Understanding Minor League Performance in OOTP 26

[OutS|der]

Originally Posted by Zooerp

I'm not too concerned about the output for scenarios that won't occur in the game

Are we not able to create our out leagues and players? So these are scenarios that would occur in the game cause not everyone is playing real MLB, but I guess those of us are just not important because we play with scenarios that don't occur within MLB.


People get so hung up with using a 1 as a rating, as if the number even matters. Make it a 100 or any number that's drastically lower then everyone else at that level. That player should struggle and the fact they aren't is game breaking.


I'm glad I didn't upgrade to 26

[Eckstein 4 Prez]

I am just reading this now, and first of all I'm really glad my online leagues are still on 25, because I look at minor league stats to decide when to move guys from one level to the next.

First, I think there's a fundamental disconnect here with what is meant by "current ratings" and "potential ratings." Let's imagine two extreme players. One is a distant cousin of Willie Mays, but his ancestors weren't enslaved, so he's living in Africa, never having played baseball, but moves to the States at age 18 and finds that he enjoys it. This guy has excellent, maybe superstar potential ratings but his current ratings should be terrible - he has no experience at all. That's how I understand it.

Meanwhile, a second guy has lived, eaten and breathed baseball his entire life. He comes into the draft with a very polished game, but it also turns out that he's pretty close to his ceiling - maybe his "current ratings" are good enough to be a borderline big leaguer but his "potential ratings" aren't really any higher.

Seems to me the second guy could probably hold his own and do well in the high minor leagues, while the first guy should be awful even in the low minors and the question is whether he's able to develop quickly enough to get to that lofty potential.

But the point is that, to me, "potential" ratings just show a person's ceiling - nothing more. "Current" ratings show what that person is capable of now, and I thought those would be the only things ever used to determine performance. Some people here have objected that guys with high ceilings should dominate the low minors, and no doubt that is true. But they should do so because their ratings should move more quickly than others, as their current performance starts to approach their potential.

And of course, I think the real reason people are upset is because an element has been introduced that runs counter to baseball knowledge that's so fundamental that Bill James had basically proven it in the 1990s: there's nothing qualitatively different between the minor leagues and the major leagues. There was this persistent idea until that point that some guys just didn't have the mental makeup or whatever to thrive at the highest level. James proved mathematically that this was nonsense - once you were able to quantify the varying levels of competition, players performed just like they were supposed to.

Introducing an element that applies to performance in the minor leagues but not in the major leagues creates two problems. One is with OOTP's game engine as a baseball simulator: there's nothing qualitatively different between how a player should perform in the minor leagues versus the major leagues once you factor in the changing talent level. Creating some artificial thing where "guys just need time to figure it out" smacks of the bad old days, when players would have "clutch" or "RBI" ratings - it's not how the game works, and this has been understood mathematically since around OOTP1. Adding this stuff in makes the engine less realistic, and it gets people worried because it suggests a willingness to put in intangible hoodoo rather than having mathematical-based simulations that people can rely on.

The second problem is as a gameplay element. Once you've played it for a while, solo OOTP is pretty easy when you play with ratings on. I tend to either turn them off entirely or only have very rudimentary ratings, because otherwise there's not a ton of challenge. Most solo players who've been at it for a few decades that I know do the same thing. As a matter of gameplay, this makes it impossible to do anything like this because when looking at minor league stats, I now know that this is done through the "minor league engine" and not the regular one, and that it's taking different things into account. So I have to turn everything on to have any confidence in the game, and now I'm back to the point where the game is too easy. I want a challenge, but I don't want the challenge to be arbitrary and opaque as a matter of gameplay - this is supposed to be baseball, not Calvinball.

Finally, I think the biggest concern people have is that this fundamental shift - adding weird intangible factors to certain parts of the engine - was just thrown in there without any discussion and this was only discovered by a person running tests. Honestly, I think a lot of people in this thread are doing him a disservice - while he has been pretty strident in this thread, he has uncovered something that is (a) extremely important and that (b) was just put live into the sim engine without any discussion or notification. We owe him a debt of gratitude, and if he's shouting it from the proverbial rooftops, it's understandable - imagine that feeling he must have had when running that test and realizing there was something not right happening.

When I first started playing OOTP (version 4!), there was a clutch rating, pitchers had an "avoid hits" rating that was pretty fundamental to their skill, etc. Markus didn't grow up steeped in baseball lore or research, but one of the best things about him was his interest in learning those things and trying to add them to the game. His change to the DIPS-based engine in the mid-2000s was pretty cutting edge for the time and it's a big part of what helped OOTP win the battle of the baseball sims, in my opinion.

I have to say that doing something to the game engine that makes the minor league simulations happen in a way that's different from the major league engine feels to me like a big step back, back to the discredited "there's something different about the big leagues" school of thought that was debunked decades ago. It's also a very bad time to do something like this - every longtime player I know is very nervous about what the first non-Markus version would be like, and everyone also knows that Perfect Team is the revenue engine for the game, so there's concern that single-player OOTP will become something of an afterthought. This certainly isn't going to do anything to make me worry about that any less.

Finally, if these critical posts cause the discussion to be brought to an end, I'll be even more worried. The most optimistic thing I can say about this discovery is that we have developers in the thread who've been willing to engage. I appreciate that, and I think it's very important for the future of the game for this to be a discussion, where the wisdom of the community is used as a feedback mechanism. I don't think this change accurately reflects the best research about baseball skills, but really the worst thing about it is that it was simply added in without any discussion ahead of time. The best thing is that this discussion is taking place now. People here really care about making sure the sim engine is as realistic as possible, and also that it provides the best and most realistic gameplay experience. For the reasons I've explained, I think this fails on both fronts, and I hope we can continue the productive discussion about it.

[Brad K]

Originally Posted by Eckstein 4 Prez

Finally, I think the biggest concern people have is that this fundamental shift - adding weird intangible factors to certain parts of the engine - was just thrown in there without any discussion and this was only discovered by a person running tests.

I'm sure it was discussed with the customers who are close to the devs, those who help with the game. However they are not a representative sample of the customers.

[Paul Reuschel's Mustache]

Originally Posted by Eckstein 4 Prez

But the point is that, to me, "potential" ratings just show a person's ceiling - nothing more. "Current" ratings show what that person is capable of now, and I thought those would be the only things ever used to determine performance.

I think the biggest concern people have is that this fundamental shift - adding weird intangible factors to certain parts of the engine - was just thrown in there without any discussion and this was only discovered by a person running tests.

Every longtime player I know is very nervous about what the first non-Markus version would be like, and everyone also knows that Perfect Team is the revenue engine for the game, so there's concern that single-player OOTP will become something of an afterthought. This certainly isn't going to do anything to make me worry about that any less.

I don't think this change accurately reflects the best research about baseball skills, but really the worst thing about it is that it was simply added in without any discussion ahead of time. The best thing is that this discussion is taking place now.

People here really care about making sure the sim engine is as realistic as possible, and also that it provides the best and most realistic gameplay experience. For the reasons I've explained, I think this fails on both fronts, and I hope we can continue the productive discussion about it.

[kidd_05_u2]

Originally Posted by Brad K

I'm sure it was discussed with the customers who are close to the devs, those who help with the game.

No it wasn't.

[Zooerp]

Originally Posted by OutS|der

Are we not able to create our out leagues and players? So these are scenarios that would occur in the game cause not everyone is playing real MLB, but I guess those of us are just not important because we play with scenarios that don't occur within MLB.


People get so hung up with using a 1 as a rating, as if the number even matters. Make it a 100 or any number that's drastically lower then everyone else at that level. That player should struggle and the fact they aren't is game breaking.


I'm glad I didn't upgrade to 26

If you haven't played the game you're not really in a position to determine if this is game breaking or not. 26 plays better than any OOTP I've played. If you have a type of save/style/scenario within 26 where this truly breaks the game, I'm sure people would be open to hearing about it.

Where success comes from in athletics, especially at lower levels, is extremely fuzzy and there are an infinite number of variables that no sim can truly replicate. Numbers representing "Current ability" and "Potential ability" are at the end of the day fictional abstractions. If you could actually put a true number that showed someone's exact "Current ability", it would be a constantly moving number within a range from game to game, at bat to at bat, moment to moment.

[matskralc!]

This is a pretty awful change, and the number of posters that don't seem to understand jpeters' point is frustrating. The amount of "OK, but nobody actually locks 1/600 prospects to AAA, so what's the problem?" is agonizing.

The development team's argument that "well, raw athleticism is represented by potential ratings and not accounted for in current ratings" is nonsensical. Current ratings are "ability to apply raw athleticism to the task". The development team's argument boils down to "people are naturally better at something that they might be naturally better at". If this is the case, then just reflect it in their current ratings.

Which is exactly what kidd_05_u2 argues:

Originally Posted by kidd_05_u2

Wouldn't career paths like Campbell or Jackson Holliday be modeled better as coming out of the draft with current ratings that are good enough to play at AAA? Good enough to tear through the minors and not good enough to easily succeed in the MLB, while avoiding this bad situation where potential affects stats.

The challenge would be to get the AI to still put some of these players through the low minors, though nowadays teams are getting more and more aggressive with early promotions.

To which RonCo responds:

Originally Posted by RonCo

Yes. This is exactly how those players should be modeled. Some players blaze through the minors because someone made a "mistake" and put them into an environment where their current skills were considerably greater than the guys around them. They only stabilize once they are put into an environment where they are at the right level--which could even be the majors.

The idea of using potential ratings to influence current results is a bad model.​

Yes! Exactly! Top prospects don't perform better in the minor leagues because they might be good major leaguers someday. They perform better in the minor leagues because they're better at baseball than the other minor leaguers they are playing against. It just boggles my mind that the dev team doesn't see this.

fhomess elaborates:

Originally Posted by fhomess

I think this is where OOTP is not yet agreeing. I'm hearing a "ends justify the means" response from OOTP on this. However, I think the issue is that the model is bad because it confuses the end user of the game. Stats should be generated according to the same ratings at all levels of play. If we're using potential ratings in conjunction with current ratings for minor leagues but not major leagues then they're playing two different games. The end user doesn't know that, and can't be expected to logically assume that, either. It's entirely counterintuitive.

Perhaps the issue that OOTP is trying to model is the wider variance of expected outcomes that prospects have based on their skill level which OOTP doesn't really have a rating for.​

I find it interesting that RonCo, who I recognize as "that guy who did countless statistical studies of the OOTP defense model", and fhomess, who I recognize as "that guy who gave us StatsLab", are on the side of "this is a poor model".

The actual problem is that almost nobody ever comes out of the amateur draft with the current ratings to go anywhere higher than Low-A. The fact that real-life teams send everybody to R/A- and the good players tear the cover off the ball is because those good players should have 40s, 45s or 50s, not 30s, and probably could have been sent right to A or AA. It's not because they have potentials of 70+ that somehow compensate for their actual 30. Real-life teams don't send all of these guys straight to A or AA for a variety of reasons, many of which have nothing to do with ability and everything to do with things like personality, maturity, and organizational culture.

But now? When current ratings don't matter? What level do I send a guy to after the draft? Who knows! Because his ratings aren't his ratings! And then I can't trust his statistical output because it turns out that his potentials are being factored in to those, but only in the minor leagues, and if he's got good potentials he's going to perform well regardless of his actual ability and I'm only going to discover his actual ability once I place him on the big league roster.

Eckstein 4 Prez then shows up to make the point, with lots of words!, that Bill James disproved the "there's something different about the Bigs" nonsense decades ago. So why are we bringing it back?!?

I already don't use minor leagues in my solo leagues anymore because it's not possible to tune the totals to be what I want without running 474,000 test sims first (and because players develop regardless of playing time; playing time is really only for getting additional information through stats, "rotting on the bench" or "giving a guy reps to figure it out" is sadly just not much of a thing).

The two online leagues I participate in, the OTBL and the TCBA, are broadly OK looking enough that their minor league totals don't bother me. But I heavily use minor league stats when evaluating whether it's time to push players up another level. The fact that minor league stats are almost completely meaningless in OOTP26 is enough for me to say that the OTBL, the league I run, may be on OOTP25 forever. And we may be a drop in the bucket, but a lot of the ownership in that league only buys new versions of OOTP when I upgrade the OTBL.

[matskralc!]

Originally Posted by matskralc!

Yes! Exactly! Top prospects don't perform better in the minor leagues because they might be good major leaguers someday. They perform better in the minor leagues because they're better at baseball than the other minor leaguers they are playing against. It just boggles my mind that the dev team doesn't see this.

Like, the guys who have historically torn through the minor leagues and flamed out in the major leagues are guys who should be coming out of the draft with 45/50/55 current ratings across the board and potential ratings that aren't much higher.

And then what we do is having the scouting model mis-interpret the player's current ratings and amateur stats and physical attributes and assign him higher potential ratings than he really has. Now you have your Chad Hermansens who come out of the draft as a 45/65, fly through the minors because they're better than everybody they're actually playing against, OPS .850+ as a 20-21 year old in AAA, and perform like a 45/45 in the majors because it turns out the scouts got it wrong and "guy who's better than everybody in the minors" was his ceiling.

At no point do we need to factor in a player's real potential ratings in order to create the mirage of players who progress quickly through the minors and then flame out in the majors.

[Eckstein 4 Prez]

Originally Posted by matskralc!

Like, the guys who have historically torn through the minor leagues and flamed out in the major leagues are guys who should be coming out of the draft with 45/50/55 current ratings across the board and potential ratings that aren't much higher.

And then what we do is having the scouting model mis-interpret the player's current ratings and amateur stats and physical attributes and assign him higher potential ratings than he really has. Now you have your Chad Hermansens who come out of the draft as a 45/65, fly through the minors because they're better than everybody they're actually playing against, OPS .850+ as a 20-21 year old in AAA, and perform like a 45/45 in the majors because it turns out the scouts got it wrong and "guy who's better than everybody in the minors" was his ceiling.

At no point do we need to factor in a player's real potential ratings in order to create the mirage of players who progress quickly through the minors and then flame out in the majors.

Right! If you want diversity in the draft class, there should be more guys who are 21-22 and coming out of college perfectly able to hold their own in AA, but whose ceiling might not be much higher. And those draft misses should come because a team's scout misinterprets the 45/45 guy who can already play at a high level as a 45/65, and the team is frustrated because the guy never seems to get any better.

I'm certainly not arguing that the game has reached some platonic ideal of baseball and the sim engine can't be improved. This is an example of an improvement that can be made. What I am saying is that an element that has nothing to do with a player's current skill has been introduced and it makes a difference to the player's outcome in the minor leagues.

This reminds me so much of the debate in the early 2000s on these very boards about having a "clutch" rating, right down to the main arguments in favor being "I played the game and you have to understand that this is a real thing," "the game creates better outcomes than it used to so why are people complaining," and "this really isn't that big of an effect so what are people up in arms about."

I'm still friends with the old school forum crowd from those days - we're in an online league together that's been going since...2006? 2007? I don't think it's possible for me to overstate the level of dismay this creates with those folks, who are basically saying "well, OOTP25 is a good product and I can live with just using that if they're going to break the sim engine by adding nonsense into the calculations." Personally I don't feel that strongly about it, and might still get new versions for the solo experience. But we rely on how our guys are doing in the minor leagues in our (quite competitive!) online league, and I don't think I'm going out on too much of a limb to say it's unlikely that league will ever upgrade past OOTP25 unless this is taken out. You can ask mats though - he's our commissioner.

[Paul Reuschel's Mustache]

Originally Posted by Eckstein 4 Prez

What I am saying is that an element that has nothing to do with a player's current skill has been introduced and it makes a difference to the player's outcome in the minor leagues.

...we're in an online league together that's been going since...2006? 2007? I don't think it's possible for me to overstate the level of dismay this creates with those folks, who are basically saying "well, OOTP25 is a good product and I can live with just using that if they're going to break the sim engine by adding nonsense into the calculations."

[Eckstein 4 Prez]

I think this was all probably a way to try to shortcut a real problem, which is that the game really doesn't seem to create a lot of players who look like they have the talent level to play in the big leagues but they crash out for whatever reason.

I don't know anything about creating sim engines or coding or whatever - I'm just a simple guy who bull****s for a living. But it seems to me that if you are creating, say, 800 players for the draft every year, you want maybe 1 in 100 to have a talent level at creation that would make them among the best in the league if they achieved it, and maybe you also want around 1 in 100 players who are already at or near their talent levels at creation. (And of course, you'd have many players who are far below their talent levels, some of whom will never reach that ceiling.)

That would mean that, say, one generated player in 10,000 would be created at a level where they were already almost ready to be one of the best players out there, which implies that it would happen once every 12 years or so. That's historically accurate-ish. (In my lifetime: Mike Trout, Ken Griffey, Stephen Strasburg, Alex Rodriguez, a few others if I broaden the definition a bit.)

Now I just made up those rates, but my point is that you can tinker with this stuff until you get realistic results by using the current/potential ratings system as they previously existed and were commonly understood. As is true with most situations where people take shortcuts, it ends up a muddled mess. As the developers are now realizing, this game does not have the kind of fanbase where people are going to go "oh well, the overall sim numbers look decent, no need to drill down at too much of a granular level."

[sixto]

I tried to make this short, I really did.

I've been playing OOTP long enough to remember the bugs of the OOTP4-6 days, when after 40 years of simming, minor leaguers couldn't hit over .130 because the new pitchers the game generated were too good. Or was that Baseball Mogul 2000? Anyway. Never played college baseball though.

Let's go back to basics: you are a MLB scout, evaluating an amateur player. Traditionally, you do this on the 20-80 scale, but let's use 1-600. In real life, ratings above the midpoint are rare, and reserved for the best prospects, except when rating foot speed and, for pitchers, fastballs, where max ratings are sometimes given. It does not work this way in OOTP, but I'm trying not to digress too much. It's a struggle.

Long story short: the vast majority of players who are candidates for organized ball would have potential ratings of just 200-300 on a 600 scale. That is how you are supposed to account for double-super stars like Babe Ruth and Pedro Martinez and early Mike Trout on a 600 scale: a 550 potential hit tool should almost never happen. You absolutely shouldn't have so many players with ratings over 400 that you now need a double booster from the potential ratings to differentiate.

This has always been a stupendous challenge for OOTP, maybe the central challenge for gameplay: how do you make a video game interesting when most of the players — the key to immersion in a sim — ought to be interchangeable? (We'll come back to that.)

(Worth noting for a sec: if a scouted player can't manage to get at least to an average of 200 in potential ratings, he probably isn't getting into organized ball. That's relevant to the discussion, but will also muddy the waters right now.)

It's absolutely critical to note that these scout ratings — potential ratings — are 100% not data. They are observations. Opinions! I know some will try, but IMO this is not arguable. Bat speed is data. Arm speed, arm angle, home-to-first time, average angle of attack, spin rate, barrel rate: all data. 500 potential power rating? Opinion.

Therefore, at the heart of what we are discussing is this: if minor league baseball players' statistics are influenced by potential ratings, then in fact they are influenced not by data but by somebody's opinions. That is, because Johnny Q. Scout believes that Hubert H. Homerhitter has the potential to hit 40 HR in the majors someday, somehow Johnny Q. Scout's opinion actually manifests in a boost in performance for Hubert in AAA. But not in the majors, where Johnny Q. Scout's opinions suddenly mean nothing. That's what I've gleaned from this discussion.

Rather than thinking of potential ratings and current ratings as two separate ratings, I recommend visualizing them as two circles, one inside the other. Each current ratings (C1, C2, C3 etc for every separate rating) you can think of as being the smaller ball inside the larger ball with the P1, P2, P3 ratings etc. So, a player with a 100 current rating and a 500 potential rating would have a small ball inside a large ball. Room for improvement. A player with a 100 current rating inside a 250 potential rating would have a small ball inside a slightly larger ball. Not much room for improvement — but perhaps, hard to tell a difference in their performances for now.

When we think of two players at AAA, one with a 100 C and 250 P and one with a 100C and 500 P, it's easy to understand why they should perform more or less exactly the same: because they have the same current rating. However, the better player's C ball should fill up more of the P ball as he plays, gains experience, gains skill. As the C balls get larger within the P balls, we can visualize the player living up to his potential, maybe even exceeding it. There is no need to increase the size of the P ball. Because some players do exceed their potential ratings, which again, are opinions and not data. When Justin Turner suddenly became a .300 hitter, nobody said, "His contact potential has gone from a 5 to a 9!" They said, "Holy cow Justin Turner can rake now, guess we were all wrong!"

The problem with OOTP's game engine, and I've thought this for years, isn't one that needs to use opinions to boost facts. The problems could be solved with just two tweaks. Well, three.

1. If a player A has more potential than player B, regardless of current ability, then if the potential rating is accurate, his current ability ratings should go up much, much faster (and more measurably!) than they currently do in OOTP. This would completely eradicate the need to self-hack it with a potential boost.

2. The way players are rated needs to be completely overhauled so that the vast majority of players are in the bottom half of the ratings scale, as they are in real life. Ratings over 400 and especially over 500 need to be reserved for the true superstars. But here's the great thing: this would make it infinitely easier for users to reliably create superstars in their games. As a fictional simmer 4 life, I would be so, so, so happy if I could create a few Jimmie Foxxes who actually play like Jimmie Foxx and not Andy Fox.

That brings us, unfortunately, to the third tweak.

3. Some genius has to figure out how to create a game where the code doesn't just see the vast majority of players as essentially replacement-level players who really could be interchanged one with the other. I have no proof, but I believe this is the problem that probably vexed Markus more than any other ("OOTP only knows numbers not names" IYKYK.) I think he did a truly fine job of faking it — but it's not perfect. And if OOTP were to implement the recommendations above, this problem would come roaring back. I don't have an answer for it.

This is all an even bigger issue, by the way, for players like me who like to simulate earlier times in history. I suspect some or many of the people who see this engine change as negative are historical and fictional simmers, with those not bothered being largely modern day users.

Nowadays, and with MLB awash in data, you can argue that the vast majority of players are kind of treated as fungible. Roster turnover has never been as high as it is today. 80, 100 years ago, you had 15 guys at the start of the season, and chances are you had a good chunk of the same guys at the end of it, even the sucky ones. Without data to guide them, teams relied on managers who did everything on feel and reputation and favoritism. That's also hard gameplay to model — and completely different gameplay from modern times.

[OutS|der]

Originally Posted by Zooerp

If you have a type of save/style/scenario within 26 where this truly breaks the game, I'm sure people would be open to hearing about it.

I have taken the past Cooperstown All Time Greats, Negro and The League that can't be named all time greats, played around with their ratings and now have a whole bunch with overall ratings set to 1 with all their potentials set for their peaks. These players get dumped into a HS/College setup before the draft.
IE The Josh Gibson has a HR potential of 600 yet current HR ability is 20.

The idea is for these players to develop through HS/College and see how many turn out to be who they were in history.

This is obviously not doable now in 26 since they would all have inflated MILB stats and I'd have no clue when to promote

[Paul Reuschel's Mustache]

Originally Posted by sixto

Long story short: the vast majority of players who are candidates for organized ball would have potential ratings of just 200-300 on a 600 scale. That is how you are supposed to account for double-super stars like Babe Ruth and Pedro Martinez and early Mike Trout on a 600 scale: a 550 potential hit tool should almost never happen. You absolutely shouldn't have so many players with ratings over 400 that you now need a double booster from the potential ratings to differentiate.

Therefore, at the heart of what we are discussing is this: if minor league baseball players' statistics are influenced by potential ratings, then in fact they are influenced not by data but by somebody's opinions. That is, because Johnny Q. Scout believes that Hubert H. Homerhitter has the potential to hit 40 HR in the majors someday, somehow Johnny Q. Scout's opinion actually manifests in a boost in performance for Hubert in AAA. But not in the majors, where Johnny Q. Scout's opinions suddenly mean nothing. That's what I've gleaned from this discussion.

The problem with OOTP's game engine, and I've thought this for years, isn't one that needs to use opinions to boost facts. The problems could be solved with just two tweaks. Well, three.

1. If a player A has more potential than player B, regardless of current ability, then if the potential rating is accurate, his current ability ratings should go up much, much faster (and more measurably!) than they currently do in OOTP. This would completely eradicate the need to self-hack it with a potential boost.

2. The way players are rated needs to be completely overhauled so that the vast majority of players are in the bottom half of the ratings scale, as they are in real life. Ratings over 400 and especially over 500 need to be reserved for the true superstars. But here's the great thing: this would make it infinitely easier for users to reliably create superstars in their games. As a fictional simmer 4 life, I would be so, so, so happy if I could create a few Jimmie Foxxes who actually play like Jimmie Foxx and not Andy Fox.

That brings us, unfortunately, to the third tweak.

3. Some genius has to figure out how to create a game where the code doesn't just see the vast majority of players as essentially replacement-level players who really could be interchanged one with the other. I have no proof, but I believe this is the problem that probably vexed Markus more than any other ("OOTP only knows numbers not names" IYKYK.) I think he did a truly fine job of faking it — but it's not perfect. And if OOTP were to implement the recommendations above, this problem would come roaring back. I don't have an answer for it.

Thank you for that. I read all of it twice and it makes perfect sense. I'd be interested in hearing what Matt and/or Lukas think about your comments.

[OutS|der]

Originally Posted by sixto

I tried to make this short, I really did.

I've been playing OOTP long enough to remember the bugs of the OOTP4-6 days, when after 40 years of simming, minor leaguers couldn't hit over .130 because the new pitchers the game generated were too good. Or was that Baseball Mogul 2000? Anyway. Never played college baseball though.

Let's go back to basics: you are a MLB scout, evaluating an amateur player. Traditionally, you do this on the 20-80 scale, but let's use 1-600. In real life, ratings above the midpoint are rare, and reserved for the best prospects, except when rating foot speed and, for pitchers, fastballs, where max ratings are sometimes given. It does not work this way in OOTP, but I'm trying not to digress too much. It's a struggle.

Long story short: the vast majority of players who are candidates for organized ball would have potential ratings of just 200-300 on a 600 scale. That is how you are supposed to account for double-super stars like Babe Ruth and Pedro Martinez and early Mike Trout on a 600 scale: a 550 potential hit tool should almost never happen. You absolutely shouldn't have so many players with ratings over 400 that you now need a double booster from the potential ratings to differentiate.

This has always been a stupendous challenge for OOTP, maybe the central challenge for gameplay: how do you make a video game interesting when most of the players — the key to immersion in a sim — ought to be interchangeable? (We'll come back to that.)

(Worth noting for a sec: if a scouted player can't manage to get at least to an average of 200 in potential ratings, he probably isn't getting into organized ball. That's relevant to the discussion, but will also muddy the waters right now.)

It's absolutely critical to note that these scout ratings — potential ratings — are 100% not data. They are observations. Opinions! I know some will try, but IMO this is not arguable. Bat speed is data. Arm speed, arm angle, home-to-first time, average angle of attack, spin rate, barrel rate: all data. 500 potential power rating? Opinion.

Therefore, at the heart of what we are discussing is this: if minor league baseball players' statistics are influenced by potential ratings, then in fact they are influenced not by data but by somebody's opinions. That is, because Johnny Q. Scout believes that Hubert H. Homerhitter has the potential to hit 40 HR in the majors someday, somehow Johnny Q. Scout's opinion actually manifests in a boost in performance for Hubert in AAA. But not in the majors, where Johnny Q. Scout's opinions suddenly mean nothing. That's what I've gleaned from this discussion.

Rather than thinking of potential ratings and current ratings as two separate ratings, I recommend visualizing them as two circles, one inside the other. Each current ratings (C1, C2, C3 etc for every separate rating) you can think of as being the smaller ball inside the larger ball with the P1, P2, P3 ratings etc. So, a player with a 100 current rating and a 500 potential rating would have a small ball inside a large ball. Room for improvement. A player with a 100 current rating inside a 250 potential rating would have a small ball inside a slightly larger ball. Not much room for improvement — but perhaps, hard to tell a difference in their performances for now.

When we think of two players at AAA, one with a 100 C and 250 P and one with a 100C and 500 P, it's easy to understand why they should perform more or less exactly the same: because they have the same current rating. However, the better player's C ball should fill up more of the P ball as he plays, gains experience, gains skill. As the C balls get larger within the P balls, we can visualize the player living up to his potential, maybe even exceeding it. There is no need to increase the size of the P ball. Because some players do exceed their potential ratings, which again, are opinions and not data. When Justin Turner suddenly became a .300 hitter, nobody said, "His contact potential has gone from a 5 to a 9!" They said, "Holy cow Justin Turner can rake now, guess we were all wrong!"

The problem with OOTP's game engine, and I've thought this for years, isn't one that needs to use opinions to boost facts. The problems could be solved with just two tweaks. Well, three.

1. If a player A has more potential than player B, regardless of current ability, then if the potential rating is accurate, his current ability ratings should go up much, much faster (and more measurably!) than they currently do in OOTP. This would completely eradicate the need to self-hack it with a potential boost.

2. The way players are rated needs to be completely overhauled so that the vast majority of players are in the bottom half of the ratings scale, as they are in real life. Ratings over 400 and especially over 500 need to be reserved for the true superstars. But here's the great thing: this would make it infinitely easier for users to reliably create superstars in their games. As a fictional simmer 4 life, I would be so, so, so happy if I could create a few Jimmie Foxxes who actually play like Jimmie Foxx and not Andy Fox.

That brings us, unfortunately, to the third tweak.

3. Some genius has to figure out how to create a game where the code doesn't just see the vast majority of players as essentially replacement-level players who really could be interchanged one with the other. I have no proof, but I believe this is the problem that probably vexed Markus more than any other ("OOTP only knows numbers not names" IYKYK.) I think he did a truly fine job of faking it — but it's not perfect. And if OOTP were to implement the recommendations above, this problem would come roaring back. I don't have an answer for it.

This is all an even bigger issue, by the way, for players like me who like to simulate earlier times in history. I suspect some or many of the people who see this engine change as negative are historical and fictional simmers, with those not bothered being largely modern day users.

Nowadays, and with MLB awash in data, you can argue that the vast majority of players are kind of treated as fungible. Roster turnover has never been as high as it is today. 80, 100 years ago, you had 15 guys at the start of the season, and chances are you had a good chunk of the same guys at the end of it, even the sucky ones. Without data to guide them, teams relied on managers who did everything on feel and reputation and favoritism. That's also hard gameplay to model — and completely different gameplay from modern times.

I couldn’t agree more with everything said here.

[Paul Reuschel's Mustache]

Originally Posted by jpeters1734

I want to share something I’ve been noticing in OOTP 26 that has real implications for how we evaluate minor league performance.

What I found is that potential ratings now influence statistical output in the minors, The higher the potential, the greater the effect.

...players with high potential consistently produce better minor league stats than players with equal current ratings but lower ceilings. That’s new in OOTP 26. In previous versions, performance was more clearly driven by current ratings alone.

Originally Posted by Lukas Berger

The whole reason for some of the complaints in the thread is that we've actually made major changes to the game engine in the last few versions to make the engine more expansive and more realistic in areas like expanding out the whole behind the scenes ratings scale to 600 (which was responsible for some of the hiccups around development and some other engine areas the last year or two, which caused some frustration).

Originally Posted by Lukas Berger

The basic principle here though, that two players with equal current ratings and greatly divergent potential ratings, that the guy with the very high potential ratings would get a slight bump to current performance, I think generally makes sense.

In real-life, top prospects do not normally put up .150/.180/.220 lines anywhere, whereas lesser players might, even if in theory both should have somewhat similar current ratings (by default of being assigned to the same level, or by the lesser prospect even being significantly older than the top prospect).

Originally Posted by Lukas Berger

The question of whether top prospects should get a slight boost to current performance is a valid one to debate given that they do in fact perform better in general in reality.

Your contention that this is because of their current skills solely is certainly valid, but is also not the only possible explanation, as even MLB scouts and personnel would likely debate the exact role of skills versus tools here, where a top player can get by on raw physical ability at lower levels, but get exposed once reaching MLB.

Originally Posted by Lukas Berger

As far as the purpose of minor league stats though, I think you're trying to use these for something that even real life minor league stats cannot be solely used for, to gauge MLB readiness.

Originally Posted by jpeters1734

I think you’re shifting the discussion away from what the actual problem is.

This is about whether minor league stats in OOTP reflect actual current ability. And what these tests show repeatedly is that they don’t.

You’re saying this is “just like real life,” but it’s not. In real life, a guy who hits .300 in AAA has real tools, even if they don’t carry over. In OOTP, a guy with no tools at all is hitting .300 just because he might someday develop them.

If the engine is giving players a performance bump because of potential, then it’s not simulating baseball. It’s simulating projection models. That’s fine for development curves. It’s not fine for in-game performance.

Originally Posted by Lukas Berger

Ultimately, we probably have a fundamental difference in our conception of how this should work, which is fine.

Potential ratings are tools, current ratings are ability. The fact is, that players in real-life with loud tools (potential ratings) often play over their ability (current ratings) in the minors and only get exposed in MLB.

So then, what's happening here is that the game is really just simulating that part of reality, that loud tools do help players perform better, up to a certain point.

Originally Posted by LD84

I cannot believe what I am reading

So Lukas
You are pretty much stating that current ratings should have no bearing on minor league performance but should for major league performance?

That is what I am taking away from your responses.

If so, then I have no words.

Originally Posted by Lukas Berger;5196742

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.

Originally Posted by LD84

Then we have a very fundamental disagreement in how sports management games should be designed and coded.

To me, current ratings should be just that. Current ratings that apply to where the player is currently at relative to other players ratings and the league they are in.

Potential ratings should be where the player could potentially reach one day.

I find it utterly absurd that potential ratings should have any bearing on statistical output at any level.

Originally Posted by jpeters1734

At this point, I understand Lukas isn’t likely to acknowledge this as a flaw, and I get that. He has a job to do and a product to protect.

But for anyone else reading: the tests speak for themselves. Whether you use extreme values or more common ratings, the pattern holds. Minor league sim performance is affected by potential, even when current ability is nonexistent. That makes stats misleading and forces users to rely more on ratings than results whether they want to or not.

Originally Posted by Lukas Berger

You and anyone else are certainly free to disagree that things should be that way or feel they should work another way.

So now that I've clarified why some things might work the way they do, I'll bow out and let you folks debate if that's the right way for them to work or not.

Four-plus pages to go through. Here's an overview of the first two pages. These quotes are not the complete respective posts. Lots of good posts followed the last one here. Fascinating discussion.

[WhiskyTango]

Classic top-down program design: decide what the final product looks like then break it down into finer layers of detail. All those calc settings determine the final season product before the season is played. That's the top. Playing out the season fills in the details. The ratings decide who gets what percentage of the total. Apparently, the devs feel combining potential with current ratings substantially expands engine performance, somehow. It hasn't been discussed how and that may be an important consideration in understanding the context of these changes and the broader contribution to the game.

[jcard]

Originally Posted by WhiskyTango

Classic top-down program design: decide what the final product looks like then break it down into finer layers of detail. All those calc settings determine the final season product before the season is played. That's the top. Playing out the season fills in the details. The ratings decide who gets what percentage of the total. Apparently, the devs feel combining potential with current ratings substantially expands engine performance, somehow. It hasn't been discussed how and that may be an important consideration in understanding the context of these changes and the broader contribution to the game.

Lingering—if anything, increasing—concerns along these lines are, directly and indirectly, leading factors in why I stopped purchasing the game this year after buying the previous 12 editions. The suspicion that targeted outputs were exerting a centripetal force on league totals at the cost of individual plate appearance outcome integrity was undermining my experience. That may well be unfair, but over time and for various reasons the OOTP team lost my confidence to trust them with such extensive black box presentation. I am not saying anyone else should feel the same. At the same time, I have seen nothing over the past four months that has caused me to question whether I made the right choice for me personally.

[OutS|der]

What levels are boosting stats? Is it all minors? Do Indy leagues boost anything? International? Are you boosting stats in the winter/development leagues or feeder leagues?

[Zooerp]

Originally Posted by sixto

It's absolutely critical to note that these scout ratings — potential ratings — are 100% not data. They are observations. Opinions! I know some will try, but IMO this is not arguable. Bat speed is data. Arm speed, arm angle, home-to-first time, average angle of attack, spin rate, barrel rate: all data. 500 potential power rating? Opinion.

Therefore, at the heart of what we are discussing is this: if minor league baseball players' statistics are influenced by potential ratings, then in fact they are influenced not by data but by somebody's opinions. That is, because Johnny Q. Scout believes that Hubert H. Homerhitter has the potential to hit 40 HR in the majors someday, somehow Johnny Q. Scout's opinion actually manifests in a boost in performance for Hubert in AAA. But not in the majors, where Johnny Q. Scout's opinions suddenly mean nothing. That's what I've gleaned from this discussion.

Except OOTP doesn't use bat speed, attack angle, barrel rate, etc for current ratings. There is no formula you can use to aggregate this data into one specific number that exactly represents a player's "Current Power" rating. A scout that gives a 20-80 scale rating based exclusively on data, is giving their opinion based on their interpretation of that data, not plugging numbers directly into a formula. If current ability was so easy to deduce based on data then in real life every team would have the exact same evaluation of the current ability of players and prospects given the amount of raw batted ball and pitch data now available.

Whatever "Current ability" means, it absolutely plays differently in different levels of baseball.