This Has My Attention: New Questions

https://365.strat-o-matic.com/league/stats/teams/462895

What has to happen when the hitting runs and pitching runs don’t match? (Total is off by one, 9387 to 9388). I don’t think it’s a nefarious mistake, by the way, but I think we can all agree it’s a mistake of some kind.

I opened this thread because I’m seeing a strange pattern. It is 100% possible I’m seeing pictures in the clouds because of a long, weird streak of numbers. My new Rookie level username can be unluckyman! But other people seem to share in the uneasiness, too. My frustration might sound loud sometimes, esp when it’s so easy to jump to crazy conclusions trying to make it make sense. But is it impossible that something could be slightly wrong in the programming and all this is some unlikely and unintended consequence?

Probably not, as I’m likely just mistaken. But again, I do have questions not outlined in the Wiki pages.

This is what gets me. Strat's engine is like 97% transparent. Its primary alternative, which is 1000x more popular, is 0% transparent.

Yup. In fact, I think the "1000x more popular" is conservative. When I logged in just now to their SOM365 equivalent product, there are 737 leagues and 22,060 teams. And that does not include the tournaments which run daily and weekly at multiple times a day (32, 64, and 128 team leagues that fill quickly) and the "sit n go" type tournaments on a constant ongoing basis as they fill, which are usually 32 team leagues that fill in minutes constantly.

So, something tells me that the transparency angle is not the draw for most people. It is for our tribe here, but I don't suspect that is the path to shangri-la. That is a niche that is shrinking.

I think the draw here is simply nostalgia. And unfortunately, the pool of people who share it is not getting any younger.

You are certianly right on the latter...but on the former, the "other" product also has its nostalgia. I say "other" in fairness to SOM since we are on their boards, and I think everyone here agrees on one thing--they would love this product to thrive.

But on your former statement... all-time greats--old-time classic sets of single year cards, "peak" cards, special edition versions etc. are all in that other product too. The player pool and the play style rules are highly varied. I mean you can set your team setting to saber style or small ball or countless other styles. or tweak individual settings. But how they really work no one knows. They are just magic sliders from never to always.

It should be a very straightforward bell curve with the VAST majority of records within +/- 3 games (per the nerds). Then, for every outlier, there should be a (somewhat) matching outlier on the other tail.

Actually went back and re-read what you wrote J-Pav.
This statement above is not accurate, most especially in the red.

Outliers do not have paired structures. These are not paired events. And there is no such thing as a straightforward bell curve except as a conceptual theory--it is a way to understand "perfect" distributions when you let the prob density function's number of data points/constituents/samples etc go to infinity. It also is free of bias, skewing, multi-modal influences, etc.

We don't have a perfect system--we have all kinds of artificial weird things that will play havoc with a perfected conception of pythag. Which even in and of itself is an imperfect approximation that smoothes out details.

For the same reason that fluid dynamics cannot explain how an electric field forms across a plasma shock boundary, this does not explain anything excpet noise. Certainly your explanation is POSSIBLE, but not expected.

You know, yes the hits, runs, etc don't match up. But occam's razor...do we know how the data is called from a "team page" to a "team stats" page? I go back to the software--it could be an error that is due to using a numerical function that has errors. Seems silly, but geez you can set up a database with an incorrect precision on a field or an argument very easily. Especially if not a skilled database admin.

I haven't checked each team to see if all the data at least correlates to each team's page, and it is weird no matter what. Your explanation is possible, but again there are many possible explanations.

The baseball stat nerds do not agree with your contention:

https://www.beyondtheboxscore.com/2006/4/10/174829/417

Teams won't always meet their Pythagorean fate, but most come close. It's easy to imagine that there is a distribution around a Pythagorean record, and that this distribution has a certain width. Those with a statistics background will recognize that the distribution around the Pythagorean projection will be normal (shaped like a bell-curve) as a consequence of the Central Limit Theorem. The width of this curve will be defined by the standard deviation, which basic statistics tells us will be 1/[2 x sqrt(N)], where N is the number of games.

J-Pav--I don't know who those guys are but I am confident that what I said is true.

Been a physicist for quite a long time, and stats are not closed end "formulas" to me--they are tools derived from calculus used to describe things, whether it is electron energy distribution functions, quantum states, or yes, an average of a group of numbers. We didn't arrive at the central limit theorem a priori. We derive these things. But some forget the assumptions that create the simplified relations, because they are usually close enough. But they are not absolute.

A "bell curve" or gaussian distribution, or normal distribution, is a benchmark for a perfect --fictional--distribution. It is a model we use to make the natural world's problems solvable, but not completely accurate.

We get lazy because we view the neatly packaged formulas that we memorize as some kind of canon, when in fact they are simply the result of applying many approximations. Yes, even the central limit theorem--hence the theorem part--it requires some assumptions that require calculus and then we can agree it is an approximation.
https://mathworld.wolfram.com/CentralLimitTheorem.html
Lots of "ifs" and BTW I still have my hardcopy reference of abramowitz and stegun's Hb of Mathematical Functions for those pesky fourier transforms.

The problem with "imagining" the below...Using the results, theorems, simplified formulas in descriptive statistics or other approximations is, as I said, what you get AFTER you take calculus-based mathematical stats to infinity--in order to create a perfect model that is closed and conceptually descriptive. That is not up for debate.

That statement is missing all the assumptions that are baked into what is needed for a normal distribution.
So in simpler form, can an average of 6 numbers be informative if it is 5 zeroes and a 10? probably not. But if you look at why, it derives from the calculus-based derivation of what a bell/normal distribution is and the only way we satisfy the concept is with a perfect set to infinity.

Basically every conspiracy/black box question is addressed by this section of the help page:

The Baseball 365 game engine will randomly determine whether each individual "at bat" falls on the hitter's card or the pitcher's card. There is a 50% chance of each.

After determining which card the at bat takes place on, the game engine then randomly determines which of the three columns the result will be found on, depending on whether the person being faced (the pitcher when rolling on the hitter's card, for example) is lefthanded or righthanded. Each of the three columns has the same chance of being selected.

Finally, the system simulates rolling two six-sided dice and adding the results (also referred to as 2d6)-- this represents which specific result will be applied to the at bat.

Considering the probability of different results when rolling 2d6, the most common result will be a "7" in any particular column. The next most common results being "6" and "8", then "5" and "9" and so on. "2" and "12" are the least likely results to occur.

Note: The Baseball 365 by Strat-O-Matic game engine is essentially the same as the one the Strat-O-Matic Baseball Windows game uses with some significant improvements. This engine allows the customization of many different optional rules, including those that take advantage of a computer's processing power to improve upon minor limitations of the original cards-and-dice game. Please be aware that some of the play results as shown on the cards may be slightly altered due to these rules. You can read all about which of these rules are being used, their effects, and the improvements we've made to the game engine here.

https://365.strat-o-matic.com/help/rules/baseball#games

That's the answer. The outcomes are determined by the dice with a few exceptions as detailed in the Max Rules wiki. There is no room in that description for any major departure from the dice such as a secret redistribution of wins from veteran managers to rookie managers. So the problem is not that Strat has failed to address these questions. The problem is that people disbelieve Strat's statements on the issue. But there is nothing further Strat can say to convince the doubters, because you can't reason someone out of a position they didn't reason themselves into. If I thought there was any statement Strat could make that would satisfy these folks then of course I would hope that they would make such a statement. But I do not think that any such statement would be found satisfactory. Truly, what more is there for them to say apart from mounting that section of the help page on a big sign and tapping the sign whenever the question comes up? https://i.imgur.com/dgObTK5.jpg

Firstly, let me start off with a mea culpa for trying to rationalize what might be happening. I shouldn’t wonder out loud, if you know what I mean, even just for the purposes of open discussion. I care zero about the “black box” probably in exactly the same way the players of the other game care zero about the black box.

But I am facing an extremely weird distribution of numbers that has been going on for a very long time. Despite Frankie’s argument, there must be a bell shaped curve around the pythag results, at least eventually. I do not think I should be two, three and four plus standards of deviation off on the loss side, and zero to the win side. There is an argument to be made that “very long time” has not been long enough for the tails to equal out.

I guess more time should tell then.

I do agree that the distribution should normalize after a long enough time simply because a run scored is someone else's run allowed. But yeah I don't know enough about math to say what a long enough time would be. And I also think it's possible that certain playstyles and strategies could push individual managers to one tail or the other, despite the overall distribution for all managers being normal. Anyway I don't think you have anything to apologize for, it's been an interesting discussion.