Historical Simulation Accuracy

tayloraj · 08-13-2018, 07:52 AM

I've been running a couple of historical simulations lately with no recalc and talent change randomness set to about 60. I was noticing anecdotally that offense seemed strangely low (on the left) and took a look at the simulation accuracy page to find that, for some reason, run scoring has run pretty consistently 5-8% below historical norms while most of the other stats have been much closer to real life. Thinking this might be caused by using Real Historical Minor Leagues, I tried a game without them(on the right), and have gotten pretty much the same result.
I've been running all strategy/league total settings as per normal. Does anyone have an idea what might be causing this?

scott1964 · 08-13-2018, 07:54 PM

Here is mine, my runs per game is all over the place.

Reed · 08-13-2018, 09:26 PM

I have noticed it with every league I have tried-- Real Stats, Neutralized stats, 1 yr recalc, 3 yr recalc., etc.. I guess it could be several things such as weaken/adjust hitters pitchers, pitcher settings in Stats & AI, etc.. I also believe it has something to do with questionable baserunning/stealing by the AI. I just accept it as my baseball universe which is slightly different than MLB.

tayloraj · 08-14-2018, 07:42 AM

From what I've seen with some of my leagues that have run longer, it seems the game does a much more accurate job at mimicking run scoring from the 1950s onward; I guess the more modern home run and stikeout:walk ratios that became prevalent that decade fit more neatly into the game engine?
It's strange, because I don't recall having this particular issue with older editions of the game. Could just be my faulty memory.

NoOne · 08-14-2018, 08:44 AM

those numbers are from different eras and lumped together to form some average.

even if they attempt to neutralize for different factors from teh different eras, it still isn't perfect.

if #'s are 'off' to your eye, go ahead and manipulate the LTM for that stat. you can get it the way you want. if it's consistentlyt 5% below, go bump that LTM a few percent... think of how that affects other things, though. if you increase walks, that reduces AB which reduces h,2b,3b,hr by a small amount. think that increases babip by reducing the denominator? you get the point.

i would suggest not worrying so much about this too much if only ~5% or less. or, i guess i'd ask a question. which do you care about more? the league-wide stats year to year and cumulative or individual player results? if it is the latter, i'd set stats based on how individual perform with little to no care about the total league-wide results - within reason. if it's far from default, i'd be surprised

tayloraj · 08-14-2018, 08:34 PM

The issue I'm seeing is that almost all the inputs that make up run scoring appear to be very close to accurate - I did a pretty deep dive into 1921 before posting this thread, and in terms of BA, 2B, 3B and HR, BABIP and SH everything was pretty much right on the money versus what actually happened in 1921 (BB was a little bit low, but not by much) ; and yet teams were scoring 8% or so fewer runs than they really did. I didn't check other years quite as extensively, but they were consistently scoring below historical norms.

08-13-2018, 07:52 AM	#1
tayloraj Minors (Double A) Join Date: Mar 2012 Location: Philadelphia, PA Posts: 162	Historical Simulation Accuracy I've been running a couple of historical simulations lately with no recalc and talent change randomness set to about 60. I was noticing anecdotally that offense seemed strangely low (on the left) and took a look at the simulation accuracy page to find that, for some reason, run scoring has run pretty consistently 5-8% below historical norms while most of the other stats have been much closer to real life. Thinking this might be caused by using Real Historical Minor Leagues, I tried a game without them(on the right), and have gotten pretty much the same result. I've been running all strategy/league total settings as per normal. Does anyone have an idea what might be causing this? Attached Images

08-13-2018, 07:54 PM	#2
scott1964 Hall Of Famer Join Date: Oct 2005 Location: Denver Posts: 5,561 Blog Entries: 19	Here is mine, my runs per game is all over the place. __________________ This just feels more like waiting in line at the Department of Motor Vehicles. PETA.....People Eating Tasty Animals.

08-13-2018, 09:26 PM	#3
Reed All Star Starter Join Date: Oct 2013 Posts: 1,942	I have noticed it with every league I have tried-- Real Stats, Neutralized stats, 1 yr recalc, 3 yr recalc., etc.. I guess it could be several things such as weaken/adjust hitters pitchers, pitcher settings in Stats & AI, etc.. I also believe it has something to do with questionable baserunning/stealing by the AI. I just accept it as my baseball universe which is slightly different than MLB.

08-14-2018, 07:42 AM	#4
tayloraj Minors (Double A) Join Date: Mar 2012 Location: Philadelphia, PA Posts: 162	From what I've seen with some of my leagues that have run longer, it seems the game does a much more accurate job at mimicking run scoring from the 1950s onward; I guess the more modern home run and stikeout:walk ratios that became prevalent that decade fit more neatly into the game engine? It's strange, because I don't recall having this particular issue with older editions of the game. Could just be my faulty memory.

08-14-2018, 08:44 AM	#5
NoOne Hall Of Famer Join Date: Apr 2015 Posts: 7,167	those numbers are from different eras and lumped together to form some average. even if they attempt to neutralize for different factors from teh different eras, it still isn't perfect. if #'s are 'off' to your eye, go ahead and manipulate the LTM for that stat. you can get it the way you want. if it's consistentlyt 5% below, go bump that LTM a few percent... think of how that affects other things, though. if you increase walks, that reduces AB which reduces h,2b,3b,hr by a small amount. think that increases babip by reducing the denominator? you get the point. i would suggest not worrying so much about this too much if only ~5% or less. or, i guess i'd ask a question. which do you care about more? the league-wide stats year to year and cumulative or individual player results? if it is the latter, i'd set stats based on how individual perform with little to no care about the total league-wide results - within reason. if it's far from default, i'd be surprised

08-14-2018, 08:34 PM	#6
tayloraj Minors (Double A) Join Date: Mar 2012 Location: Philadelphia, PA Posts: 162	The issue I'm seeing is that almost all the inputs that make up run scoring appear to be very close to accurate - I did a pretty deep dive into 1921 before posting this thread, and in terms of BA, 2B, 3B and HR, BABIP and SH everything was pretty much right on the money versus what actually happened in 1921 (BB was a little bit low, but not by much) ; and yet teams were scoring 8% or so fewer runs than they really did. I didn't check other years quite as extensively, but they were consistently scoring below historical norms.