Historical Simulation Stats Issue

[majesty95]

Originally Posted by actionjackson

I've got a great 62-season random debut going in OOTP16. I hope to eventually be able to start one in a more up to date version, but until I see results, OOTP16 it is.

I found my license for 16. Woohoo! Any suggestions or things I need to know? I usually play historical sims from the early-mid 80's on.

[actionjackson]

Originally Posted by majesty95

I found my license for 16. Woohoo! Any suggestions or things I need to know? I usually play historical sims from the early-mid 80's on.

It's pretty tight. See the second thread for some of my results with stats distribution. Special seasons are really special because of their rarity. It becomes kinda dull when a player or two is having one every year, but that doesn't really happen in OOTP16, so enjoy it, and hopefully OOTP19 will bring us better results. So far (according to that thread) it looks promising.

[majesty95]

Originally Posted by actionjackson

It's pretty tight. See the second thread for some of my results with stats distribution. Special seasons are really special because of their rarity. It becomes kinda dull when a player or two is having one every year, but that doesn't really happen in OOTP16, so enjoy it, and hopefully OOTP19 will bring us better results. So far (according to that thread) it looks promising.

How do you get guys to retire at an appropriate age without using retire according to history? I seem to remember you used to need to use development and like 1.2 player aging modifier? I saw Nick Esasky in a test league and he is kind of the perfect example of why it's important lol.

[David Watts]

Take a look at the batting average leaders(1984, full minors, 1 year recalc, development on, import modifiers according to history). Go to Baseball Reference and look at each of the guys in the batting leaders real life 1984 season. Every single guy on that list hits at least 29 points higher than real life. 3 hit 40+ higher and one hit 55 points higher. Just bizarre. Plus look at the hits leader. Amazing that you can randomly pick a season to test and instantly get a individual hit season that ranks among the all-time best.

[pstrickert]

Originally Posted by majesty95

Another thing that I noticed on this is that Roger Clemens almost always has an ERA right around 2.00 Doesn't matter the year. He'll have one year where its like 2.26 but almost every other year is between 1.98 and 2.06. It's almost like he's performing off of his best season every year vs getting a new set of ratings every year. Or the ratings used are too ambiguous and unable to replicate the individual seasons.

I have not seen this happening. Last night, I ran a sim from 1901-2016 (1-yr recalc using real lineups & transactions). Here's Clemens:

(sim / real)

W: 364 / 354
L: 192 / 184
ERA: 2.78 / 3.12
WHIP: 1.07 / 1.17
BAA: .271 / .272

[pstrickert]

Originally Posted by majesty95

I'm having some issues trying to do a semi historical replay. The stats are WAY off. I've always thought OOTP did a good job of historical sims. However, this recent league and subsequent test league has things WAY out of whack.

I started off my game with 3-year recalc. I had several guys hitting over 50 home runs in the early to mid-80's when that was not seen (including Eric Davis who had 52 HR and 69 SB). I decided because of this and amateur draft compensations issues to run a test league.

In my test league I did 3-year recalc and double weighted current year. The home runs were mildly more realistic but that could have been just variance. However, I was seeing a lot of examples of guys who had never hit better than say .300 hitting .350. I then changed the settings to only use a one-year recalc. I then had John Olerud hit .340 in a season which he hit .265 in real life.

I always thought OOTP was able to do historical replays and have the stats come out more similarly yo real life. I understand things can affect this like facing different pitchers, variance, etc. I just find it hard to fathom that a guy who never hit better than .304 over a 10-year period would magically break out and hit .350 or that a rookie who hit .265 in real life would somehow hit .340 and lead the league.

Is there a setting somewhere I'm missing (I use import historical modifiers) or is the sim engine just way more random than I remember?

Standard deviation, as you know, means you'll inevitably see outliers. If the outlier happens to a season which happened to be an outlier for a player, then the issue is compounded. George Brett, for example, hit .390 in 1980. In OOTP, he could theoretically be off by 50 points either direction due to statistical variance. It (randomly) happens. But it doesn't necessarily mean something's wrong with the game.

[David Watts]

Originally Posted by pstrickert

Standard deviation, as you know, means you'll inevitably see outliers. If the outlier happens to a season which happened to be an outlier for a player, then the issue is compounded. George Brett, for example, hit .390 in 1980. In OOTP, he could theoretically be off by 50 points either direction due to statistical variance. It (randomly) happens. But it doesn't necessarily mean something's wrong with the game.

I actually enjoy standard deviation. Seeing a guy over achieve here or under achieve there is what makes things interesting. I've always felt the coolest thing about OOTP historical is not that it always nails a players single season stats to the letter, but that it has always came very close to nailing the numbers a player posted over an entire career. But, 18 has turned standard deviation into a somewhat of a farce. Super seasons are so routine that they become mundane. 200 hits in a season is a benchmark accomplishment in real life. You'd have to use 220, if not 230 as the OOTP18 benchmark. Man, I hope they can get this fixed.

[pstrickert]

Still working on it, David. If OOTP was just another baseball replay game, Markus could devote all his time to perfecting the historical replay aspect. But you know and I know OOTP is so much more. It offers so many different features, appealing to so many different users hoping to play it in so many different ways, Markus cannot over-compensate one area of the game at the expense of other areas. He’s been more than fair to us historical baseball aficionados, IMO. (Doesn’t mean I’m not greedy for even more! )

[David Watts]

Originally Posted by pstrickert

Still working on it, David. If OOTP was just another baseball replay game, Markus could devote all his time to perfecting the historical replay aspect. But you know and I know OOTP is so much more. It offers so many different features, appealing to so many different users hoping to play it in so many different ways, Markus cannot over-compensate one area of the game at the expense of other areas. He’s been more than fair to us historical baseball aficionados, IMO. (Doesn’t mean I’m not greedy for even more! )

I agree, but I also feel if something isn't broken....in fact if something may even be at a level that other games can only hope to be, then why the hell break it? 16 wasn't perfect, but it was a damn good 3/4 of the way there. As a random debut player and a historical player, I really don't need a ton of flash. I love additions like the Hall of Fame plaques. Hall of Fame voting. The achievement icons. I love the new stadium models. I play on the broadcast screen. I love the little asterisks announcing a new pitcher has entered the game. If right now we had those things and historical was the exact same game it was when it was called OOTP16, I would be perfectly happy. Having Correa, Seager etc would be enough for me. Sure, I would be complaining at times about the spinning cursor, the earthquake shakes and the RHP,LHP in front of the batters, but I could get over or at least used to that stuff. I can't get over .375 being the new .280 or 250 hits being common.

[pstrickert]

I hear you. I don't think any of us know what happened. Or exactly when. Sometimes, fixing one thing breaks another thing somewhere else. Or an enhancement has unintended consequences. Then again, I'm no programmer. I have no idea how it all works under the hood. Appreciate the comments -- and the passion, David!

[NoOne]

you can either callibrate it based on individual results or you can callibrate based upon league-wide results and then anything in between those two choices that you can think of.

basically we can't affect distribution of ratings much -- that determines how much of the pie that person is likely to get, and over time will get. a dearth or saturation as possible exclusions, a guy who is ~125% of league baseline will continue to be 125% of league baseline after most any change. (again*, extremely small or large League Totals may not adhere to this statement, same with anything that drastically changes distribution of players, like too many or too few causing an significantly smaller/larger # of better players than normal.. a dearth over time may still 'add up' but any one year could be barren)

volatility amounts to a fairly consistent percentage of baseline, and when it's a larger range, the slices get larger per '1-pt' rating change = greater variation between good and bad. (*** similar exclusions as above, assumes staying 'in the ballpark' relative to previous settings)

i care about both, but i will fudge a little on league-wide resutls to get individual results that i want to see. mainly peak performances and how often they occur. whatever your opinion on the matter is, you'll likely have to compromise on something.

[David Watts]

Originally Posted by pstrickert

I hear you. I don't think any of us know what happened. Or exactly when. Sometimes, fixing one thing breaks another thing somewhere else. Or an enhancement has unintended consequences. Then again, I'm no programmer. I have no idea how it all works under the hood. Appreciate the comments -- and the passion, David!

Ugh! I find it easier in my mixed up, muddled up world to blame the pawns for everything....heck if I get bad service at McDonalds I blame the pawns But, I've never been the most rational man on earth.

[NoOne]

Originally Posted by David Watts

I agree, but I also feel if something isn't broken....in fact if something may even be at a level that other games can only hope to be, then why the hell break it? 16 wasn't perfect, but it was a damn good 3/4 of the way there. As a random debut player and a historical player, I really don't need a ton of flash. I love additions like the Hall of Fame plaques. Hall of Fame voting. The achievement icons. I love the new stadium models. I play on the broadcast screen. I love the little asterisks announcing a new pitcher has entered the game. If right now we had those things and historical was the exact same game it was when it was called OOTP16, I would be perfectly happy. Having Correa, Seager etc would be enough for me. Sure, I would be complaining at times about the spinning cursor, the earthquake shakes and the RHP,LHP in front of the batters, but I could get over or at least used to that stuff. I can't get over .375 being the new .280 or 250 hits being common.

i really think a few changes to your modifiers and/or totals could make it look like what you saw in the past.

a side note -- i don't hink the export schemas have changes between 16 and now? if so cna still do this with a little elbow grease.

export players from 2 leagues -- one from your '16 installation and one from your '18 installation. make sure they are ~equal. the key will be to retain existing PlayerID column and paste over the rest from your '18 exported file.

you need the same or greater # of players in the destination '16 league. player id must exist! you will re-use them.

don't paste over teamID etc... if 'personality' columns are new, jsut make them and give "100" out of 200 for all, type once, copy/paste to fill -- that's exactly what it was before personalities effects were added to cause variation between players.

oh i said that backward. remove entire columns as opposed to adding info is more likely. match order of column labels to destination league. cutting info out is easier than adding.

once it is ~congruent to the right exported schema, you can now import this modified players file.

this sounds like it takes more time than it does... and can be a template for future re-uses... you can feasibly have your 'new' players with your old ootp.

[majesty95]

Originally Posted by pstrickert

Standard deviation, as you know, means you'll inevitably see outliers. If the outlier happens to a season which happened to be an outlier for a player, then the issue is compounded. George Brett, for example, hit .390 in 1980. In OOTP, he could theoretically be off by 50 points either direction due to statistical variance. It (randomly) happens. But it doesn't necessarily mean something's wrong with the game.

Standard deviation, to me, shouldn't equate to a 50 point swing in either direction. Maybe I don't understand probability well enough to make this argument, but to me, I really don't think any player should deviate more than 25 points or so from real life. Think about it. If a .300 hitter hits .275 for a season he had a very off year. Seeing a consistent .300 hitter hit .250 is unfathomable unless there was some something serious going on in his life. Now, there are hitters who are streaky and can have wild ranges for production. But I just can't imagine Tony Gwynn ever hitting .270 nor Tony Pena hitting .340. But that's what we can get very easily in 18. As David said, it would be one thing if we never had realistic outcomes. The fact that they've gone so drastically in the undesired direction is the issue.

[majesty95]

Adding to that, I worry the rabbit hole we're going down with 3D models, challenge mode and now Perfect Team will further draw the focus away from what built the game to what it is, accuracy in statistical simulation and AI management.