Stats on import look very, very wrong - why?

[dewelar]

OK, rather than wait for the answer to my ratings question (https://forums.ootpdevelopments.com/...d.php?t=353361), I went ahead and imported my manually-curated league to see how things looked. The answer is that things look very, very wrong. Mainly, every player's base stats seem to have taken a hit.


For example, here's before-and-after Kirby Puckett, vintage 1988:



And before-and-after Ron Guidry, vintage 1978:



So what's going on here? I was hoping, since I saw the post about the patch fixing import stats, but these look pretty much the same as they were when I tried this yesterday. I don't want to have to re-enter over four thousand players to account for this, so if this isn't something that can be fixed or easily worked around I may have to stick with v24.

[dewelar]

Bump.

I will also note that things are the same whether I select "Yes" or "No" when asked if I want to update the modifiers on import. Of course, I'm operating under the assumption that the base stats SHOULD be the same when importing -- if that's incorrect, let me know and I'll just throw up my hands in despair now ...

[Lukas Berger]

Originally Posted by dewelar

Bump.

I will also note that whether I select "Yes" or "No" when asked if I want to update the modifiers on import. Of course, I'm operating under the assumption that the base stats SHOULD be the same when importing -- if that's incorrect, let me know and I'll just throw up my hands in despair now ...

They definitely shouldn't be the same, because the ratings will change with the import, which changes the stats you see in the editor. The ratings set the stats you see there, not the stats set the ratings.

[dewelar]

Originally Posted by Lukas Berger

They definitely shouldn't be the same, because the ratings will change with the import, which changes the stats you see in the editor. The ratings set the stats you see there, not the stats set the ratings.

I had a feeling that was the case. My main question in response is: Why? None of my settings have changed, all I'm doing is moving my game from OOTP24 to OOTP25. How much has the calculation of the modern, neutral environment changed to cause such a vast disparity in the stats tied to it?



My problem is that I have spent years manually creating neutralized stats for something like 10,000 players, using the modern environment as defined in the game, and backwards-calculating the ratings -- in other words, all my ratings were set using the "Create current ratings based on stats" button. I started this project in OOTP 22, and for the intervening versions things have been stable. What that tells me is that the definition of the modern, neutral environment ITSELF must have changed. Weirdly, it has changed in a manner that somehow causes BOTH pitchers AND hitters to be substantially worse than they were, which makes no logical sense.


So...again, I ask, what gives?

[Lukas Berger]

Originally Posted by dewelar

So...again, I ask, what gives?

The entire ratings scale changed for OOTP25. Which changes everything to do with how the ratings are applied.

This actually creates huge improvements with our being able to be far more accurate with historical results, especially outliers, and much better able to handle universes with leagues at multiple talent levels, even with MLB/MiLB.

You can see some more info on how this affects imports here: https://docs.google.com/document/d/1...73Rl9KSzo/edit

[dewelar]

Originally Posted by Lukas Berger

The entire ratings scale changed for OOTP25. Which changes everything to do with how the ratings are applied.

Right, I get that. I will re-read that document to see if I missed anything. If you'll indulge me for a moment, though, I'd like to present a scenario, and you can let me know where I'm going wrong.


Let's take random player X. In OOTP 24, his rating for power is 125. In that version, in the modern, neutral environment on the stats page, player X hits 25 home runs in 550 AB, or about 56% more than the average (100 rating) rated hitter, who hit 16. Upon importing to OOTP 25, should I not expect him to have a power rating that allows him to produce 56% more home runs (~23.4) than the average (400 rating) rated hitter, who has 15? If not, then what SHOULD I expect? Because in order for my league to work, I need player X to hit 56% more home runs than the average hitter , because that's what my neutralized stats tell me, and if I don't get that on import I need to know how to edit his ratings to get him back there.

[Lukas Berger]

Originally Posted by dewelar

Right, I get that. I will re-read that document to see if I missed anything. If you'll indulge me for a moment, though, I'd like to present a scenario, and you can let me know where I'm going wrong.


Let's take random player X. In OOTP 24, his rating for power is 125. In that version, in the modern, neutral environment on the stats page, player X hits 25 home runs in 550 AB, or about 56% more than the average (100 rating) rated hitter, who hit 16. Upon importing to OOTP 25, should I not expect him to have a power rating that allows him to produce 56% more home runs (~23.4) than the average (400 rating) rated hitter, who has 15? If not, then what SHOULD I expect? Because in order for my league to work, I need player X to hit 56% more home runs than the average hitter , because that's what my neutralized stats tell me, and if I don't get that on import I need to know how to edit his ratings to get him back there.

The ratings are not really linear in the sense that we can be that precise when doing an import. We're also not working back from the stats that show in the editor when we adjust the ratings. Most leagues are not set up with the goal of your league, so the process is optimized differently than we would do if most leagues were created how yours is.

The import will also adjust the talent levels intentionally in some cases, if the league's talent levels are off for whatever reason.

Plus, given the actual engine has changed, the modern neutral environment that is represented in the editor for 25 is not the same as the one for 24. So even if the ratings were to change exactly proportionately, the stats would not necessarily follow.

Basically, the whole ratings matrix and import process is a very complicated thing and there are a lot of considerations involved.

One thing about 25 that should actually make what you want to do in your league even better, is that you now have a display in the editor that will show the targeted/expected stats in your actual league environment. Not only in a theoretical environment that may or may not even remotely correspond to your leagues talent level and thus may give far different results from what you'd expect.

[dewelar]

So the upshot of all this seems to be that if I want my neutralized stats to carry over from OOTP 24 to OOTP 25, the only way to achieve it is to manually re-enter all ten thousand players. That is...sub-optimal, and will probably take me years. And then I'll probably have to do it all over again for OOTP 28 or 29, to the point that I'll never actually be able to, you know, PLAY the game.

So, it looks like I'll be staying with v24 until this league is done doing what I want it to do, which last time I calculated it (based on how much time it takes me to play out each day) came out to about 4-5 years at minimum. That doesn't even get into my question about how to calculate the ratings spread for splits linked in my first post, which remains unanswered, and without which I'm stuck anyway. That's very unfortunate, as I was hoping to be able to use the new running and catcher ratings.

Ah, well, it's been fun while it lasted...

[dewelar]

So I played around with this, and decided to go ahead with reentering all my data. I have to say that the new feature allowing me to see the expected production in the current season is VERY helpful. Kudos for that. It's possible that my previous calculations may have been off, but the base stats are pretty much consistently worse by the same amount for every player -- pitchers and hitters both worse, with hitters getting the worst of it. Every single player looks similar to Puckett and Guidry above, with every stat taking a hit (e.g., batting average almost always appears to be about 60 points lower than the same player in v.24, pitcher strikeouts are almost always about 30% less, etc.). Still, if I have to do this, the ability to see the resulting stats in-line is a godsend.

[Lukas Berger]

Originally Posted by dewelar

So I played around with this, and decided to go ahead with reentering all my data. I have to say that the new feature allowing me to see the expected production in the current season is VERY helpful. Kudos for that. It's possible that my previous calculations may have been off, but the base stats are pretty much consistently worse by the same amount for every player -- pitchers and hitters both worse, with hitters getting the worst of it. Every single player looks similar to Puckett and Guidry above, with every stat taking a hit (e.g., batting average almost always appears to be about 60 points lower than the same player in v.24, pitcher strikeouts are almost always about 30% less, etc.). Still, if I have to do this, the ability to see the resulting stats in-line is a godsend.

This is probably an indication that the overall talent level in the league was too high before. Given that, guys would likely have under-performed the targets you were setting for them.

Which it was pretty easy to have that happen when doing a lot of manual editing with the old editor not taking into account the league environment as a whole.

Just one note, as you do adjust things in 25, you'll want to fairly frequently hit the rescout button (even if you have scouting off) as that will actually adjust the league average talent levels based on any changes you've made, and then show you the current up to date projected performances based on the league as a whole, including any edits you've made.

[dewelar]

Originally Posted by Lukas Berger

This is probably an indication that the overall talent level in the league was too high before. Given that, guys would likely have under-performed the targets you were setting for them.

Actually, this is kind of the point of this league. Essentially, I'm taking every player from history during their best season and running a one-year league, playing out each game. The thing is, I need to have everyone adjusted to a baseline first. So, yes, I expect that once the league starts, those could be Puckett's and Guidry's season stat lines, but first I want to know how they'd perform in a vacuum, because I am always adding new players to the league and I need to have a fixed target. That's why I won't be doing this:

Just one note, as you do adjust things in 25, you'll want to fairly frequently hit the rescout button (even if you have scouting off) as that will actually adjust the league average talent levels based on any changes you've made, and then show you the current up to date projected performances based on the league as a whole, including any edits you've made.

...like, EVER. My guess is that the game did a rescout when I imported, which is why I have to go through this whole process? I understand why given the new ratings system, but it messed me up big time.

[Lukas Berger]

Originally Posted by dewelar

That's why I won't be doing this:





...like, EVER. My guess is that the game did a rescout when I imported, which is why I have to go through this whole process? I understand why given the new ratings system, but it messed me up big time.

Up to you, but if you don't do this, you'll be setting players to a level they won't be able to perform at in your league.

Because the league averages will keep rising (I would assume, given what you've described) as you adjust more players, but the editor will keep showing you the expected performance in a league with lower averages.

So players will not actually perform how you've set them to once you start playing out games, because the entire league's baseline will have shifted.

[dewelar]

Originally Posted by Lukas Berger

Up to you, but if you don't do this, you'll be setting players to a level they won't be able to perform at in your league.

Because the league averages will keep rising (I would assume, given what you've described) as you adjust more players, but the editor will keep showing you the expected performance in a league with lower averages.

So players will not actually perform how you've set them to once you start playing out games, because the entire league's baseline will have shifted.

And, odd as it may seem, that's exactly what I want . Hopefully there won't be any further major ratings system changes anytime soon, but now I know that if there are, this is the kind of thing I can expect unless future versions include some way to turn off rescouting on importing old games.


Anyway, thanks very much for taking the time to walk me through this. You folks are awesome as always!