Players that go way beyond statistical norm

[DetroitStyle]

While there's a lot of discussion on player development, I want to talk about extreme players who have developed.

Specifically I noticed in most saves I get a guy or two who go beyond generational talent, to a point that they break the sim. At least in my head.

Here's Elly De La Cruz:

The level of performance and consistency breaks baseball for nearly every era after the 1950s. Not even Mickey Mantle sustained this.



Here's Michael Harris. This is what I would expect from a generational talent. Not just a star.



But for whatever reason De La Cruz goes above and beyond a generational talent.

So my questions are - what can cause this? Is this a player development issue? If the entire universe is shifting more towards mediocrity (as people have mentioned) is it allowing his 80 rating to dominate further?

For reference this is a near default modern day game with evolution turned off. So this is not influenced by more/less offense.

[DrSatan]

Wait until you get the rando that hits 80+ HRs and 200 RBI's for no reason.

[Rain King]

I'm curious why you would consider those player as "breaking the sim"? How are they different than real life players like Albert Pujols, Mike Trout, or currently Judge and Ohtani?

[DetroitStyle]

Originally Posted by Rain King

I'm curious why you would consider those player as "breaking the sim"? How are they different than real life players like Albert Pujols, Mike Trout, or currently Judge and Ohtani?

Michael Harris, no.

Elly is absolutely breaking the sim. 10+ WAR in 4/6 seasons is insane. It hasn't been done since like Babe Ruth. Even the best of the best of best unicorn today wouldn't be able to sustain that level of production that long.

Not Trout, not Ohtani, not Pujols. Pujols never had a 10 WAR season at all.

I added Harris as an example of what the top of the top unicorn generational player should achieve. De La Cruz is straight up game breaking.

[kidd_05_u2]

Originally Posted by DetroitStyle

Elly is absolutely breaking the sim. 10+ WAR in 4/6 seasons is insane. It hasn't been done since like Babe Ruth. Even the best of the best of best unicorn today wouldn't be able to sustain that level of production that long.

Not Trout, not Ohtani, not Pujols. Pujols never had a 10 WAR season at all.

You have Trout or Judge without injuries. Play with injuries on very high if you want realism.

Also, if you run ten sims and you have a sustained 10 WAR player in 50% of the sims, OOTP has a problem. For just one sim, your Elly is not enough for me to see a problem.

[spartacus007]

Would the sim be more broken if Ruthian players *never* developed? He was a real person, after all.

[uruguru]

Let me provide a theory on why I think this happens.

In the historical game, statistical accuracy is paramount. They derive ratings from players' real-world performance, and then rigorously test the sim by ensure the simmed performance is close to the real-world performance.

Therefore, ratings should track closely to simmed performance, within some variation of course. This is good.

Now, in the historical game, these ratings are recalculated for every season so that the ups and downs of the simmed players match the ups and downs of the real players. Make sense?

Now imagine a fictional player that starts at some lower ratings and then slowly progresses to his potential ratings before they start falling because of age. Now there is variation in that due to injuries or players not reaching their potential, but that's the general gist of how it works.

The issue is that the fictional players progress to their potential far more evenly than historical players. So while historical players may jump up and down from season to season, fictional players will have less season-to-season variation.

So when you have a legendary fictional player with high potential and he doesn't get derailed by injuries or random chance, then he is going to have those high ratings year and year. This is not realistic. And because the simulation is so accurate to the ratings, those superstar fictional players will have mega-season after mega-season.

What is the solution for this? IMO, one possibility would be to create more variability in performance for fictional players (and historical players not using recalc). Obviously changing the engine to have more variability would be like lighting a stick of dynamite, but having hidden negatives (or positives) to ratings on a season by season basis might make the performance of fictional players seem more realistic.

[LansdowneSt]

uruguru,

Accepting the premise, isn't the answer then to make sure you have TCR and injuries at a level you are comfortable with. Sure some guy may thread the needle and never get TCR'ed or hurt but those are customizable settings.

[uruguru]

Originally Posted by LansdowneSt

uruguru,

Accepting the premise, isn't the answer then to make sure you have TCR and injuries at a level you are comfortable with. Sure some guy may thread the needle and never get TCR'ed or hurt but those are customizable settings.

Yeah that's a possibility, too, but the default TCR is already 100. And some guys have really low injury proneness and can skate through most of their career without any serious downtime.

The reality is that "annual ratings reverse engineered from real players" are a LOT more variable than "annual ratings generated by the OOTP sim engine"

And, imo, the OOTP engine is really good at translating ratings into performance.

[NoOne]

looks to be more a problem with league total modifiers and totals.

wrc+ and probably ops+ aren't too whacky and that'd tell you if it is outside the norm.

The exact numbers you see are more about overall league totals and modifier settings... he'll get a particular share of them.. if league-wide numbers are high, he'll get larger numbers but still similar ops+ etc etc.

if his numbers drifted upward without improvement to his ratings, then that points toward 'statistical drift.'

use a backup and test this out so it doesn't ruin anything -- try restoring / renaming the league and run 'calculate modifiers' function in ST or right before opening day. I bet the numbers shift downward. but again, he'll have similar ops+ type numbers. With more offense you tend to have more tiers of stratification. if leaague numbers get too small it can affect distribution of stats.

e.g. distributing 100hr accross 900 players vs 5000hr. With 100hr there's going to be very small differences from high power to lower power in regard to hr output. Proportionally it may be similar, but 5hr vs 9 hr might be the difference from median to elite. So you could probably impact ops+ type stats with extreme changes, but if staying in similar ballpark, should be similar output for various quality of talent.

historic replay and shifting from 'seed' players to fictionally created players are contexts where odd statistical shifts can occur. There are probably other contexts you can extrapolate that this will happen, too. Major shifts in talent distirbution are key. A shift in historic era settings causes it from the other end ...

If that particular player did benefit from TCr, you could re-run that sime several times and it won't happen again, lol. I remember andy dirks turned into a mike trout.. it's not something you can expect, but it certainly isn't unrealistic or gamebreaking. Look at hamilton's early flop or jd martinez's first 1500 PA. gigantic shifts do happen. they are rare.

Don't run that 'calculate modifiers' too often unless you want your statistical variation year to year to be near zero -- for league totals, not individual players, but it does filter down in some ways. 'that' would be unrealistic. With a historical replay, i can't imagine what is required to tightly control and get what you want... with a fictional sim you can zoom out past the seed players, then run some long-term 100-150 year sims... if things are in good balance it doesn't drift. I do things this way. I callibrate it based on high water marks i want to see for individual seasons as well as career numbers after 100+ years. I make sure hr can get into 60's or even low 70 is fine. I want somenoe eventualyl to go over 100sb in a year ... shouldn't happen often but in 100 years shoul see it (personal opinions, here). ...

each to their own as to exactly what they want to see... if sekptical about the statistical outcome, can always sim a year and rview.. be forgiving... oddities happen. you can tell more from overall distribution than focusign on high-water marks of 1 year's results... if significant portions are hitting 40+ hr or batting >.300 and isn't supposed to be that way, then you know something needs to be tweaked before you run that season... each time a historic era institutes new PCM or League totals, you may need to tweak again... each time a major shift like that occurs.. Or, just don't care so much and enjoy the sim... one way isn't 'better'

[jksander]

Originally Posted by kidd_05_u2

You have Trout or Judge without injuries. Play with injuries on very high if you want realism.

Also, if you run ten sims and you have a sustained 10 WAR player in 50% of the sims, OOTP has a problem. For just one sim, your Elly is not enough for me to see a problem.

On high (modern) injuries this guy, Josh Zeno, is the best I have ever seen in my main online league. He was NL-MVP in 2035, 2036, 2037, 2038 and 2039 and finished second in 2040 voting. But I would call him a unicorn, bc he has had just seven injuries, all small, in a league where fragility by age 30 is the norm.

I agree with uruguru that outliers will happen in OOTP, especially when a player gets TCR early AND has very good injury luck.

[jksander]

I have also noticed that WAR is consistently high with defensive wunderkinds who also slug. Zeno, above, has been consistently the NL's best SS since he arrived.