Rasnell's Bugs - HR too low

[Garlon]

I can't imagine HRs could be off by that much when using the "auto-adjust league totals modifiers for historical accuracy" option. Please post your simulated HR and AB league totals for each of these seasons.

[rasnell]

Originally Posted by Garlon

I can't imagine HRs could be off by that much when using the "auto-adjust league totals modifiers for historical accuracy" option. Please post your simulated HR and AB league totals for each of these seasons.

I can give you more, but here's the start, looking at the 8 fictional AL teams, with this list beginning in 1929 and going to 1935 (I've now gone 30 years):

HR/AB
233/45,230
825/46,100
553/45,216
708/45,010
611/44,736
693/44,877
802/45,476

Garlon, you would understand all of this a lot better but this seems to be a tremendous fluctuation. Also, I have no idea how all these numbers affect each other. Realize that clicking on the historical player creation and historical modifiers were adjusting the pcm and modifier numbers each year, too.

The real MLB boom years of the 1930s saw some big HR numbers, as you know. In this fictional league, only one player reached 30 in any given season in the 1930s.

Looking at real MLB league averages and ERAs and comparing to the simmed season results, the batting averages were pretty accurate throughout, but ERAs are about a full run too high in the sim results.

[rasnell]

After 30 years of simming this fictional league of 16 teams, here is where the modifiers stand in 1959. It confuses me why PCM would be turned down so low and league modifiers would be turned up so high.

Perhaps this sheds some light for you, Garlon. What do you think? I've not touched any of the settings since the default start of the season in 1929 and, as you see, the historical modifier buttons are checked on.

[battists]

Rasnell's bug from another thread:

Originally Posted by rasnell

Home runs in the default settings are too low by about 15 percent, even using the historical modifiers in the default game.

[battists]

Sorry for the goofy order in this thread. :P

[battists]

Can anyone help me flesh out the information in this thread, please?

Not being a historical simmer, I can't see where any of the values are out of line. It would help if you could provide something like the "expected range" for each of the years that you simmed, as well, so I can see where the problems are falling.

Thanks,

Steve

[MadMax58]

Just an off the cuff answer FWIW.

The PCM on Batting power looks strangely low. But without knowing how many years back they have been that low it's difficult to know whether that is affecting the stat output.

In contrast, the League totals for HR's looks very high, but again difficult to understand without knowing how long it's been creating/importing players at that above mentioned PCM, or if the HR LT looks that way every year, or just 1 year, etc.

There were some big HR seasons in the early to late 30's era...but I don't have exact stats on league totals handy at the moment to compare to today's totals.

But 30 HR's by 1 guy is low for that era, as guys like Foxx, Wilson, and Greenburg all hit well into the 50 HR range in this decade. And they weren't the only ones above 30.

[rasnell]

I changed nothing, used the defaults and clicked on the historical modifiers and PCM, importing from the Lahman database.

I did not track how it changed the settings automatically each year, but showed the screenshot for the most current season. The sim has been off for all 30 years -- 1935-1965.

Same thing happens when you use the default settings and ask it to use historical modifiers in a fictional league. Steals are way too high, home runs way too low.

[battists]

Can someone please post the "acceptable" range for homeruns from 1901 - 2005, so I can know what the heck "off by X%" means?

Thank you!

Steve

[Markmeister]

Actually, I've been looking at the home run numbers from 1901 to 1930.

In my opinion the player creation modifiers are probably too low. The reason they are too low is different than you would expect.

The PCMs create enough players with power so that the league totals are fairly accurate. However, from 1901 to 1920 using the same numbers the variance between players is not enough. What I am thinking is that when the PCMs are close to 0, OOTP is generating too many players with similar HR power.

I think that I have come up with good modifieid numbers to come up with fairly accurate league numbers for 1901 to 1920. I need to test some more though.

I haven't come up with a good adjustment for 1921 forward yet. Using the same formula as I did for 1901 to 1920 only gave me league leaders with roughly 35 homeruns. Better than what I was seeing with the original PCMs though.

[Markmeister]

dola,

I think that I immediately see the problem with the league numbers.

Rasnell - Unless I am reading your post wrong, your first year was 1929?

I was seeing the same problem starting in 1901. Numbers way off the first year, jumping wildly for a couple years, and then becoming more realistic.

The problem is caused by the first year league totals not being correct. OOTP is trying to adjust, but is over compensating when the first year deviates too far from historical totals. It takes a few years, but the totals eventually will become correct - usually by 1905 in my tests.

I adjusted the league total modifier for HRs from 1.000 to 0.800 for my 1901 tests. The 1901 number is better which leads to better numbers for 1902.

[Markmeister]

trola,

Most of the testing that I have done is by sight as I could immediately see that HR numbers were way off.

I have a league sitting around that appears to have fairly good numbers, though.

Give me some time to compile the HR totals and I will post them later tonight.

[rasnell]

When I say that it's off by about 15 percent, I'm looking at the simmed totals for the league in HR for a year and comparing to the Baseball Encyclopedia for the historical totals for that same season and league.

Do the league settings correlate with the historical numbers? In other words, if you looked at MLB in 1929 and copied down the AB, HR, etc. would it sim accurately or is there no correlation?

I assumed that the import from Lahman somehow set those first year modifiers. How are the default numbers generated for the first year?

My sims remained consistently low for HRs even after 30 years of simming. It never straightened out.

Yes, started in 1929 and continued through 1959.

[Markmeister]

Rasnell,

The League Totals numbers do not correspond to the historical totals. That has been a 'problem' with every version of OOTP. Don't worry though, enough people play around with the League Totals that someone will come up with good League Totals.

I've attached the results of a 1901 to 1910 test with what I believe are improved PCMs for 1901 to 1920.

The only odd year is 1909 with a huge jump - the NL had a huge jump in HR compared to 1908 and 1910. The overall comparison for 1901 to 1910 is only 1.68% less than historical, though.

p.s. The AB are low for the test as I forgot to change from 140 to 154 games in 1904.

[Markmeister]

Okay, found the reason for the jump in stats for 1909 - was using Garlon/Arod's modified era stats (5 year wave)

[Markmeister]

Ran a 1930 to 1939 test using my modified era_modifiers

Only problematic year was the first year. 1938 had a big jump, but that may be just an anomally.

No super-power hitters during the era though.

[rasnell]

Originally Posted by Markmeister

Ran a 1930 to 1939 test using my modified era_modifiers

Only problematic year was the first year. 1938 had a big jump, but that may be just an anomally.

No super-power hitters during the era though.

Your test matches the same problems that I reported in initial thread. Thanks for checking and verifying.

[Markmeister]

Okay, did some more research and testing.

First the testing - increasing the PCMs to closer to 1.0 after 1920 did not help create great power hitters. The most I saw was a single occurance of 40 HR in a season. Instead of getting league leaders with 45-55 HR for the time period, I am getting leaders with 30-40 HR. A 15 HR drop.

I think I found the problem though. It is the lack of power in the average player during 1930 to 1939.

Comparison of historical 2005 to 1935:

Year,AB,HR,AB per HR (league average),AB per HR (league leaders)
1935, 86437, 1325, 65.24, 19.98
2005, 166335, 5017, 33.15, 13.14

If you take 1935 and project out to 166,355 AB you would end up with 2,550 HR. Roughly 50% of 2005 - that's reflected in the AB per HR figures.

The reason that great power hitters aren't being generated is the disparity between the AB per HR for the league average and league leaders. A difference of 32.09 vs. 6.84. I don't think that OOTP can't compensate for this disparity using the default PCMs and League Totals. If you try to generate great power hitters by increasing the individual player power modifier, OOTP looks at the league-wide HR totals and makes sure that those numbers are correct. This essentially negated my adjustment to the power modifier preventing great power hitters from showing up.

I have a couple more ideas to try, but won't get around to trying them for a couple of days.

[MadMax58]

Mark...your results seem to make a lot of sense, sort of like the Ruth factor, only to a lesser degree with only a few guys hitting a majority of HR's.

But how does it explain guys like Gehrig, Foxx, Greenburg & co. not getting into the 40's much (or at all in rasnell's case)? I mean, it makes complete sense if it were a fictional league, thus making everybody closer in HR power, but he's importing real players who should have ratings reflective of the disparity already, right?

Rasnell...are any or a lot of those players I listed not in the league from CEI or early retirement by any chance?

[rasnell]

Originally Posted by MadMax58

Mark...your results seem to make a lot of sense, sort of like the Ruth factor, only to a lesser degree with only a few guys hitting a majority of HR's.

But how does it explain guys like Gehrig, Foxx, Greenburg & co. not getting into the 40's much (or at all in rasnell's case)? I mean, it makes complete sense if it were a fictional league, thus making everybody closer in HR power, but he's importing real players who should have ratings reflective of the disparity already, right?

Rasnell...are any or a lot of those players I listed not in the league from CEI or early retirement by any chance?

You're assumption is right MadMax. Everything is normal with their health and historical ratings. The game is just all over the board each year trying to swing the historical modifiers back into line. You hardly get any of the big guns into the 30-hr neighborhood during an incredible power era of the early 1930s.

To me, the easiest override would be to turn off the historical modifiers and simply add about 15-20 percent higher league modifiers manually for home runs. But that seems to defeat the whole point of having default settings and of clicking the historical modifier button to synch this every year.