Alternate Reality: A Little Flat

[Cobby]

I've just finished the 2017 season in a fictional alternate reality simulation that started in 1871. (I've been working on this for 9 months or so). I was going for a fictional simulation that still nevertheless kept fairly close to how baseball developed IRL. I used historical financials, historical player creation modifiers and did the auto-recalc each year to keep the stats close to real life.

Now it's time to see how the two worlds compare statistically. I'm kind of thinking I'll do this with a handful of different threads.

This is the third thread on the topic. Previously I've mentioned that I had no great modern pitchers. I did have several great hitters - but none comparing to the truly great real-life hitters. No Babe Ruth, no Ty Cobb, no Ted Williams, no Willie Mays, no Ted Williams, no Honus Wagner. But I did have a few that compared fairly well with the next tier of real-life players. Following the careers of the great players is what I enjoy most about doing this sim. So I was a little bit disappointed at the lack of any of those top-tier players in my sim. In general it felt a little bit flat for hitters and very flat for pitchers.

So, how do I address this in the next sim? I think I can save NoOne the trouble of posting his answer to that question: I had yearly auto-recalc on. That brings the stats in line with the historical results. But if the stats are being skewed by an exceptionally good player - supposedly, unskewing via auto-recalc has the effect reducing the impact of that exceptional player. Or something? I'm not quite sure I buy that - but I could be convinced, I suppose.

It seems to me that the key characteristic of an exceptional player is that they have ratings considerably higher than their peers and thereby get a larger percentage of the statistical pie on average. So, if I want more exceptional players I need to increase the standard deviation of players in the league. There doesn't seem to be an obvious parameter for that.

I have a couple of ideas though. One is to decrease the number of draft rounds - perhaps coupling that with increasing the number of international scouting discoveries. International scouting discoveries are usually of lower quality. If you could figure out the right balance, possibly you could lower the mean quality of players in the league while still keeping the best players from the top of the draft. Those players would then be further above the mean than they would otherwise and have better careers. Maybe something like that could work.

Or maybe I could start tweaking the player creation modifiers every year. You'd probably want to create worse players for a couple of years and then make up for it with better players for one year, then repeat. Maybe you'd get an exceptional player in the good year that really stood out more given the less than stellar players from the previous two years.

Anyway. I'm interested in ideas about this and what people with more experience have discovered.

[NoOne]

here's any easy/funcitonal way to look at it.

*** how each LTM changes is inverseley related to that LTM/LT. ***

*assumes previous auto-calc.

* 2 forces at play -- rise and fall of pitching and/or batting talent each year at time of clicking auto-calck (snapshot of time).

so if most of the offensive LTM drop: it's possibly any combination of the 2 forces that add up to the resulting change.

it won't neccessarily tell you it's "more" talent than other times in history. if i said that, i mispoke. easy to "see" though. you'd have to simply make a high-end filter and count how many players there are. with a good baseline you know about current talent levels. league-wide, by position, just depends on how dirty you want to get.

their competitors will always have somethign to say about what their stats look like.

flat results -- Auto-calc less often.

when era changes, players from preivous era remain at same ratings? if true, it's similar for turning over the league with seed players. once the new "era" players start hitting the MLB it's a good idea to auto-calc 2-3-5years depending on where you are at on the curve.

you can wait longer at the 'ends' and want to click more often in the 'middle' when the most change is occuring due to new players. -- that's for league-wide statistical "reining" without making it too flat. trial and error to taste.

based on what you said, you may want to do it in a reactive way. ie once a few key stats hit a particular level you click it. (too high or too low). whatever trend starts, it's going to go that direction for as long as you ran the new PCMs for the draft and ~20years for league turnover context.

1 player point of view..

don't make fewer players. that will have a bad effect, unless you have an extreme # of players being created currently. don't starve your minors, a little more is good too.

playing with PCM can do what you want too. it's a rising tide. it will shift any baseline related to stats that rating plays a role in. if you don't auto-calc, you can use this to eye-ball how how individual resutls get (remember 1 year is nothing, but if the league never breaches ~40hr, then in first year these "new" pcm-created palyers hit league 1 hits 50, you know something is up.

obviously quite a few lag years to see PCMs make their influence known on a league-wide manner, but individual stats will shift accordingly, if you do not auto-calc since the PCM change -- 100% gauranteed with time.

*assumes no other oddity like "seed" players or in midst of changing eras -- all similar contexts. that would get dicey choosing when to auto-calc and when to let it roll, maybe?

the more tightly you want to control your league-wide stats may be the bigger cause of your obstacles. (not necessarily, though)

starving talent may help a slight amount, but it will make the extreme talent hitters/pitchers 1/50years instead of 1/10 or whatever you saw before. (30T, 44rounds, 60int'l ama. FA, fewer than default other stuff -- i get at least 1-2 amazing fictional batters every ~20 years. ~180+ contact and power out of 200 type players of various positions.)

@ a little less than 5000hr baseline league, i've had 1 guy hit 60HR. 50 wasn't as rare, but fairly rare, i'd have to look.

based on each era's LTs, you can 'see' what a peak # will be with a little bit of time. i'd use your current sim as a little research for this knowledge because you auto-calced so often.

if you want more doubles in a particular era, remeber that, take note, or something more general etc

what you want may take a bit of organization ahead of time. the constant mix of fictional and historic players may limit viable choices in how much to manipulate stats through ltm/pcm etc.

anyway, you could use the history and check leaderboards of the long-run sim you already have... look into specific eras... what needs to be changed the entire time vs. what you want to modify per era. you can easily make a map, and a little trial and errror at various points you implement things over time.

you will get different league-wide stats if you change PCMs or the ltm/lt's after auto-calc, obviously. key to remember here is that the RL results were just 1 possible result of many. Trying to stay ~near it will make you go crazy if too rigid about it. may need to make compromises between what you want to see on the leaderboard per player and what you want to see in the league history.

can't affect std. dev, but you can shift it left or right. with larger numbers you do get larger ranges of variances, but same % of baseline -- at least anywhere near a ~4.00ERA type league with ~modern-ish power.

it's amazing, actually... bumping league-wide baseline and creating enough-to-a-little-more players each year can actually create what you want to see, i promise. greater range in individual results at the cost of a slightly higher league baseline.

ps, i have a spreadsheet somewhere in forums that randomizes PCM within a stated range. i made 2 i think, one that deviates based on the previous year and one that just does it each year randomly accross the board. you'll have to search. google with "site:www.ootp...com" their base url appended to pcm spreadsheet ?

i hammer one set of LTM out for a baseline and use that. i was going to use this to implement some additional volatility in the results. i know better about myself... i'd change it one year then 20 years later get twisted about why hr are so low or too high... absent minded.

[Cobby]

Thanks for your detailed reply. There's a lot to think about there.

I did do some thinking about recalc - and I played around with that log5 equation a bit - and now I can see how recalc can flatten results. The whole thing is an interesting interplay between the talent pool (affected by player creation modifiers) and the expected results historically. Ideally, one would never recalc and have the whole thing driven by the player creation modifiers. I'm thinking now, for a period where there were abrupt changes (like the transition from the dead ball era) that it's best to let it be driven by changes in the talent pool. I'm thinking the recalc should be used mainly after things start to stabilize - i.e. in the middle of eras rather than at the ends and beginnings. Although, as you say, it could also just be driven by when you notice that things are getting a little out-of-whack.

Abandoning the excessive re-calcing might be enough in itself to give me what I'm looking for.

But I also wonder about things like injury rates and player development and aging. I think these are calibrated to modern baseball - and the default values do well for the modern era. But I suspect they have also changed through time - I know that has to be true for injuries. We can fix a lot of things these days that we couldn't fix before.

Most (not all) of my discrepancies center around careers rather than seasons. Not completely, but for instance my season home run totals are eerily similar to real life totals. On the other hand, I only had one player hit over .400 and that occurred in the 1800's. But, some of my player's best seasons matched those of the all-time greats - it's just that they didn't have careers that matched them. If just reining in the auto-recalc doesn't help this - perhaps I can give a tiny little tweak to player aging in some eras to lengthen careers a tad and to give my best players a couple of extra good years. That might be enough too.

Anyway. I guess I have some experimenting to do.

[NoOne]

" Ideally, one would never recalc and have the whole thing driven by the player creation modifiers."

I think i love you, in a platonic way of course.

same with LTM/LT. the totals and mods essentially lay out physics of the game world... how much power equates to X newtons of force which equates to Y-hr/year on average. (not speaking of code, just general logic).

by changing ltm/lt's you change how many Netwons equate to 1 Power. that's just not right! obfuscates feedback too -- not unlike "relative" rating systems.

that's nearly forced to use now, and default in '19. otherwise RP look like all <50/80 players, spare a few.

had they went the other route and affected player create ratings through PCM, it would be so much more intuitive and you'd rarely have to change your LTM/LT -- wouldn't even be needed.

i'd love in addition to PCM to affect the distribution and range of deviation. that would be the most intuitive way... you put in a baseline environtment, ~max deviation sought, a few distribution options, and "it" calculates the values you need behind-the-scenes.

So, an era changes, a new set of PCM kick-in according to what you just plugged in, and those players start being created for drafts... slowly they infiltrate and change the league statistical environment... driven by what seems to be the cause in RL too, lol... not God changing our LT/LTM values.

it could be how the game actually functions. it's certainly not a scentifically modelled sim. it's partly for entertainment and leisure so some things are skewed that way.. and just common sense of an indie budget. i wouldn't doubt some patches here and there to "keep it in line" so to speak, lol.