Avoiding statistical drift

[treedom]

Hey guys. I'm looking for some help. I'll go into a bit more depth below, but basically the heart of my question is, how do I avoid the downward offensive drift I'm seeing in my fictional league? Please see screenshot, which shows the league ERA and AVG going down gradually over the span of 18 seasons.



I know it's something to do with the "League Totals", but they kind of confuse me. I tried reading the manual first and it says:

OOTP generates a league total for each category, noted in the left column. The league total acts as the basis for the calculation engine. In historical leagues, these are the real league totals from the imported year. In fictional leagues it's the major league totals from the most recently completed season.

So about that last sentence, it's not what I'm seeing. My league totals are the exact same as those entered when I created the league. (I based them on the 1985 season using the dropdown.) Did I forget to change a setting?

Anyway, the manual goes on to recommend that I use the "Modifiers" in the right column. Okay, fair enough, except...given that the values in "League Totals" never changed, but the actual output of the engine seems to be drifting, I'm not sure what the modifiers are applied against. Are they applied against last year's stats or against the static "League Totals" from 1985 I still have there?

The way I see it, there's a distinction there. Say I have modified HRs to be 10% more. Is that a one-time bump, or will they go up by 10% every season if I leave that modifier there?

Anyway, I'm even confusing myself as I write this. I'm probably just overthinking it. Please help me see straight.

[actionjackson]

Originally Posted by treedom

Hey guys. I'm looking for some help. I'll go into a bit more depth below, but basically the heart of my question is, how do I avoid the downward offensive drift I'm seeing in my fictional league? Please see screenshot, which shows the league ERA and AVG going down gradually over the span of 18 seasons.



I know it's something to do with the "League Totals", but they kind of confuse me. I tried reading the manual first and it says:



So about that last sentence, it's not what I'm seeing. My league totals are the exact same as those entered when I created the league. (I based them on the 1985 season using the dropdown.) Did I forget to change a setting?

Anyway, the manual goes on to recommend that I use the "Modifiers" in the right column. Okay, fair enough, except...given that the values in "League Totals" never changed, but the actual output of the engine seems to be drifting, I'm not sure what the modifiers are applied against. Are they applied against last year's stats or against the static "League Totals" from 1985 I still have there?

The way I see it, there's a distinction there. Say I have modified HRs to be 10% more. Is that a one-time bump, or will they go up by 10% every season if I leave that modifier there?

Anyway, I'm even confusing myself as I write this. I'm probably just overthinking it. Please help me see straight.

I think that if you want the league totals from 1985 to work you need to apply them the day before Opening Day every year. What I mean by this is that you should select 1985 in the drop down menu every year, or if your input year is always set to 1985, hit auto-calc the day before the season starts. That's what I do, and I use 1984 which is pretty similar. I've done this with a random debut historical league that has run from 1901 through 1941 and the outputs I get are very reasonable to the point where the MLB AVG and ERA are very close to what they were in 1984 when you look at the 41 seasons and average it out. They can vary quite a bit from season to season, but they average out to within a point of the 1984 AVG (.260) and 1984 ERA (3.81) last I checked. I don't know if it makes a difference that yours is fictional and mine is historical because I've never really played fictional.

I'd also be careful about messing around with modifiers too much because you may get more of a change than you're looking for in the opposite direction, which may bother you just as much. It's a fine line with these things. I've come around to just selecting 1984 in the drop down the day before Opening Day and letting the chips fall where they may. Hope this helps.

[jpeters1734]

for some reason with default fictionals, the league will always drift. the league average will start around .250-.260 and will drift down to .235-245

I've seen my minor leagues drop to .200-.210!

[NoOne]

starting before 2016 had an odd effect for me this year - even after the patch notes said they fixed something related... it would use or influenced by historical eras even if you didn't implement them through team strats and LT/LTMs.

so, assuming that's not real.... make sure Player creation modifiers are all 1.000 (or whtever for 1985 - just assume this from now on if i speak of 2016 values). make sure your league totals are copied in there from 1985. hit auto-calc modifiers on opening day. no matter what's going on, this will force the players that are in your league to proportionately have a probabilty to get their share of stats relative to their ratings. as i explain below, you still get quite a bit of fluctuation from autocalc because it's purposely slightly inaccurate or has some element of random plus/minus to various results. (taking an average of many "autocalc" clicks will eliminate that randomness, if you want)

the seed players in a newly created league are slightly different in (distribution and age? not sure how best to describe) than what results from drafts, and all the other ways players are created and lost over time. so, the first 20 or so years you may see a steady change or drift. auto-clacling on opening day will completely avoid this during that transition.

how long can a career be? 18-40ish? so ~22+ years for all of the seed players to be fully replaced by players created per year.

After that you should merely fluctuate based on randomness of newly created talent reachign the majors - some drafts are better than others right? so, if you get enough of the same luck year after year you will have a boom or bust period on occasion. you might get lucky have have 100 years of fairly consistent results too.

anyway, once you reach that point, you can stop Auto-calcing the league total modifiers, if you want. the problem at this point is whether the last time you clicked autocalc the league was somewhat average in nature relative to what i just described about randomness of quality of newly created players per year.

you can either eye-ball it over time and get fairly close, continue to autocalc each year which will stil have some fluctuation (you notice if you click it multiple times it has different results - it's not perfect on purpose), or you can run some long-term sims at this point on the side and base soem LTM's on ~100years or data or more (or less, but less accurate if you go less).

easy way - click autocalc one year, then let it do a few seasons with those LTM and keep an eye on leageu-wide season stats and make sure they aren't 50%? or more different from what you want (relative to the stat at hand and something extreme - if the results are an outlier for the intended results you want, you know that it needs some adjustment).

clicking autocalc a bunch of times and averaging the reuslts will very liekly result in the league totals, proportionately to your league. that doesn't mean that you will get them for certain, but it can't be far off under these circumstances.

figure BA can sway +/- ~.15 from the mean you want, due to various levels of talent in your league - while maintaining the same LTM (modifiers).

so .235-.250-.265 roughly... probably a bit less than that, even. if you see a .280 or a .220 result, you know you need to adjust the hits LTM a bit or reduce K's or something that causes more hits to occur. even reducing walks LTM could increase batting average in some situations.

if you make your own LT, as opposed to copying some from a 1985 created league, make sure it all adds up properly - like babip jiving with certain other totals etc etc... a spreadsheet makes this simple all you have to do is put in the formulas for the LT that are actually caclulated by some of the other LT that are used. (* you'll need to know a rough average of sac hits or whatever from your league results - just be in the ballpark, off by10-50 is no big deal when workign with 165k ab.)

[sprague]

In these cases auto-recalc before the season is your friend

[jpeters1734]

if you auto calc, you get guys hitting 50-60 HRs every year

[David Watts]

Originally Posted by actionjackson

I think that if you want the league totals from 1985 to work you need to apply them the day before Opening Day every year. What I mean by this is that you should select 1985 in the drop down menu every year, or if your input year is always set to 1985, hit auto-calc the day before the season starts. That's what I do, and I use 1984 which is pretty similar. I've done this with a random debut historical league that has run from 1901 through 1941 and the outputs I get are very reasonable to the point where the MLB AVG and ERA are very close to what they were in 1984 when you look at the 41 seasons and average it out. They can vary quite a bit from season to season, but they average out to within a point of the 1984 AVG (.260) and 1984 ERA (3.81) last I checked. I don't know if it makes a difference that yours is fictional and mine is historical because I've never really played fictional.

I'd also be careful about messing around with modifiers too much because you may get more of a change than you're looking for in the opposite direction, which may bother you just as much. It's a fine line with these things. I've come around to just selecting 1984 in the drop down the day before Opening Day and letting the chips fall where they may. Hope this helps.

Mr Jackson taught me this method a ways back and it works great.

[Markus Heinsohn]

Originally Posted by jpeters1734

if you auto calc, you get guys hitting 50-60 HRs every year

Nope. Depends entirely on the totals you enter

[jpeters1734]

Originally Posted by Markus Heinsohn

Nope. Depends entirely on the totals you enter

i dont enter totals. i know that i shouldn't bring up statistical anomalies if it's not default.

[jpeters1734]

or very high 40's sometimes. rarely do you get low 40's.

I guess my point is that it's far too easy for a guy to hit 60 HR's all he has to do is play in an era with fewer power hitters and his number will spike.

Quick sim on default for 13 seasons

[jpeters1734]

here's another quick sim of 25 years on default.

[treedom]

Thanks guys. I'll try auto-calc at the beginning of each season and see how it goes. I've been afraid of that button till now.

[actionjackson]

Originally Posted by treedom

Thanks guys. I'll try auto-calc at the beginning of each season and see how it goes. I've been afraid of that button till now.

Make sure to do it the day before Opening Day.

P.S. I checked over my 41 year old random debut historical league after posting above, and the batting average over 41 years is .260 (should be .260), while the ERA is just barely rounded up to 3.82 (should be 3.81) for that time. Remember that even a season is a small sample size in the grand scheme of things due to variances in talent level in your league. It should all shake out in the end though. Again, neither sure if my using historical is what's giving me these solid results, nor if using fictional is what's causing your league to drift, but that's been my experience so far.

[NoOne]

Originally Posted by jpeters1734

if you auto calc, you get guys hitting 50-60 HRs every year

is the league grossly over it's HR total form league totals (maybe ~700-1000 above or so)? if the values are 'roughly' legaut totals, then it's the total that is causing too many 50-60hr guys. if you adjust HR you much adjust other LT so that babip and ?other things? calculate out correctly.

autocalc can result in a range of results... some clicks will have extra offense, some clicks will be fairly average, other clicks will result in lower levels of offense. otherwise the people that click autocalc all the timewould have virtually no fluctuation in league-wide results (not necessarily individual players, but a rising tide lifts all boats).

maybe the last few clicks you've gotten some 'better' autocalced values relative to offense.

[NoOne]

Originally Posted by actionjackson

Make sure to do it the day before Opening Day.

P.S. I checked over my 41 year old random debut historical league after posting above, and the batting average over 41 years is .260 (should be .260), while the ERA is just barely rounded up to 3.82 (should be 3.81) for that time. Remember that even a season is a small sample size in the grand scheme of things due to variances in talent level in your league. It should all shake out in the end though. Again, neither sure if my using historical is what's giving me these solid results, nor if using fictional is what's causing your league to drift, but that's been my experience so far.

opening day works, too... i wait until opening day for more accurate preseason predictions too.



you know for certain the 25-man rosters are the ones they will use - maybe it's based onthe 40-man, but jsut in case opening day avoids the concern. i see distinctly different values when i autocalc for my minors too far before opening day.

if you watched the ai... does it set everythgn the day after ST ends or does it set rosters for its org. on opening day? whenever it completes this task, it is safe to click autocalc

[NoOne]

Originally Posted by treedom

Thanks guys. I'll try auto-calc at the beginning of each season and see how it goes. I've been afraid of that button till now.

after 20-25 years, or as you approach that mark, it will be less important to do it each year...


if after that transition period you still see steadily rising (or decreasing) values wihtout reaching a ceiling, then it's something else causing it.... if you are mostly defualt, i wouldn't worry about this, though.

as long as everyrthing has hit an equillibrium state (player creation / player loss etc) then it will merely fluctuate around a specific point. At this point clicking autocalc isn't necessary if it's at least clost to callibrated correctly for your LT. if you don't like the averages it's fluctuating around you need to adjust the LTM or the LT (* the modifiers are easier to work with, and each time you changs the LT you must autocalc the modifiers, anyway). but, at this point once you adjust them you can leave them.

just the logic of a simulation makes it a bit wonky to force the stats to be somewhere near average, instead of actually fixing the cause of it. there's nothing you can do the frist 20-25 years with real players, but after that there's not reason not to fix it. i'd rather the the game's player creation and aging and development result in the fluctuations i see... not changing some modifier constantly (by hand or autocalc). *this has nothign to do with a change of era or tangible reasons to change LTM/LT - including roleplay reasons.

[swampdragon]

? It looks to me like offense has held steady since 1939.

[actionjackson]

Originally Posted by swampdragon

? It looks to me like offense has held steady since 1939.

Steadily down though. He's using 1985 totals for his league and that year the MLB batting average was .257 and the ERA was 3.89. It should be closer to those numbers than it has been. The batting average averaged over 18 seasons is roughly .244. The ERA over that time is right around where it should be though at roughly 3.88 to 3.89, so that's good. That probably means either slugging or on base (or both) are up to make up for the 13 point gap in batting average between what he's seeing so far and what 1985 actually produced.

[jpeters1734]

I know that I can adjust the league statistical output how ever I see fit. I'm specifically talking about default values

[RchW]

Originally Posted by NoOne

opening day works, too... i wait until opening day for more accurate preseason predictions too.



you know for certain the 25-man rosters are the ones they will use - maybe it's based onthe 40-man, but jsut in case opening day avoids the concern. i see distinctly different values when i autocalc for my minors too far before opening day.

if you watched the ai... does it set everythgn the day after ST ends or does it set rosters for its org. on opening day? whenever it completes this task, it is safe to click autocalc

I was sure that the MLB autocalc button disappears on opening day. I do the minor leagues after MLB opening day to get the AI after it has parsed it's minor league stock.