might... this is something Markus needs to either explain or have it where you can set month by month settings.. I would like ratings to way more early on.. then stats close to the trade deadline.. might be a way to help the AI?

[SirMichaelJordan]

Originally Posted by MizzouRah

He doesn't show any signs of declining..

I personally do not think the AI evaul settings matter as much as we think they might... this is something Markus needs to either explain or have it where you can set month by month settings.. I would like ratings to way more early on.. then stats close to the trade deadline.. might be a way to help the AI?

Still doesn't mean his numbers under the hood aren't declining. He's 39 so his attributes are on the downward even though it may not look like it is on his profile. His stats are actually showing bit of a decline.

Sent from my Z958 using Tapatalk

[MizzouRah]

Originally Posted by SirMichaelJordan

Still doesn't mean his numbers under the hood aren't declining. He's 39 so his attributes are on the downward even though it may not look like it is on his profile. His stats are actually showing bit of a decline.

Sent from my Z958 using Tapatalk

True.. maybe I need to weigh more towards stats then ratings?

[SirMichaelJordan]

Originally Posted by MizzouRah

True.. maybe I need to weigh more towards stats then ratings?

Also the manager or GM might prefer younger players than vets. So he may have been the only option to be sent down if the other younger guys didn't have any more options years left. I'm guessing the vet refused demotion and the team released him. Would be cool if they shopped him around first.

I believe a couple of things may have played a role in his release.
Sent from my Z958 using Tapatalk

[PSUColonel]

I am still torn between two settings that I think are pretty good all around since we don't have dynamic ratings yet.

55/25/15/5. In this case, last year is weighted 3 times more than 2 years ago....the current year is weighted about 1.65 x more than last years. It means also that this years stats are weighted 5x more than 2 years ago


60/25/10/5. In this case, last year is weighted twice as much as 2 years ago, this year is weighted 2.5 x more times than last year, and again, this year is weighted 5x more than two years ago.

[PSUColonel]

So the big question here is:

Should this year be worth 2.5 x more than last year, and last year be worth twice two years ago?

or

Should this year be worth only 1.65x more than last year, but last year is worth 3x two years ago?



EDIT: The more I play around with these settings though...I am leaning towards 55/25/15/5.....it just seems to strike the right balance, and I think the AI is actually a little tougher in trade negotiations as a result. It won't trade you a great defensive catcher with minimal offensive skills for junk, weras with the higher emphasis on ratings, it will. Odd I know, but somehow this settings just hits the sweet spot it seems. Don't know why.

[PSUColonel]

of course with the default 65/20/10/5


every year is worth 2x more than the last & this year is worth 4x more two years ago.

[MizzouRah]

Originally Posted by PSUColonel

of course with the default 65/20/10/5


every year is worth 2x more than the last & this year is worth 4x more two years ago.

Right.. but it's only 35% of the evaluation.That's why I like 50/30/15/5.. 50% ratings and 50% stats where 2 years ago is 3x of 3 years ago and this years stats is 2x more important than last season and 6x more than 3 years ago.

[PSUColonel]

Originally Posted by MizzouRah

Right.. but it's only 35% of the evaluation.That's why I like 50/30/15/5.. 50% ratings and 50% stats where 2 years ago is 3x of 3 years ago and this years stats is 2x more important than last season and 6x more than 3 years ago.

This is just me, but I personally feel you need to give more than 50 percent to ratings. Not much as you can see, but I do think it makes a difference. It could be the reason you have seen some odd AI behavior. I can understand why you like the idea of 50/30/15/5, but I just feel that little extra nudge toward stats, and a little less of a difference between the current and previous years makes a difference. I could even argue that while 55/25/15/5 may put a little more emphasis on ratings, the stats "body of work" is considered as opposed to one year. It heavily favors this year and last year, but doesn't weigh any one of them too heavily....but the current year still gets higher consideration by 1.65 x

[Rain King]

Originally Posted by PSUColonel

of course with the default 65/20/10/5


every year is worth 2x more than the last & this year is worth 4x more two years ago.

That is the default for the MLB quickstart. The default for fictional games is 30/50/15/5. The MLB quickstart has to have a higher "ratings" factor in order to get the AI to use/assign players more like their real life counterparts.

The game, IMO, functions more realistically at something closer to the fictional defaults because it becomes reactive (to results) instead of proactive to ratings changes.

My guess is that in the OP situation the Dodgers scout saw a large ratings drop (pretty likely at 39) and dumped him.

[MizzouRah]

Originally Posted by Rain King

That is the default for the MLB quickstart. The default for fictional games is 30/50/15/5. The MLB quickstart has to have a higher "ratings" factor in order to get the AI to use/assign players more like their real life counterparts.

The game, IMO, functions more realistically at something closer to the fictional defaults because it becomes reactive (to results) instead of proactive to ratings changes.

My guess is that in the OP situation the Dodgers scout saw a large ratings drop (pretty likely at 39) and dumped him.

You could very well be right. If this is the only odd AI decision then Markus is a genius Seriously.. I remember seeing a lot of this in previous versions but it gets better every year.

I still believe dynamic ai eval settings would be awesome.

Since I'm several years into my universe and have a ton of fictional players now, I just might try 30/50/15/5 next season.

[PSUColonel]

Originally Posted by Rain King

That is the default for the MLB quickstart. The default for fictional games is 30/50/15/5. The MLB quickstart has to have a higher "ratings" factor in order to get the AI to use/assign players more like their real life counterparts.

The game, IMO, functions more realistically at something closer to the fictional defaults because it becomes reactive (to results) instead of proactive to ratings changes.

My guess is that in the OP situation the Dodgers scout saw a large ratings drop (pretty likely at 39) and dumped him.

If this is true....and it very well may be...I am experimenting again, and seem to see what your are saying....then something in the line of 30/40/20/10 might not be a bad choice. For years, I used 40/30/20/10 with pretty good results, and am wondering if the just flipping the 30 & 40 makes a bit of difference, since that would in fact be closer to the fictional default.

[jpeters1734]

question about the ratings portion. Does it use true ratings or scout ratings? I ask because i use low scouting accuracy.

[PSUColonel]

scouted

[NoOne]

Originally Posted by MizzouRah

He was a BEAST for me in 2023, AL reliever of the year and 2nd in Cy Young voting.

I don't really look at age when you see his stats.. stats always trumps age. Like you said.. why sign him - he's pitched well so at least let him help you to a playoff birth and don't sign him in the off season. He's still 4* and one of the best options you had coming out of the pen.

There's more to the decisions than that for the AI. you can look at a situation and without thought exclude many options without even thinking about it.. for the ai, if the qualifications are met, it will try to do it without question. so, age as well as other factors likely caused this. it's neither right or wrong in this context, but rather comes down to opinions. it's not an odd move. (with that velocity and anywhere near that age, he'd never be in my pen unless a LR or last-used MR)

Age is a very important factor to be ignoring even when looking at stats.

as a player ages recent history becomes less predictive of the future, since they are likely decrapitating. especially since his numbers have dropped recently, despite being good. if the scouting reports show a decline coinciding with that... it's likely to continue at a precipitious pace due to his age... and if it doesn't tha'ts the same as going to las vegas and choosing of your own volition to bet against the odds (bad odds*).

same way you likely look at younger, not fully developed players in the opposite way... their #'s aren't as predictive of the future as when they come from his prime years, but they are likely to get better if there is still room for improvement. not considering either of these forces is a bad idea.

stats are just results... results can be wrong in a the short-term. a changing player equates to short-term since they are no longer rated the same as previously when they put up those stats (meaningful change, like a significant increase or decrease in ability). even more so with releivers... you need upto 10 years to know what you go, lol. i just had a guy put up 4 straight 2+ WAR years out of the pen with whips around ~.800. I couldn't believe it relative to his ratings... then bam... after i sign him for 4 years he starts pitching like his ratings... luck happens. it's decieving... trust the ratings especially when the player is changing shapes.

[MizzouRah]

Originally Posted by PSUColonel

If this is true....and it very well may be...I am experimenting again, and seem to see what your are saying....then something in the line of 30/40/20/10 might not be a bad choice. For years, I used 40/30/20/10 with pretty good results, and am wondering if the just flipping the 30 & 40 makes a bit of difference, since that would in fact be closer to the fictional default.

I'm going to switch to fictional default at the end of this current season. I'll post back on anything I find noticeable.. if anything.

[MizzouRah]

Here is another one using my current eval settings.. (haven't changed yet)

San Francisco is leading the NL West by 5 games with 25 games to go. There is no reason to release this player. He was an All Star this year to boot. Now Washington signs him to help them stay atop the NL East.

The news article even mentions this signing as Washington winning the lottery for his services - BUT does mention his age as a concern. So, I believe you guys are right when you say the AI does these moves due to a players age.

I'm just not sure changing eval settings will help fix this... but maybe changing to 30/50/15/5 will help eliminate these types of releases. I'm currently using 50/30/15/5, so essentially I'll be switching the 50 and 30 (ratings and stats around).

[NoOne]

Originally Posted by MizzouRah

Here is another one using my current eval settings.. (haven't changed yet)

San Francisco is leading the NL West by 5 games with 25 games to go. There is no reason to release this player. He was an All Star this year to boot. Now Washington signs him to help them stay atop the NL East.

The news article even mentions this signing as Washington winning the lottery for his services - BUT does mention his age as a concern. So, I believe you guys are right when you say the AI does these moves due to a players age.

I'm just not sure changing eval settings will help fix this... but maybe changing to 30/50/15/5 will help eliminate these types of releases. I'm currently using 50/30/15/5, so essentially I'll be switching the 50 and 30 (ratings and stats around).

it would only help if those types of players consistently perform better than their ratings... so, probably not going to matter.

but, in the context of a player that does do this, they are probably valued more highly due to the higher overall ratings... that would be one tick for the "improve" column, but doesn't guarantee a decisions... it might be simply at that age if a viable destination is found, they just do it, so that they get something, rather than nothing.

if these types of player rotatate through more than 2 teams a season then i'd definitely bring that to their attention. e.g. if the destination teams takes the trade, they should be more willing to retain longer, since that's the only reason they'd go for it.

if you want the AI to make decisions based on data that isn't necessarily sound, increase stats portion of the ai evaluation (it's using small sample sizes). this only affects overall, so it's quite inconsequential to most human beings -> it's a common suggestion to ignore overall rating when making a decision - e.g. i only use it for filtering lists. i currently use 50-60% ratings and the rest stats, but i'm thinking of testing even less stats the more i think about it. i wish i could set the AI to have a better scouting accuracy, too...

i like being hamstrung, but the AI needs all the help it can get. giving it bad info can't be helpful. if i recall offensive numbers increase 5-10% with 100% ratings... but i could be confusing that with the accuracy settings. it's been a few years at this point.