1 year recalc---is it broken?

[pstrickert]

Oops! It's on the beta forum. Sorry about that!

[pstrickert]

I noticed robc still checks in from time to time. Would you mind sending him a PM? I'm sure he'd share the file with you (and instructions). Otherwise, if you don't hear from him, I could share the file via PM.

[pstrickert]

Here's a screenshot after simming the 1957 season. Since I highlighted Ted Williams, his info appears in the box at the top. Below is the data for the season as a whole (including only players with 350+ ABs).

[actionjackson]

Originally Posted by pstrickert

I noticed robc still checks in from time to time. Would you mind sending him a PM? I'm sure he'd share the file with you (and instructions). Otherwise, if you don't hear from him, I could share the file via PM.

I'm currently running four tests from 1882-2014. 2 with straight historical, and 2 with random debut historical, OOTP16 vs OOTP18. Are these stats comparisons that RobC's file looks at to do with individual players or the league as a whole? I ask because it's been clear that league-wide stats are just fine in both OOTP16 and OOTP18. What's off is the distribution of those stats in OOTP18.

[actionjackson]

Originally Posted by pstrickert

Here's a screenshot after simming the 1957 season. Since I highlighted Ted Williams, his info appears in the box at the top. Below is the data for the season as a whole (including only players with 350+ ABs).

Whoa! That's an awesome little doo-hickey right there! My four sims will be 3-year recalc, double weighted for the current year, using Real Stats, with player development on.

[actionjackson]

Because of that 3-year recalc, double weighted for the current year setup, my results may not be ideal for comparison with what happened IRL in whatever season it happened in. 1-year recalc would work better for that, but my personal preference is 3-year, double weighted recalc, so that's what I go with.

[pstrickert]

Originally Posted by actionjackson

I'm currently running four tests from 1882-2014. 2 with straight historical, and 2 with random debut historical, OOTP16 vs OOTP18. Are these stats comparisons that RobC's file looks at to do with individual players or the league as a whole? I ask because it's been clear that league-wide stats are just fine in both OOTP16 and OOTP18. What's off is the distribution of those stats in OOTP18.

Individual players. The tool might help us notice where outliers happen more often than usual.

[actionjackson]

Originally Posted by pstrickert

Oops! It's on the beta forum. Sorry about that!

Yes, I backed out of beta. Hence no access.

[pstrickert]

Originally Posted by actionjackson

Because of that 3-year recalc, double weighted for the current year setup, my results may not be ideal for comparison with what happened IRL in whatever season it happened in. 1-year recalc would work better for that, but my personal preference is 3-year, double weighted recalc, so that's what I go with.

Yes, it works better with 1-yr reacalc. But if there’s a problem with 1-yr, there’s probably a problem with 3-yr.

[pstrickert]

Originally Posted by actionjackson

Yes, I backed out of beta. Hence no access.

Boooooooooooooo!!!

[actionjackson]

Approaching the "steroid" era now with the first test. Please fasten your seatbelts ladies and gentlemen...It's about to get bumpy.

[David Watts]

Guys, I know this will sound insane.....you may even tell me I should seek professional help, but I often think about these issues all night long. Toss and turn, trying to figure out this or that. Well one of the things I got stuck in my head last night was did OOTP advertise a new feature for the fictional side of 18 that claimed to enhance or improve the way outliers performed? For some reason I convinced myself that this was one of the things 18 touted when it was released. What I'm getting at is, if so, could this be bleeding into the historical game? Especially, in a historical game when development is left on? Could that be why the players in the screenshots I posted are performing like superstars? Does the game spot a player having a good season and then make some determination to turn it great? To make it stand out? I know I'm grasping at straws here, but in my sleep last night, I became convinced that the feature list mentioned outliers.......am I nuts?

[pstrickert]

Oh, boy, David. You’re pitchin’ softballs now!

[Lukas Berger]

I seem to recall asking David this before on a similar issue, but what amount of ab's are you guys using for the make worse/make bad option?

That's probably the one single setting in historical that will have the biggest effect on the distribution of stats from the top of the league to the bottom, or vice-versa.

[pstrickert]

Usually, I reduce the first figure (ABs) from 300 to 150. The rest I leave alone.

[David Watts]

Originally Posted by Lukas Berger

I seem to recall asking David this before on a similar issue, but what amount of ab's are you guys using for the make worse/make bad option?

That's probably the one single setting in historical that will have the biggest effect on the distribution of stats from the top of the league to the bottom, or vice-versa.

For me, the only time I ever change those from default is when I do a real lineups sim. I do it then because the game tells you to do so....I think it says to reduce hitters side by 1/2.

All the screenshots and stats I've posted are with the default make worse/make bad settings. I remember several versions ago, I decided to mess with the make bad settings in a league and got some really crazy results. Herb Score whiffed something like 450 batters in a season. Had some hitters do some bizarre things as well. Since then, I don't mess with those.

[David Watts]

Folks, please go look at that Andre Ethier screenshot.

Over the first 5 years of Ethier's real life career he went 722-2481. .
Over his first 5 OOTP years he goes 1118-3136

So he had 655 more OOTP at bats. He had 396 more hits.

Distribution is clearly an issue, but creating super heroes is just as big an issue. A .300 hitter over 5 real playing years, becomes a .360 hitter in OOTP.

[Lukas Berger]

Originally Posted by David Watts

Folks, please go look at that Andre Ethier screenshot.

Over the first 5 years of Ethier's real life career he went 722-2481. .
Over his first 5 OOTP years he goes 1118-3136

So he had 655 more OOTP at bats. He had 396 more hits.

Distribution is clearly an issue, but creating super heroes is just as big an issue. A .300 hitter over 5 real playing years, becomes a .360 hitter in OOTP.

I would guess this is in part somehow a user settings/user error thing, because there's clearly a correlation between a guy having an unusually good year in game and then following it up with more good years.

If it was purely a distribution issue, you wouldn't necessarily see the correlation of one good year in-game leading to more good years like with the Shane Mack or Andre Ethier screenshots. You'd see guys having too good years, but it would vary from year to year as to which guys had those extra good years. It wouldn't be the same guys all the time.

So I think there's something about your settings that means you're not strictly using the base historical recalc but the development engine is somehow getting involved.

Now that's not to say there isn't a problem with distribution, Markus has already fixed a couple things he found and we'll take a harder look at that during beta as well.

Still, I think there's more going on here than just a distribution issue, unless everyone who had better historical season than Ethier or Mack are also putting up better stats than they are in-game, which it doesn't sound like is the case.

[David Watts]

Originally Posted by Lukas Berger

I would guess this is in part somehow a user settings/user error thing, because there's clearly a correlation between a guy having an unusually good year in game and then following it up with more good years.

If it was purely a distribution issue, you wouldn't necessarily see the correlation of one good year in-game leading to more good years like with the Shane Mack or Andre Ethier screenshots. You'd see guys having too good years, but it would vary from year to year as to which guys had those extra good years. It wouldn't be the same guys all the time.

So I think there's something about your settings that means you're not strictly using the base historical recalc but the development engine is somehow getting involved.

Now that's not to say there isn't a problem with distribution, Markus has already fixed a couple things he found and we'll take a harder look at that during beta as well.

Still, I think there's more going on here than just a distribution issue, unless everyone who had better historical season than Ethier or MAck are also putting up better stats than they are in-game, which it doesn't sound like is the case.

Like I said at the start of this thread, Ethier, Mack, Hosmer and Burks could be a product of recalc and development no longer meshing. That being said, those two features worked perfectly fine together when playing OOTP16. There is a chance that development is over taking recalc and turning players fictional. Afterall, Mack, Ethier and Burks appear more like the insane fictional Pumpkin player I took a screenshot of than they do real players.

So, if that's the user setting errror, how do you explain all the other numbers I posted. The ones from straight historical leagues with development off. The random league with development off. Take a look at the Andres Galarraga screenshot. What about the numbers Pstrickert posted.....do you really think it's user setup errors there as well? What about Actionjackson?

Guys are reaching 240-270 hits in a season on a regular basis. That's insane. Ugh, sorry, just reread your post and I may have jumped the gun a little. My bad. This whole thing tends to get me a little too heated. Again, sorry about that.

[Lukas Berger]

Originally Posted by David Watts

Like I said at the start of this thread, Ethier, Mack, Hosmer and Burks could be a product of recalc and development no longer meshing. That being said, those two features worked perfectly fine together when playing OOTP16. There is a chance that development is over taking recalc and turning players fictional. Afterall, Mack, Ethier and Burks appear more like the insane fictional Pumpkin player I took a screenshot of than they do real players.

So, if that's the user setting errror, how do you explain all the other numbers I posted. The ones from straight historical leagues with development off. The random league with development off. Take a look at the Andres Galarraga screenshot. What about the numbers Pstrickert posted.....do you really think it's user setup errors there as well? What about Actionjackson?

Guys are reaching 240-270 hits in a season on a regular basis. That's insane.

I don't necessarily have a ready explanation, and as I mentioned I think there are multiple things going on here.

If things did work in a way you preferred with development on in 16, obviously that's cool, but I don't think that was ever really an advertised feature, so I'm not really sure how high a priority it is to "fix" that.

The other stuff though, yeah it's a definite priority and as I mentioned Markus has already done some tweaking and we'll certainly give it a hard look over the next couple months.