Long Sim vs MLB

[Eiskrap]

Before I dig into a new version of OOTP properly I like to run a number of long sims to get initial impressions of how the universe compares to current day MLB. I thought I would share my analysis here this year in case it is any use to anyone else. The primary things I want to see the impact of are:

* Changing ageing and development modifiers
* Impact of talent randomness
* Impact of more or less offence in League evolution
* Impact of injury setting on ageing

First things first, to view how the game runs 'out of the box' I have run an MLB quick start, disabled league evolution and then simmed through to 2085.

My first comparison is how the distribution of the age of players matches up against what has happened in the MLB. I have exported the csv files and used these to compare the age distribution of innings pitched and plate appearances against MLB data from the Lahman database. The samples I have taken are:

* 1996-2005 from Lahman database
* 2006-2015 from Lahman database
* 2016-2025 from OOTP quick start to see age distribution of original players
* 2051-2085 from OOTP quick start to see age distribution of created players

Results as per the attached images:

Initial thoughts are these are not a bad starting point.

Pitchers: everything seems to be shifted to the right slightly meaning there are not enough young pitchers, so might be worth toying with the pitcher development speed.

Batters: is a bit of the opposite not enough 30+ players resulting in too many 25-28 year plate appearances.

A few versions ago a few of us came to a consensus on ageing and development modifiers in this thread. I'm going to run another sim with these in to see how these change the age distribution.

[koohead]

Love this analysis every year. Thank you (and your CPU!) for taking the time to do this

[NoOne]

definitely keep real and fictional as different data. since the real players are a small sample, i'd just ignore them, tbh. otherwise you have to continually replay the first X number of years and account for the fictional players slowly creeping in yadd yadd yadd... a bunch of crap ruins the data.

don't bother looking at 10-year periods. you want the bulk over a large period of time... 85 years should be fairly precise... still may be a percent or so off though, give or take.

10-year periods are just going to give a glimpse at volatility (or variance if you prefer). it won't help with your initial question. your question is best answered with the entirity of the data (fictional players).

Also, since the RL data is only 10-20 years, you need to assume a larger %error from that data...

the batting looks like it could have a bit added, but pitching looks spot on, relative to RL comparison sample size limitations. do they separate batting and pitching aging in the settings?

maybe favoring veterans in trades would shift it? there could be other settings that can help manipulate that curve.

[Eiskrap]

Right so progress over the weekend:

I tried the ageing and development settings from this thread. I got to 2061 and had a look and these are complete nonsense now, so thats good news a lot has changed since back then. If I can be bothered I'll pull together the graphs at that point to show what impact they have had.

For the next long sim I have tried:

* Batting Ageing 0.600 and Development 1.000
* Pitching Ageing 1.000 and Development 1.400

Results as per the below graphs. In my opinion this has:

* Innings pitched - Aligned OOTP to actuals a little more by shifting the OOTP lines slightly to the left. I think it can still be improved a little bit more. I like to increase the talent randomness and injury frequency modifiers so I'm hoping this might have some impact.
* Plate appearances - Reducing the ageing speed does't appear to have had any affect as line distribution still seems to be in line with the default sim.

Things to try:
* Change AI evaluation weightings
* Increase injury frequency
* Increase talent change randomness
* Tweak ageing and development further

So for my next sim I am going to keep the ageing and development as per this sim, and just change the AI evaluation weightings to 25/40/25/10 and the lineup selection to sabermetric which are my preferred settings.

[sprague]

Your results are similar to what I have found in mine (without the data)
that I make pitcher development .200 more than batting. 1.100 batters 1.300 pitchers
and the same with my aging, batters for me are .500, pitchers are .700

Perhaps I will change the differences a bit more now