Talent Change Randomness

[OutS|der]

Originally Posted by SirMichaelJordan

That's not a TCR thing. That is most likely a scouting thing. What setting are you using to where current ratings are higher than potential ratings?

According to devs, talent is referring to potential ratings, not current ratings. If current ratings are being hit, it is because potential rating have changed, or in other words the game adjusting. TCR does not directly hit current ratings.

Can you link where the Devs said that?

The more I learn about TCR and league totals the more I dislike the game.

[Rain King]

Originally Posted by Dutch Alexander

Below a link to the league quickstart. This is the base file from which I run all my tests. There are no players and coaches at the start. When I do a test I fill all 4 leagues with fictional players and coaches and then simulate 12 seasons to get rid of the statistical "noise" and delete all leagues history.

https://app.mediafire.com/myfiles

If someone has time to look at it I would really appreciate that. Maybe there is a setting I should or shouldn't use. Maybe there is a silly mistake that I overlook. File contains the latest settings I've been using in my tests.

I've noticed that if I change the scouting rating scale from 1-5 to 2-8 and when I switch on the current rating scale things improve. With those settings it is now rare that such young players are put on the active roster. I guess with a "larger" rating scale the AI can distiguish better between players? But there still a few teams that occasionally insist on putting these players on the active roster. I could put a minimum age of 18 on the major league teams and highest level of minors but then you are fighting the symptoms instead of the disease.

This link doesn't seem to go to your file. It goes to the "MyFiles" area of my own account for me.

[Dutch Alexander]

Originally Posted by Rain King

This link doesn't seem to go to your file. It goes to the "MyFiles" area of my own account for me.

Ah yes, I messed it up. Below link should work.

https://www.mediafire.com/file/ap7w1...quick.zip/file

It seems not setting roster limits to the minor league teams helps. AI can't handle roster limits well at all and this leads to strange behaviour. Still doen's make sense why such young players are put on the active roster but without roster limits it seems to happen only occasionally.

[Brad K]

Originally Posted by jcard

This is absurd.

In lieu of the adequate documentation one would expect, will a developer please provide an answer to the simple question of whom is affected by TCR?

They don't know. There's no internal documentation that's better than the game manual that they can refer to. All they have is memories. That's why devs start their answer with phrases like "If I remember correctly..." and "I think what we did there was..."

[Rain King]

Originally Posted by Dutch Alexander

Ah yes, I messed it up. Below link should work.

https://www.mediafire.com/file/ap7w1...quick.zip/file

It seems not setting roster limits to the minor league teams helps. AI can't handle roster limits well at all and this leads to strange behaviour. Still doen's make sense why such young players are put on the active roster but without roster limits it seems to happen only occasionally.

Ok, so my first thoughts when just looking over your settings and before I sim forward at all.

You have TCR jacked up, Batter Dev. Speed jacked up and Development Target Age set to "Younger". This is all seemingly geared towards young players progressing very quickly.

You also have player evaluation set to a large (65%) Ratings Weight and Scouting Report Updates set to "Rare" and Scouting Accuracy set to "Low".

So, I am thinking that what could be happening is that these young players actually do have major league quality ratings, but that the Scouted ratings just haven't caught up with that yet. It might be worth looking under the hood at any questionable players you are seeing on your end to see what their underlying ratings actually are. Or maybe it is the Low scouting accuracy and the teams think the players are much better than you are seeing (I think to check this you can act as the team and look at their scout evaluations).

[Rain King]

Some other things I wonder if could be causing weirdness.

Your highest talent league has Player Creation Modifiers at .800. While the ratings are set to be relative to that league I still wonder if the game isn't expecting an environment where the talent is muted in that way. Essentially, I think this squeezes the portion of the under-the-hood rating scale that is being used to a much smaller range and so the AI probably has more trouble distinguishing between players.

Your minor leagues have a mix of teams that are affiliated and unaffiliated. It has been a long time since I have used a setup like that and I don't remember the AI ever handling it particularly well.

[Dutch Alexander]

Originally Posted by Rain King

You have TCR jacked up, Batter Dev. Speed jacked up and Development Target Age set to "Younger". This is all seemingly geared towards young players progressing very quickly

Well, I am trying to make young players develop quicker. Looking at the best players in the league they seem to come up at around age 21. It is supposed to be a low level league were players develop a bit quicker. But the 16-17 year olds that are being promoted aren't talented, they're scrubs with next to no playing time in the minors. Regarding the TCR I've tried different settings, this is just the lastest. An attempt to have more scrubs turn into usefull players.

Originally Posted by Rain King

You also have player evaluation set to a large (65%) Ratings Weight and Scouting Report Updates set to "Rare" and Scouting Accuracy set to "Low".

Again, these are the latest settings, I've tried all kinds od AI eval settings. Low scouting accuracy and rare updates is to make it more challenging for myself.

Originally Posted by Rain King

So, I am thinking that what could be happening is that these young players actually do have major league quality ratings, but that the Scouted ratings just haven't caught up with that yet. It might be worth looking under the hood at any questionable players you are seeing on your end to see what their underlying ratings actually are. Or maybe it is the Low scouting accuracy and the teams think the players are much better than you are seeing (I think to check this you can act as the team and look at their scout evaluations).

No, I've checked these pitchers (allways pitchers) and they are really scrubs with no minor league experience. They produce stats like in the OP showing they have no business playing for the top team.

[Dutch Alexander]

Originally Posted by Rain King

Some other things I wonder if could be causing weirdness.

Your highest talent league has Player Creation Modifiers at .800. While the ratings are set to be relative to that league I still wonder if the game isn't expecting an environment where the talent is muted in that way. Essentially, I think this squeezes the portion of the under-the-hood rating scale that is being used to a much smaller range and so the AI probably has more trouble distinguishing between players.

I chose the PCM at .800 since the league is supposed to be a low level professional league. Off course with no other leagues in the universe this doesn't really a whole lot but I was hoping it would produce more outliers, more dominant players. Not sure if it does. It could indeed make it a bit more difficult for the AI to distinguish between players but I think it should be fine, .800 is not that low. The outliers I'm seeing now are extreme and must have another cause. For the most part the league is functioning fine.

Originally Posted by Rain King

Your minor leagues have a mix of teams that are affiliated and unaffiliated. It has been a long time since I have used a setup like that and I don't remember the AI ever handling it particularly well.

My setup is a bit unusual. I have 12 teams in the top league and 8 in the lower league. These two leagues do not have separate minors, all 20 minor league teams at each level play in a combined league. The highest level of minors is affiliated to the highest league and the lowest level of minors is affiliated to the lowest league. That means that for teams in the lowest league their highest minors teams will show as being unafiliated (and for teams in the highest league their lowest minors team). But these teams are still part of the organisation and function as such. If a team promotes then the game will recognize the minor league team is part of the organisation and it will show as being affiliated (and the other that was affiliated will now show as unaffiliated).

I know it is not intended to function this way, but I'm pretty sure it works. Players that are part of an organisation promote/demote between the two levels normally and I have not seen any issues. After promotion/demotion players remain in the organisation and everthing seems to function fine.
I'm quite sure this is unrelated to the problem.

[Rain King]

In your original screenshot, the player has been given 1.5 stars...which is legitimately a back of the bullpen overall rating.

My best guess is that this is some sort of bug/weirdness when your top-level league has low Player Creation Modifiers and you are using relative ratings. I would maybe try turning relative ratings off and doing a re-scout.

[PSUColonel]

Too many settings.

[Dutch Alexander]

Originally Posted by PSUColonel

Too many settings.

Exactly. It is great to have options, but there are just so many settings and we don't know precisely what will happen when we change them. It's basically trial and error, hit and hope.
As a smart man once said regarding the OOTP settings:

"Because as we know, there are known knowns; there are things we know we know.
We also know there are known unknowns; that is to say we know there are some things we do not know.
But there are also unknown unknowns—the ones we don't know we don't know."

[Dutch Alexander]

I ran another test with below settings

Scouting accuracy: LOW
Current Ratings: 2-8
Potential Ratings: 2-8
Other Ratings: 2-8
Overall/Potential Ratings: OFF
TCR: 200
AI Evaluation: 65/20/10/5
No roster limits for both minor leagues.

I simulated 12 years into the future and still saw the same issue. Progressed further another 8 years and the issue was all but gone. After that I simulated another 15 seasons, year by year and each year the youngest players that appeared in the league were ussually in their twenties. In the highest league the issue was basically gone. In the lower league there was an occasional iffy AI decision but usually for just a few games. It seems removing the roster limits on the minors seems to have helped. I think the AI is just very poor at roster management and with limits odd behaviour occurs. I don't really like to remove the limits as now teams sometimes have more than 50 players on their roster, but I guess I have no choice.

It seems that it takes quite some for the league to stabilise after the initial startup but once it is stable it remains so. Still more testing to be done.

[Rain King]

Part of the stabilizing issue comes from the way players are coming into your league.

1.) You aren't starting with initial rosters, but instead using "Fill Teams With Fictional Players" which will produce much lower quality on the initial rosters and essentially no true prospects. Basically that option creates replacement level players (based on your Creation Modifiers) for each level.

2.) All new players coming into your league are extremely young. While you have settings to speed up their development that still creates a talent "gap" and until you are past the place in your history where those missing gap players would be represented things aren't going to become completely stable.

I think that just means you need to sim longer to get things to a balanced state.

[Dutch Alexander]

Originally Posted by Rain King

I think that just means you need to sim longer to get things to a balanced state.

Yeah, I was aware of this but underestimated the time it takes a bit. Should've started the league in 1975. During tests I first focussed mainly on the stats output which stabilises after about 12 years. During other tests that ran for more than 100 years I still saw teenagers apearing on active rosters frequently. But during these tests I changed settings every few years. It seems I have to run the league with steady settings longer for it to stabilise.

In the last test I switched roster limits back on for the minors and it seems things remain stable. So roster limits seem not to have caused a real problem after all. Anyway thanks for your help and input. I will continue to run more tests. Very tricky and tedious to get eveything to work properly.

[Brad K]

Originally Posted by PSUColonel

Too many settings.

Yea. Well, perhaps someone could try them all and make notes and tell the devs what the settings do.

[jcard]

Originally Posted by PSUColonel

Too many settings.

Not to mention that (as I believe you have pointed out) the claim that you can just run the game out of the box on default settings and have a bona fide satisfactory experience is dubious. At some point all of the settings and options tacked on are at best ineffective and at worst counterproductive if you do not go in and really commit to changing the underlying code. Yearly additional dials and buttons make for great release note bullet points, but are they really improving the game at this juncture. I would argue in the negative, especially since so many seem to be orphaned from a optimization and fine-tuning perspective by a lack of sustained commitment of development resources.

[jcard]

Originally Posted by jcard

Not to mention that (as I believe you have pointed out) the claim that you can just run the game out of the box on default settings and have a bona fide satisfactory experience is dubious. At some point all of the settings and options tacked on are at best ineffective and at worst counterproductive if you do not go in and really commit to changing the underlying code. Yearly additional dials and buttons make for great release note bullet points, but are they really improving the game at this juncture. I would argue in the negative, especially since so many seem to be orphaned from a optimization and fine-tuning perspective by a lack of sustained commitment of development resources.

Actually, I just realized that I inadvertently hi-jacked this thread and want to apologize to the original poster. It was an honest mistake on my part, and I do hope that you and those helping you are able to find a satisfactory resolution to your concerns.

[OutS|der]

What happens when TCR is set to 0?

[Dutch Alexander]

Originally Posted by jcard

Actually, I just realized that I inadvertently hi-jacked this thread and want to apologize to the original poster. It was an honest mistake on my part, and I do hope that you and those helping you are able to find a satisfactory resolution to your concerns.

No worries, I don't mind at all.

[Dutch Alexander]

After some more testing it seems the main solution for my league is a very high TCR. With TCR at 200 I was able to turn scouting accuracy back to low with current rating off and potential and other ratings to the 1-5 scale. I also switched back to 25/25/25/25 eval settings. After this the league remained stable with only occasionally teenagers appearing on the active roster. When I turn TCR down to 125 more teenagers started to appear over time.

I guess the problem for my league is twofold. Firstly, it is a small league with a small talent pool. As a consequence there isn’t enough talent to fill all the roster spots. TCR at a high setting turn more scrubs into useful players to fill the roster spots. The second issue is the AI itself. It seems it does not take age and experience into account at all. A 16 year old scrub with 0 experience is viewed the same way as a 30 year old scrub with 10 years’ experience. At the same time the AI seems to prefer to hang onto prospects, even if they are scrubs. If an older slightly better scrub is available as a FA the AI hangs on to the 16 year old scrub. Hence 16 year olds start appearing on active rosters.

Also the AI is just very bad at roster management. In my league all prospects are under team control until their age 21 season. If the team wants to keep control over the player after that it needs to put him on the secondary roster. This means no player should EVER be put on the secondary roster before the end of their age 21 season until he is clearly ready and the team plans to use him as a regular. I think the AI is simply completely incapable of understanding this and thus makes ridiculous roster moves.

Now to be clear I'm not trying to put down the game. I understand it is incredibly hard to program a competent AI. Also the league I created is rather unusual which probably doesn't help. But the AI is certainly imperfect which means we need to find the right combination of settings to steer the AI in the right direction while understanding the final destination will never be reached. But it is just incredibly hard sometimes to find the correct combination. How do you figure out that a high TCR would improve things? And mind you only after 15-20 years into the league. I experimented with a TCR of 200 and simulated 10 years and saw no results and therefor dismissed it as a solution. Later by pure chance I found that a high TCR takes 15-20 years to have full effect. How can you expect anybody to figure stuff like that?

Anyway I seem to be on the right track now. Off course I might find later that a high TCR has other unwanted consequence. And in the newest update base stealing has been tweaked which means that I have to check and adjust my modifiers again first before I can proceed further. There is always something else, Why am I doing this?