Historical Random Debut

[David Watts]

[Lukas Berger]

Originally Posted by David Watts

Attachment 543678

Again, not seeing what's wrong here

Both Ordonez's best seasons (.364/.444/.572 in your sim and .363/.434/.595 in real life) (221 hits, 30 HR single-season bests in your sim, 216 hits and 32 HR's in real life) and his career averages (.323/.396/.478 in your sim and .309/.369/.502 in real life) are pretty similar. The slight differences are well within expected mathematical variance, a difference in run scored environments due to different eras or the potential effects of a slightly weaker/stronger player pool.

[Markus Heinsohn]

Originally Posted by Lukas Berger

Again, not seeing what's wrong here

Both Ordonez's best seasons (.364/.444/.572 in your sim and .363/.434/.595 in real life) (221 hits, 30 HR single-season bests in your sim, 216 hits and 32 HR's in real life) and his career averages (.323/.396/.478 in your sim and .309/.369/.502 in real life) are pretty similar. The slight differences are well within expected mathematical variance or the potential effects of a slightly weaker/stronger player pool.

Agreed, nothing wrong here.

[swampdragon]

The real Ordonez hit .309 for his career and won a batting title with a .363 average, in the same age 33 season where he hit .364 in your game.These numbers actually aren't bad. Considering the respective run environments, they're probably a touch high, but they're well within the standard deviations you should expect.

Edit: Sorry to duplicate the others. I'm slow sometimes.

[Lukas Berger]

Originally Posted by swampdragon

The real Ordonez hit .309 for his career and won a batting title with a .363 average, in the same age 33 season where he hit .364 in your game.These numbers actually aren't bad. Considering the respective run environments, they're probably a touch high, but they're well within the standard deviations you should expect.

Exactly.

If anything, this is an argument for the game getting it right, not for it getting it wrong.

[swampdragon]

This reminded me that the real problem here is that Dave Parker isn't in the Hall of Fame, a problem I think a future Veterans Committee will solve.

[Lukas Berger]

Originally Posted by swampdragon

This reminded me that the real problem here is that Dave Parker isn't in the Hall of Fame, a problem I think a future Veterans Committee will solve.

Agreed, I think he belongs in as well.

[David Watts]

Originally Posted by Lukas Berger

I don't understand what you think is wrong there

There's nothing there that's all that far off from anything Cash did in real life.

The difference in environment between the 1960's, when Cash really played, and which were an extreme pitchers era and the 50's which were more of a hitters era (or the 80's if you're using 84 as the baseline), along with random variance are more than enough to explain the marginally higher HR totals and higher BA's in a few seasons.

That's almost exactly what you'd expect to see if you dropped the real Cash into either the 50's or the 80's. He would hit better in either of those periods, because the league as a whole hit better.

Norm Cash hit .300 once over his entire career. A monster outlier season in which he hit .361. Never sniffed .300 again. His career batting average was .271. In that sim he hit .300 plus 9 times. I just don't buy that the league numbers can come out dead on if it's based on a "if you put Norm Cash in the 40's or 50's he would challenge for a batting title every year" argument. If the game was actually making those types of determinations, the league numbers would go haywire.
Thing is, when Norms hitting that .332 in 1956, a bunch of other guys are all stinking to make those league numbers look great. I bet a bunch of the stinkers aren't having the period their playing in factored in.

[David Watts]

Originally Posted by swampdragon

The real Ordonez hit .309 for his career and won a batting title with a .363 average, in the same age 33 season where he hit .364 in your game.These numbers actually aren't bad. Considering the respective run environments, they're probably a touch high, but they're well within the standard deviations you should expect.

Edit: Sorry to duplicate the others. I'm slow sometimes.

Yeah I realized after posting the Ordonez screenshot that I made a mistake. Mainly in not noticing that two of this ootp years were injury plagued. I just seen the clump of high averages. Trying to do this and a fantasy draft at the same time was a mistake.

[swampdragon]

I'm still trying to learn about recalc. I'd guess you used one year for Ordonez and three year for Cash. Right or wrong?

With Cash, I'd point out that the 60s really were different. Transfer a 60s season into any era but deadball, and the averages should go way up.

[Markus Heinsohn]

Originally Posted by David Watts

Norm Cash hit .300 once over his entire career. A monster outlier season in which he hit .361. Never sniffed .300 again. His career batting average was .271. In that sim he hit .300 plus 9 times. I just don't buy that the league numbers can come out dead on if it's based on a "if you put Norm Cash in the 40's or 50's he would challenge for a batting title every year" argument. If the game was actually making those types of determinations, the league numbers would go haywire.
Thing is, when Norms hitting that .332 in 1956, a bunch of other guys are all stinking to make those league numbers look great. I bet a bunch of the stinkers aren't having the period their playing in factored in.

What you don't seem to understand is that when you import Cash, his ratings get era-neutralized and he gets dumped into a random player pool whose ratings are also neutralized. Then the LTM recalculation comes and puts the entire league average stats output on a certain realistic level. Now, how one player performs in this environment depends on a) his ratings and b) the environment (competition) and c) the league stats output / LTMs. And then the usual small sample size / statistical variation issue comes in, of course it is easy to cherry-pick someone like Cash from a large pool of players. None of the examples you have just posted illustrates a problem to me, sorry.

[Lukas Berger]

Originally Posted by David Watts

Norm Cash hit .300 once over his entire career. A monster outlier season in which he hit .361. Never sniffed .300 again. His career batting average was .271. In that sim he hit .300 plus 9 times. I just don't buy that the league numbers can come out dead on if it's based on a "if you put Norm Cash in the 40's or 50's he would challenge for a batting title every year" argument. If the game was actually making those types of determinations, the league numbers would go haywire.
Thing is, when Norms hitting that .332 in 1956, a bunch of other guys are all stinking to make those league numbers look great. I bet a bunch of the stinkers aren't having the period their playing in factored in.

He hit that low literally because he was in the 60's though. It was the worst hitters era in baseball history, by far (post dead ball era). Put Rogers Hornsby in the 60's and he'd have probably hit a lot like Al Kaline. You can buy the argument or not, but it's a definite factor in the discussion. Especially when you're mixing players from all eras, the run environment they put their original stats up in is going to make a big difference.

For his career he comes out about 27 points higher in BA in your sim than in real life.

He played his real career in leagues where the average BA was around .245-.250

If you're playing in the 1984 environment, the average BA was around .260. That explains around 15 points of average right there.

The other 12 points can easily be explained by the vagaries of the random debut environment, player pool strength and typical statistical variance.

EDIT: I see Markus beat me to it with a more concise explanation

[sprague]

Question for David or the guys. As this thread has me interested in trying random debut again.
The question
If I start a league. Delete the entire first crop before the inaugural draft.
Then create enough fictional players to start the league...
All the original deleted guys will show in later amateur drafts. Am I right.

[Markus Heinsohn]

Originally Posted by sprague

Question for David or the guys. As this thread has me interested in trying random debut again.
The question
If I start a league. Delete the entire first crop before the inaugural draft.
Then create enough fictional players to start the league...
All the original deleted guys will show in later amateur drafts. Am I right.

Historical random debut has nothing to do with fictional players really, but if you want it that way, what you write in theory works, yes.

But why would you want to start out with fictional players when they get replaced with the historical players eventually anyway?

[sprague]

Originally Posted by Markus Heinsohn

Historical random debut has nothing to do with fictional players really, but if you want it that way, what you write in theory works, yes.

But why would you want to start out with fictional players when they get replaced with the historical players eventually anyway?

Hey Marcus
I just like all the historical players to have full careers.
Last time I played I created fictional players between the ages of 27 and 33 so they only lasted a few years before the new drafts began taking over the league.
So each historical player starts at 19

[Markus Heinsohn]

Originally Posted by sprague

Hey Marcus
I just like all the historical players to have full careers.
Last time I played I created fictional players between the ages of 27 and 33 so they only lasted a few years before the new drafts began taking over the league.
So each historical player starts at 19

OK, that makes sense.Yes, that should work...

[actionjackson]

Originally Posted by Lukas Berger

He hit that low literally because he was in the 60's though. It was the worst hitters era in baseball history, by far. Put Rogers Hornsby in the 60's and he'd have probably hit a lot like Al Kaline. You can buy the argument or not, but it's a definite factor in the discussion. Especially when you're mixing players from all eras, the run environment they put their original stats up in is going to make a big difference.

For his career he comes out about 30 points higher in BA in your sim than in real life.

He played his real career in leagues where the average BA was around .245-.250

If you're playing in the 1984 environment, the average BA was around .260 That explains around 15 points of average right there.

The other 15 points can easily be explained by the vagaries of the random debut environment, player pool strength and typical statistical variance.

EDIT: I see Markus beat me to it with a more concise explanation

Here's my Norm Cash from an OOTP16 random debut league: 3-year recalc double weighted, neutralized stats, all ballparks neutralized, weather neutralized, stats output set to 1984 (4.26 R/G, .260/.323/.385/.708) for every season. The Norm Cash in David Watts game went from 1947-1964 with I believe (correct me if I'm wrong David Watts) stats progressing as they did IRL from 1947-1964. My Norm Cash had an OPS+ of 140, and a wRC+ of 140. RL Norm Cash had an OPS+ of 139, and a wRC+ of 139. OOTP16 abso-freakin'-lutely nailed it. David Watts' OOTP18 Norm Cash had an OPS+ of 151 (I'd be curious to know what his wRC+ is David Watts if you could). As far as I'm concerned David Watts' Norm Cash is within a reasonable range of where he should be. It's a little bit high, but it's nowhere near some of the other ones that were raised in the bug reports threads.

I'm more concerned about Jean Segura, Ron Santo, Norm Cash, Hal Newhouser from the start of this thread. There are also Jerry Helper's results from that thread. I'm not sure what David Watts used for recalc in that thread, but perhaps if he used 1-year recalc it would explain the extremes. There are some other outlier guys in the 1-year recalc thread, but Markus has said that we should use 3-year recalc at a minimum in historical random debuts so I'll definitely take him at his word on that. I would suggest we stay away from 1-year recalc per his advice for this reason.

Two things I'm very encouraged by as far as OOTP19 is concerned are: 1) pstrickert's beta testing results, which I believe Lukas touched on earlier. OOTP19 appears to be flattening out the outliers that we were seeing, and 2) Markus has said he has improved defensive ratings in historical games. These two things make me believe that the possibility exists that we can leave the extreme seasons behind in OOTP19. That would be cause for celebration. These two things, in combination with listening to Markus' suggestions re: settings to use in random debut historicals give me hope that this problem might be solved. BUT, as always, the proof will be in the pudding.

Now, can we all stop fighting on OOTPmas please? I've gotta get to bed early so that Santa Heinsohn can come down the chimney (even though I do live in an apartment...technicalities ) with my new version of OOTP. I hope you understand Markus that we're just trying to keep it real with the stats, and certainly some of the results we were seeing needed to be looked at. I say, bring on OOTP19 and I'll put it through it's paces, and be brutally honest with what I'm seeing good (hopefully) or not so good (hopefully not). OK? Peace out.

[Reed]

Thank you Markus (and Lukas) for your inputs. I know how busy you must be now so I appreciate you taking time out to look at this. I think OOTP usually does a very good job of simulating a players career and this discussion reinforces those feelings.

[waittilnextyear]

Prime Dave Parker batting .407 in 1948 is something that seems a little high. But, given (n=1) and that Parker is a HoF talent that actually batted .338 in 1977 (led the league in a much tougher batter's environment, .324 league wOBA in 1977 vs .341 league wOBA in 1948) it's something I chalk up to the joys of variation and putting a great player in a very good context for him to shine. Sim that same save out 10 times and I bet he doesn't quite approach this output in any of the other 9. Ted Williams played from 1939 to 1960 and he batted .400+ exactly once in a full season. Not that he wasn't a threat to do so in every other season.

Putting Jose Altuve, Magglio Ordonez etc into the mid-1940s--I would not be surprised if either of them hit .400. Both are some of the best contact hitters of this century so far. Altuve is arguably the best contact hitter in the game at the same time that Mike Trout, Buster Posey, and Joey Votto are all a thing.

Let's not forget what huge factors league context and variation can be. Stan Musial led baseball with a .376 batting average in 1948 so you might say "but, yeah it's Stan freaking Musial," but Harry Walker led baseball in 1947 with a .363 batting average. He was a .296 career hitter and otherwise small footnote in baseball history. The game is seriously littered with Brady Anderson's and Bret Boone's and Davey Johnson's who had wild, crazy inexplicable seasons. Getting these oddball seasons out of the OOTP sim engine gives me the feeling that it's dealing properly with variation, rather than indigestion that something might be working incorrectly.

[actionjackson]

Originally Posted by waittilnextyear

Prime Dave Parker batting .407 in 1948 is something that seems a little high. But, given (n=1) and that Parker is a HoF talent that actually batted .338 in 1977 (led the league in a much tougher batter's environment, .324 league wOBA in 1977 vs .341 league wOBA in 1948) it's something I chalk up to the joys of variation and putting a great player in a very good context for him to shine. Sim that same save out 10 times and I bet he doesn't quite approach this output in any of the other 9. Ted Williams played from 1939 to 1960 and he batted .400+ exactly once in a full season. Not that he wasn't a threat to do so in every other season.

Putting Jose Altuve, Magglio Ordonez etc into the mid-1940s--I would not be surprised if either of them hit .400. Both are some of the best contact hitters of this century so far. Altuve is arguably the best contact hitter in the game at the same time that Mike Trout, Buster Posey, and Joey Votto are all a thing.

Let's not forget what huge factors league context and variation can be. Stan Musial led baseball with a .376 batting average in 1948 so you might say "but, yeah it's Stan freaking Musial," but Harry Walker led baseball in 1947 with a .363 batting average. He was a .296 career hitter and otherwise small footnote in baseball history. The game is seriously littered with Brady Anderson's and Bret Boone's and Davey Johnson's who had wild, crazy inexplicable seasons. Getting these oddball seasons out of the OOTP sim engine gives me the feeling that it's dealing properly with variation, rather than indigestion that something might be working incorrectly.

The argument was never about one outlier season. I agree those are entirely within the realm of possibility for just about any Joe Schmoe. It was about a player stacking five or so of them in a row, which was what it was looking like here and here.