Rasnell's Bugs - HR too low

[Garlon]

Rasnell, I would much prefer the entire combined league totals, not just the AL. There will of course be a difference in the AL and NL depending on the distribution of power hitters in each league. First year simulation totals are always tricky.

Your data with HR per AB calculated vs real life combined:

HR/AB
1929: 233/45,230 = .0051, real .0158
1930: 825/46,100 = .0178, real .0180
1931: 553/45,216 = .0122, real .0123
1932: 708/45,010 = .0157, real .0155
1933: 611/44,736 = .0136, real .0125
1934: 693/44,877 = .0154, real .0156
1935: 802/45,476 = .0176, real .0153

1930: Hrs up 1.12%
1931: Hrs up 0.81%
1932: Hrs down 1.28%
1933: Hrs down 8.8%
1934: Hrs up 1.29%
1935: Hrs down 15.03%

1930-1935 span: Hrs down 3.47%. Several of the individual seasons look just about perfect. The overall results are probably much closer to real life, but you only provided half (one league) of the data. As a rule of thumb, I'd say anything within 5% is very accurate for an individual season or a span of seasons. For instance, if the league hit 1400 Hrs in 100000 AB, 5% more than this would be 1470 Hrs or 70 Hrs too many. 70 Hrs too many would be like an extra 4 Hrs per team for the season, or 1 extra Hr per 40 games played.

Your league totals look right on target.

[rasnell]

What I don't understand is why no player during the era of the 1930s hit more than the low 30s in HRs?

I'll try to get you more data.

Understand that I'm talking about Hack Wilson, Hornsby, Ruth, Foxx with HR totals in the 20s and one of them maybe reaching 30. I'll try to post a screen.

Remember that it also threw the PCM numbers all over the board -- almost like the historic modifier settings are fighting against each other and the result is very, very unrealistic HR totals.

[Garlon]

Regarding the distribution of big Hr hitters (fictional players created using the PCM's):

OOTP creates a distribution of great Hr hittes based on modern day distributions. In modern baseball the league HR% from 2005 was 3%, which translates to the average batter ending up with about 17 Hrs per 550 AB. Modern league leaders are around 50 Hrs for the season in roughly the same number of AB's or essentially 3x more often than what the average player does. So when OOTP creates a great HR hitter his power is about 3x better than th average player in the league. Now when you go back to the 1920's and 1930's you can expect that the league leader in Hrs wil be about 3x better than the average player. Now, the 1920's Hr% was about 1.2%, so you can expect the big HR hitters to hit about 3.6% HRs, or about 19 Hrs per 550 AB. In th 1930's the league HR% was about 1.5% so you can expect the league leaders to be around 4.5% Hrs or about 25 Hrs per 550 AB. The disparity between the great hitters and average players of that time are greater than the disparity between the great hitters and average players of today. Your Gehrig's, Greenberg's, and Foxx's of the day only made up about 1% of the players in the league. In modern baeball if the league leader hits 50 Hrs there will also be many players in the league who hit at least 25 Hrs, whereas in the 1920's and 1930's if the league leader hit 40 Hrs there were probably only a couple players with more than half that many.

[Garlon]

Rasnell, if you are running a historical league by importing players from a DB then you do not need to have the PCM option enabled, all you need is the auto-adjust league totals modifiers option. Enabling PCM's shouldn't hurt anything though since that only affects fictional created players, which are not part of your league correct?

Many people have posted their simulated Babe Ruth careers in various threads and many of them have him hitting 50+ Hrs per season and ending up with around 800 career HRs, and usually have Ruth top out with a 70+ Hr season in the early 1920's.

I haven't tested the newest patch extensively yet but I can't imagine anything regarding Hrs would have been changed since the initially release of he game.

[Markmeister]

Rasnall,

I ran a test league while I was gone this afternoon. After a quick glance, league leaders look okay.

I don't remember seeing which database you are using. That might be why you are seeing different results.

[Markmeister]

Garlon,

I'm thinking that in order to simulate a fictional 1920's and 30's, some thinking outside the box may be necessary.

Using the original PCMs is creating "average" players with the correct power for the time period. However, the "average" players for the time period had noticeably less power than the above-average players. Probably the only way to compensate for the difference between the average and above average players is to change the talent distribution ratios. Unfortunately, unless I missed something, that doesn't have a variable that we can change.

I have a couple of ideas how to modify the PCMs and league totals to simulate the era more accurately. Mathematically I don't believe the adjustments will work, but with all of the variables that I don't have (i.e. the distribution ratios for poor, fair, average, good and superstar players) I might be surprised.

I won't have a chance to try until later this week to see if my ideas will work, but I will report back the results.

[battists]

God bless you guys.

This thread hurts my brain, but if anyone ever comes to any conclusions, please let me know...

[rasnell]

I'm just going to manually set the 1930s in the league modifiers and turn off the historical modifier functions in the future. I understand what Garlon is saying about the averages and the extremes, but previous versions of OOTP and other text sim games on the market somehow get the distribution much closer to accurate and give me results like I want.

I'm not getting anywhere close to the results that I want with the default and with the historical modifier button so I'll do it manually. Seems like we've always struggled with getting the default and the historical league settings right in all versions of OOTP and this is no different with the complete code rewrite.

To answer the most recent questions, I was using the plain, vanilla Lahman database and no fictional players.

[Garlon]

rasnell - in the analysis I provided above you were getting league results that were very close to historical totals in HRs. In many of the seasons you were within 2%. The modifiers are calculated mathematically each season from a formula based on league totals from the previous season, they are not set in stone. It is not going to be possible to be more accurate by guesswork. Many people have posted extensive studies comparing their results with historical league totals and the system works very well, in fact it may even be more consistent if you use one of the modified era_stats.txt files that come with my DB.

Try using the Arod23/Garlon2006DB v1.2 instead of the lahman DB.

Did you have injuries enabled or fatigue set to high? I have not seen anyone else report any problems with great HR hitters not hitting HRs. As I have said before many people have posted results with Foxx, Gehrig, Ruth all getting well over 500 Hrs as they should.

[rasnell]

Injuries were set to very low and fatigue to normal.

When the game was tested and default settings and historical modifiers used, do you know if the Lahman database was used for these tests or yours?

In your database, did you delete a lot of the lesser players where you might not have enough to import in a certain inaugural season?

[rasnell]

Here are some more details, as I drill down to the leaderboards in some select years:

1929 HR leaders, both leagues in order:
38, Ott
30, Ruth
30, Wilson
29, Foxx
29, Neun
28, Rice

1931
27, Leonard
25, Klein and Ott
21, Hurst

1933
35, Ott
31, Berger
29, Arlett
28, Gehrig
27, Ruth
25, Foxx

1935
35, Foxx and Ott
33, Klein
32, Greenberg
28, Trosky

NOTE: I'm going to run another historical test league and import Garlon's rosters, as he suggested above. Still curious about whether beta testing of historical sims was based on Lahman or Garlon and if everything seemed to work fine with the historical modifier and default settings.

[rasnell]

I have now done a second historical sim, this time using Garlon's database, and covering 1925-1932. I'm still getting the same unhappy results. If there is not a problem with the game defaults and historical modifiers, then I'm just not going to use them. It swings wildly and gives very poor results. Here are the details:

1925: League mod sets at 1.0; home run leaders are pretty good with Ruth at 48 and Gehrig at 35.

1926: For some reason, the league mod wildly resets to .633 and the HR leaders are about 63 percent of what should be historic. Ruth hit 34 to lead both leagues; Stearns, 27; Gehrig, 25; Bottomley, 21.

1927: League mod, .695; 32 Stearns; 29 Harper; 27 Bottomley; 27 Ruth; 25 Schulte.

1928: Mod .851; 37 Ott; 35 Simmons; 33 Bottomley; 32 Gehrig.

1929: Mod .979: 52 Gehrig; 35 Ott; 32 Klein; 30 Ruth; 28 Simmons.

1930: Mod 1.097: 49 Klein; 44 Ott; 42 Gehrig; 40 Leonard; 36 Hornsby.

I just don't understand the wild automatic fluctuations in the league modifier for homers. If I had manually locked the modifiers at 1.0, I think I would have gotten historic results.

My biggest lack of understanding is about the player creation modifiers which jumped from 1.0 to .495; .506; .519; .528; .531; .530; .534.

Is it true that these are totally irrelevant in a historic league and only influence the creation of new fictional players?

[Garlon]

The PCM's only affect fictional created players. They will not affect any historical players imported from a DB. Also, the PCM's should not have started at 1.00...sometimes they don't take immediately and you hav to uncheck and re-check the box, but this is of course irrelevant in a purely historical league eithout any fictional players.

Here are the actual HR per AB percentages from 1925-1932:

1925: .01367
1926: .01030
1927: .01091
1928: .01294
1929: .01583
1930: .01807
1931: .01234
1932: .01557

As you can see the individual season historical HR percentages were changing from year to year during that period. This is the same type of thing you are seeing in the game.

Only 75% as many HRs were hit in 1926 than in 1925...you got a .633 modifier because your league overshot the historical 1925 ratio so the modifier needed to be lower to hit the 1926 target.

Hrs were hit 5.9% more often in 1927 than in 1926. Your modifier went from .633 to .695. An increase of 9.7% because your league undershot the 1926 ratios.

Hrs were hit 18.6% more often in 1928 than in 1927. Your modifier went from .695 to .851. An increase of 22.4% because your league undershot the 1927 ratios.

Hrs were hit 22.3% more often in 1929 than in 1928. Your modifier went from .851 to .979. An increase of 15% because your league overshot the 1928 ratios.

Hrs were hit 14.1% more often in 1930 than in 1929. Your modifier went from .979 to 1.097. An increase of 12.0% because your league slightly overshot the 1929 ratios.

If you don't want the swinging back and forth as much, you can use one of my custom era_stats.txt files that I included with the DB in the subfolder inside the DB. I included a readme file of how the seasonal expectations were calculated for each one. I prefer using the 5yravg file myself since blocks of 5 seasons will stay very consistent. For example, 1925-1929 all play to the same expectation (the historical average of those seasons combined), then 1930-1934 all play to the historical average of that 5 year span. Every 5 seasons the expectancy changes. Also, my DB includes some Negro League players...in my custom era_stats expectancy files I calculated these players into the league expectancy.

I'm not understanding why you are still unhappy with the results. Your totals look good. First year totals are always a problem. After a few seasons thing really start to smoothen out. Remember that there is not guarantee that players will develop exactly as they should...it looks like in your league Gehrig was a bit more powerful than he actually was and Ruth a bit less, maybe Ruth as on the DL?

Try this, when you create your league set the "base potential ratings" to Remaining Years of Career. Do not generate random left/right splits. Disable injuries too. Set the rookie draft date to Dec 1st. Do not use DH. Set all teams to 4-man, strict order rotations. Set all the general strategic settings in the Strategy and Equivalencies tab to normal, set rotation size to 4-man.

I will start league in 1925 and play thru 1935 and report my results.

[Garlon]

I just ran a test simulation from 1925-1935 using my 3yrwave expectancy file. Here are the top 3 Hr hitters each season:

1925: Ruth 46, Hartnett 45, Hornsby 45,
1926: Hornsby 30, Brooks 28, Stearnes 28
1927: Stearnes 33, Bottomley 31, Hornsby 29
1928: Stearnes 42, Ott 39, Gehrig 32
1929: Ott 61, Klein 56, Herman 46
1930: Leonard 45, Berger 44, Foxx 43
1931: Foxx 55, Ott 48, Leonard 37
1932: Ott 43, Foxx 39, Leonard 33
1933: Leonard 53, Foxx 46, Ott 45
1934: Foxx 46, Leonard 44, Ott 38
1935: Foxx 43, Gehrig 41, Ott 39

I'm not seeing any problems with a lack of players hitting 40+ Hrs. There were even four 50+ Hr seasons in this span. Note, I did set the schedule length to 162 games instead of 154, but an extra 8 games on the schedule wouldn't amount to more than 1 or 2 Hrs even for the best Hr hitters.

[rasnell]

Originally Posted by Garlon

Try this, when you create your league set the "base potential ratings" to Remaining Years of Career. Do not generate random left/right splits. Disable injuries too. Set the rookie draft date to Dec 1st. Do not use DH. Set all teams to 4-man, strict order rotations. Set all the general strategic settings in the Strategy and Equivalencies tab to normal, set rotation size to 4-man.

Surprisingly, my settings matched your recommendations except for the draft date and disabling the injury settings. Perhaps your other database will smooth things out.

I'm not sure how you think my sim and your sim are matching up. I'm not even coming close to historical league-leading totals but you are.

I'm going to try another sim with your database and disable the historic modifiers and simply keep everything at 1.0 to see what happens.

Looking more closely at your 5-year average text file, I like what I see and think that may dramatically resolve my issue. I'll test it and report back.

[Garlon]

Rasnell, do not disable the automatic historical modifiers. Keeping things at 1.0 will give you very poor results after a few seasons. Even using my 5yr average file you cannot keep the modifiers at 1.00, they still need to be recalculated each season because you will always be slightly over or under the expectancy...the modifiers correct for this to keep things on track. Your league will end up completely out of whack after a few seasons. If you ran that simulation into the 1950's with 1.00 as the modifiers you will have guys hitting over 80 hrs per season.

It sounds to me like the problems you were having were that some of the top HR hitters were just getting injured. Keep in mind that league leaders do not tell the whole story either, the entire league HR% should be very close to the historical ratio as well, not to mention all the other batting categories. Modifiers keep all of the stat categories in check.

[rasnell]

I appreciate your patience, Garlon, and I think you've dramatically resolved my issues with what was happening in historical sims. The default game database or era_stats.txt file was simply too erratic for me and each year jumped too severely.

Using your five-year averages txt file and moving it over to be my new game default file is getting the results that you report.

I also ran a similar sim leaving the mods at 1.0 in every category. I'm getting extremely close to the real numbers for league leaders in all the key categories and this has continued from 1925-1931. I'll take your word that this will eventually get entirely out of whack by the 1950s and closer to our current era. I may sim just to find out.

I might still prefer to manually adjust myself for greater satisfaction of what I'm seeing, but your 5-year average text file should clearly be replaced by Markus, Mark, Battists and the gang as the new game default file because it does a much better job of adjusting the historical modifiers. This also will apply to fictional leagues if you choose to use the auto historical modifiers and it refers to this database.

Thanks again for all of your messages and your patience in trying to help me.

I'm now satisfied that there is not a bug in the game related to HRs in historical sims. However, I would recommend that Markus give serious thought to using your 5-year average text file as the new era_stats default for the game. If it's not incorporated as the new game default, all historical simmers should note that Garlon has found a viable solution to the current sim woes.