Understanding Minor League Performance in OOTP 26

[jpeters1734]

FWIW, this is something that specifically changed with OOTP 26. Here's the same exact test in OOTP 25 and the player is rightfully terrible in AAA. This proves that this is an undocumented major change in the OOTP 26 sim engine and it's not just me being "upset with OOTPD for not meeting my personal expectations".

[MisterTidster]

If I were to speculate, I’d say that there probably is an algorithm that calculates the outcome in bulk rather individually for each player and each at bat… mostly to address performance issues so it doesn’t take 5-7 minutes to simulate each day. Probably works well overall but will produce wonky results processing statistical outliers, which is a shame, since half the fascination with this game as a player is hunting for statistical outliers that could give us an edge.


Makes me wonder how this formula is tied into other stat anomalies that I’ve noticed…

*High School and College players typically having decreasing production in High School or College year over year… often exaggerated if the player participates in a DL.

*Draft busts less than 22yo that achieve 1.5+ WAR in each of their first two seasons in A ball, but experience declining production at the same league level as their potential drops, even if their overall rises, and struggle to stay in the positive WAR.

*MLB veterans who lose 5 overall and potential in a stat, which then triggers a complete statistical tank rather than a slight decline in the affected stat.

[Guthrien]

It gives funny spin to the default of setting depth charts for each minor league lineup/staff by potential and not current ability.

This clearly is something funky with the new 26 engine, I don't think many people get what jpeters is saying and I sympathize with that. There are a lot of rough things in the game right now IMO. So much so I went back to 25, but we all care about different parts of the game.

I do respect what Lukas and others are trying to do, I don't enjoy the "you all shill for PT" talk. Sports games are really trapped between change something or break something, and it already is a pretty opaque and undocumented game outside of some dedicated work by fans (not a great thing). So many of the threads are competing theory and folklore about how you do this, or how this works. Then someone runs numbers and it turns out, oh wait, the game is designed, not some platonic shadow of the MLB. How exactly does development work? How was it designed to manage your minor leagues?

[panda234]

I have always used stats as a major consideration when promoting or demoting MiL players. Simple question—am I doing it wrong?

[whaleheader]

After conducting further testing and reviewing some of the posts, this appears to be a bug related to the expansion of the rating range from a maximum of 550 to 600 in OOTP 26.

This doesn't appear to impact players other than those with artificially unrealistic number gaps between current ratings and potential ratings over 550 IN THE MINORS. As others have noted, such players suck, as they should, in the majors.

[mytreds]

Originally Posted by Lukas Berger

What you're stating here might apply in a regular job, but doesn't really apply to baseball as such though.

Players viewed as having high potentials, are generally seen that way based on things like physical tools, which provide them with with an advantage compared to other players, especially at the lower levels of the game.

Often, players with raw ability can dominate at lower levels just based on those tools, even if their game remains raw overall.

Once they get to higher levels, especially MLB, then that advantage goes away, because the overall physical level is much higher. Many of those formerly highly regarded players, who have performed well at every level up until MLB then fail, because their actual abilities (current ratings) are not high enough and they can no longer depend on their physical tools (potentials).

This is a pretty well known phenomenon in most sports.

I think Billy Beane would disagree with these statements.

[mytreds]

Originally Posted by panda234

I have always used stats as a major consideration when promoting or demoting MiL players. Simple question—am I doing it wrong?

Tbf to the devs, I’ve always used ratings as a factor for my high rated prospects. Stats comes secondary, for me, when promoting. But ymmv. Minors stats are shiny things, and help me role play.

[whaleheader]

I tool the worst rated hitter in the game. I cloned him twice. One clone was set at 550 batting potentials; the other clone was set at 600 batting potentials. I turned off Development, set their strategies to force play, and I played them each one month at each level.

As I said, there is a bug.

[highandoutside]

Originally Posted by Lukas Berger

Not in the least. I'm saying that players with very loud potentials (tools), more or less get a boost to their current ratings in the minors.

That's it, that's all that's actually being discussed here once you cut to the chase.

This discussion doesn't even affect the vast majority of the players in OOTP. We're only talking about a very, very small subset of players with elite potentials here. The vast majority of of guys in OOTP who do not have stratospheric potentials will more or less play to their current ratings, even in the minors.

This a good discussion and I’m actually excited to read this from Lukas. It’s pretty cool approach to simulating the minors.

[jpeters1734]

Originally Posted by whaleheader

I tool the worst rated hitter in the game. I cloned him twice. One clone was set at 550 batting potentials; the other clone was set at 600 batting potentials. I turned off Development, set their strategies to force play, and I played them each one month at each level.

As I said, there is a bug.

It's a little hard to glean from your screenshots what exactly you're referring to. It would also be helpful if you turned off scouting so we could see the 100% ratings for each test.

However, it's not a bug. I've since learned this is an intended change in the engine for 26. Maxing out potentials just show the effect more clearly. The effect is still there with any rating.

[LansdowneSt]

I think it is worth considering that whatever this new behavior precisely is with regard to extremely higher potential prospects in the minors, it presumably has, in a normal game with actual players that the game creates, a good effect on how the game handles promotions, evaluates and values MiLB talent, etc. So being a purposeful change, it was likely done for a reason.

Here is the line from the Changelist for 26:

Updated minor league AI roster management, to be more aggressive at releasing blocking players, and to pay more attention to minor league placement for top talents

So, now I guess we know how they accomplished that.

If this mechanic has resulted in better AI promotions, less releases of top prospects, better evaluations of their talent when trading, smarter AI draft selections, is it worth it? I mean, I have no manually edited 1/600 players force locked to start in AAA. Do you normally? Does anyone? I made one and he got demoted down several levels until I locked him.

If it makes 99.999999999% of the customers have an AI that is smarter but bugs people because when they do this weird thing that never actually happens in the game, and they don't like it... isn't that a trade you'd make?

We can argue that because this didn't exist in OOTP 25, as jpeters showed, it should have been explained better in some documentation - (and I for one, never, ever feel like the one-liners on changes are sufficiently informative enough), but if OOTPD feels the AI is better at its job handling minor-league promotions and prospects, that's what matters. I could see them reversing this or overcorrecting this and getting worse AI across the minors in everyone's games - and why? So a 1/600 experiment yields expected pre-26 results?

It's true that a stats-only player doesn't have the same insight that we who have read this thread now have - that MiLB stats have some potential-driven extra oomph - is that bad? I'm a little annoyed I know this and it makes stats-only more tempting. Should I be concerned if I play with only stats that my next promotion to the majors would be less ready than I thought and might fail? Isn't that baseball? Are great MiLB stats truly a guarantee of MLB success? Has it ever been? Of course, he might not fail either. Does everyone want the game as predictable as that? I wish I didn't even know this was the "how" of that changelist adjustment. I now can't unsee it.

[whaleheader]

Nakamura was not touched.

Nakamura550's current ratings were set to 1; his potential ratings were set to 550.

Nakamura600's current ratings were set to 1; his potential ratings were set to 600.

I then played them at the same time and moved them up a level each month.

The first two performed as we'd expect. The third didn't. That suggests it's a bug.

I'm assuming nobody caught this because, and I haven't checked, newly created players don't exceed 550. They only raised the limit to catch the most extreme outliers, who'd only get assigned a rating above 550, if they were playing in the majors.

[LansdowneSt]

If the game doesn't create fictional players over 550 in potential and 551-600 is only for historical MLB extremes like Ruth vs contemporaries in the early 20's, then the bug is fixed by the 550 fictional player cap of 550 on potential (assuming that's a thing). Is that what the screenshots implied, that 1/550 was fine? EDIT: Because the changelist implies there is some of this (though less noticeable) all the way down the potential scale).

[whaleheader]

Yes, the 1/550 version of the player was fine. I included the screenshots so people can view the stats generated.

[jpeters1734]

Originally Posted by LansdowneSt

I think it is worth considering that whatever this new behavior precisely is with regard to extremely higher potential prospects in the minors, it presumably has, in a normal game with actual players that the game creates, a good effect on how the game handles promotions, evaluates and values MiLB talent, etc. So being a purposeful change, it was likely done for a reason.

Here is the line from the Changelist for 26:

Updated minor league AI roster management, to be more aggressive at releasing blocking players, and to pay more attention to minor league placement for top talents

So, now I guess we know how they accomplished that.

If this mechanic has resulted in better AI promotions, less releases of top prospects, better evaluations of their talent when trading, smarter AI draft selections, is it worth it? I mean, I have no manually edited 1/600 players force locked to start in AAA. Do you normally? Does anyone? I made one and he got demoted down several levels until I locked him.

If it makes 99.999999999% of the customers have an AI that is smarter but bugs people because when they do this weird thing that never actually happens in the game, and they don't like it... isn't that a trade you'd make?

We can argue that because this didn't exist in OOTP 25, as jpeters showed, it should have been explained better in some documentation - (and I for one, never, ever feel like the one-liners on changes are sufficiently informative enough), but if OOTPD feels the AI is better at its job handling minor-league promotions and prospects, that's what matters. I could see them reversing this or overcorrecting this and getting worse AI across the minors in everyone's games - and why? So a 1/600 experiment yields expected pre-26 results?

It's true that a stats-only player doesn't have the same insight that we who have read this thread now have - that MiLB stats have some potential-driven extra oomph - is that bad? I'm a little annoyed I know this and it makes stats-only more tempting. Should I be concerned if I play with only stats that my next promotion to the majors would be less ready than I thought and might fail? Isn't that baseball? Are great MiLB stats truly a guarantee of MLB success? Has it ever been? Of course, he might not fail either. Does everyone want the game as predictable as that? I wish I didn't even know this was the "how" of that changelist adjustment. I now can't unsee it.

Appreciate the thoughtful reply. I don’t disagree that improving AI behavior around prospect handling is a worthwhile goal, and if this adjustment helped with that, then fair enough. But I think two things can be true at once.

Yes, users are not locking 1/600 players in AAA. I think there’s too much focus on the test parameters and not enough on what the test revealed. If potential is now influencing minor league performance across the board, that changes what stats actually mean, even for players created normally in the game.

For those of us who rely on stats to supplement or even override ratings, this has real gameplay impact. I want to trust that on-field results reflect current ability. That’s always been the truth, and it’s not a small thing to change without telling users.

If this mechanic is necessary to support smarter AI, I get it. But I still think players deserve clarity about how the system works, even if it doesn’t make it into the changelog. That’s really what I’ve been pushing for.

[whaleheader]

Originally Posted by LansdowneSt

If the game doesn't create fictional players over 550 in potential and 551-600 is only for historical MLB extremes like Ruth vs contemporaries in the early 20's, then the bug is fixed by the 550 fictional player cap of 550 on potential (assuming that's a thing). Is that what the screenshots implied, that 1/550 was fine? EDIT: Because the changelist implies there is some of this (though less noticeable) all the way down the potential scale).

I would think, and the developers can tell us, that the change impacts the valuation of the players, not the statistical output.

[jpeters1734]

Originally Posted by LansdowneSt

If the game doesn't create fictional players over 550 in potential and 551-600 is only for historical MLB extremes like Ruth vs contemporaries in the early 20's, then the bug is fixed by the 550 fictional player cap of 550 on potential (assuming that's a thing). Is that what the screenshots implied, that 1/550 was fine? EDIT: Because the changelist implies there is some of this (though less noticeable) all the way down the potential scale).

FWIW, the Dev Lab can push potentials above 550 with outstanding completions

[Brad K]

Originally Posted by Lukas Berger

I don't think running a test with parameters that would never naturally occur in-game provides any particularly useful information.

Given that the game would never create a player with a 1 current rating and a 550 potential, this is basically just testing whether entering extreme values manually can break the engine, which I don't think anyone should be surprised that it can.

Products are commonly tested under extreme conditions unlikely to be encountered real life. A failure at the extremes can reveal less than optimal performance under moderate conditions.

https://www.youtube.com/watch?v=a0SSLnwVRwA

[Brad K]

Potential is not current.

[mr-cpu-geek]

Maybe I’m the minority, but I have to agree with Lukas. I really don’t see the point of these extreme test. You are testing something that will not happen organically in the game.


Sent from my iPhone using Tapatalk