the catcher lead-off & other bad ideas

[Fishbreath]

Originally Posted by NoOne

i have my doubts about whether this game adheres to the law of independent results, too. however, it's possible that the 'rolls' were already made prior, so it only seems odd.

i think the general outcome of a single AB gets put into certain profiles and those adhere to probabilities, but that isn't the same nor is it a good practice, if it is done that way.

i saved a game before an AB, made a backup and exited the game. i restored the backup numerous times for that AB and got the same result 100 times over (no exagerration and excluding fielding errors, those are calculated after a ball goes into play).

it would be impossible for any data for an ab that never took place to reside in the backup. it was a fresh start to that AB everytime.

i was trying to get a 4hr game in ootp15, it happened during an early ab and i only tried that one 20-30times before i moved on, then it happened again 2 ab later. that AB i redid 100+times out of spite. i have a decent pc, but restoring a game 100times still took a heft chunk of time over 2 sittings, lol. i just couldn't believe that the result was predetermined.

luckily i don't play day-to-day so it isn't a concern for me anymore. even if the stats are somewhat shaped, they still fluctuate somewhat realistically from year-to-year. instead of tuning their engine, i think they took the easy path to make it realistic.

edit: i don't see what realstar sees though. there is no punishment by not subbing a catcher in late innings. this isn't as common as you believe it is. also, since the ai follows the same logic, it is an even playing field. they aren't making that substitution either. the whole immersion stuff is just personal opinion.

it's possible you start winning when the ai runs stuff becuase it is actually making better decisions relative to your team. if the talent distribution on your team fits their strategy better, you will perform better, and vice versa. also, watching a game or 50 isn't going to give you an answer one way or another. even a season. replay that season over and over again with each strategy - turning off trades, development and other factors that cause deviation not related to strategy. control all other factors and change the ai only. then you'll know better about how the varying ai strategies influences results.

Computers don't generally have access to purely random numbers: it takes specialized hardware that most home computers don't have. (Some operating systems use tricks to generate true random numbers from a variety of sources, but those aren't important here.)

To make random numbers, a computer uses a mathematical process that generates numbers that meet mathematical tests for randomness. These processes are both stateful (they have internal values which control their outputs) and deterministic (for a given set of internal values, their outputs are always the same). Random number generators have a seed number, an initial state for which their outcomes are always the same. If the game saves the seed, then the results of a hundred attempts at a given at bat might well be the same: the random numbers generated are always the next few pseudorandom numbers from a given initial state, and therefore the same every attempt.

[NoOne]

Originally Posted by Fishbreath

Computers don't generally have access to purely random numbers: it takes specialized hardware that most home computers don't have. (Some operating systems use tricks to generate true random numbers from a variety of sources, but those aren't important here.)

To make random numbers, a computer uses a mathematical process that generates numbers that meet mathematical tests for randomness. These processes are both stateful (they have internal values which control their outputs) and deterministic (for a given set of internal values, their outputs are always the same). Random number generators have a seed number, an initial state for which their outcomes are always the same. If the game saves the seed, then the results of a hundred attempts at a given at bat might well be the same: the random numbers generated are always the next few pseudorandom numbers from a given initial state, and therefore the same every attempt.

i'm not sure why you bring up random numbers, that's not really anything i brought up, so i'll skip that. a RNG is an imperfect thing, not really relative though to what i said. except for the possible explanation for the behaviour i used... which isn't a good thing. i can only hope i was wrong.

Regardless, the AB hasn't occured yet. no pitch thrown, nothing. i'm pretty sure i even tried saving in previous inning, but i'd hesitate to be sure. i encountered this more than a few times while screwing around. i tried all sorts of things to get around it - likely saving long before they came to bat. now i'm thinking of testing it again, but this is a new versio, too.

that function in the code shouldn't have even been called yet, because nothing has happened to cause a calculation based on ratings, probabilities of all the various factors they decided to use to determine results.

it shoudln't invovle rng - at least not in a significant way in regard to results. it should be a chain of probabilities. some static, some based on the player, weather, and whatever. even pitch selection should be based on the last pitch and the idea of changing speeds or not being predictable - that shouldn't use an RNG, if they tried to make it realistic... because pitches aren't randomly choosen. i'm having trouble thinking of anyhting that should be randomly choosen that would be such a strong factor to cause 100+ consecutive outs for a .500 batter?

in re to my random profile theory - that would be more of a 1-5 case select (or more), using a random 1-5 as an input. that's a total guess to explain my results. if that is the way, it is a very unsophisticated way to determine results. if that is the case, what you said about rng could hold true. i really hope this isn't how it works. but that info shouldn't stay in across innings of resimming each time you restore.

really glad i don't manage day-to-day in this game anymore, lol. i'll be blissfully ignorant just watching history pile up that at least looks okay, even if the means by which they were created is a bit suspect.

[NoOne]

Originally Posted by Orcin

Do you still have this save available?

i don't but i am curious again. i removed ootp15 a long time ago.

i'm going to load up a backup in ootp16, make a 250's player and try again for all HR in AB. i won't upload personal data, but i'll post screenshots.

process to employ:
i'lll save immediately after an ab, and make a restore point. that way it should avoid any 'seeding' problems that were brought up... i'll make sure he never makes it up a 2nd time in an inning by taking pitches with players after him.... so it will always be a new inning, too. and each restore will have to sim through the rest of the inning as well as another half inning at minimum, and more likely multiple innings before he is up to bat again.

no excuse of leftover data causing it to happen, which i am fairly sure is not the cause at this point.

when i'll do it is up in the air... maybe if it eats at me some more.

[Dyzalot]

I think you misinterpreted what he was saying. It is quite possible that the game chooses a new "seed" at the beginning of each inning or at the beginning of each AB. Then as "random" numbers are needed, say for a ball hit into play to determine the result, it goes down the list of numbers for that "seed". This would appear as random unless you saved a game at that point of the seed being generated and then ran the sim for that AB multiple times. Not saying that is definitely what happens with OOTP since I have no idea of the specifics in how the game generates random numbers but it is one possibility that might explain what you were seeing.

[NoOne]

Originally Posted by Dyzalot

I think you misinterpreted what he was saying. It is quite possible that the game chooses a new "seed" at the beginning of each inning or at the beginning of each AB. Then as "random" numbers are needed, say for a ball hit into play to determine the result, it goes down the list of numbers for that "seed". This would appear as random unless you saved a game at that point of the seed being generated and then ran the sim for that AB multiple times. Not saying that is definitely what happens with OOTP since I have no idea of the specifics in how the game generates random numbers but it is one possibility that might explain what you were seeing.

first: i started to test it again, after restoring 10 backups and reloading 10 times, i am not going to care about whether it's true or not, lol. when i saw that occur it was just screwing around when i first got the game.

what you described is exactly how i interpretted it.

this is all hypothetical, i do not know for sure what's actualyl going on, but it raises a red flag - that's not debateable for me.

the fact an AB could be determined the moment they step up to the plate is the problem, and if true, it's obvious how unsophisticatedly they determine a result of an individual AB. there is no factor at that point that should cause and absolute result.

e.g. it would be similar to using just their average the moement the stepped up to determine a hit or not. infinitely worse if it's just some randomly generated number that cuases it.

RNG's could be rationally involved in the various equations and factors and their associated probabilities and how previous events affect teh next one (markov chain). however, they should not be so strongly weighted or so prominently used to be significant enough to cause 100+ outs.

e.g. determing a gust of wind could rationally use an RNG. applying personality and it's affect to the player could make use of an RNG - e.g. +/- some % to ratings or whatever baed on chemistry or any other rational factor..

take a moment to ponder:
100+ straight groundouts or flyouts, lol. what factor, if locked in by random seed before the ab occurs, could cause that result each time and be so much more important than how the AB played out after that point? what factor before an ab starts could be so strong it causes the result to be absolute and unavoidable?

i'm pretty sure i see the website he is nearly copy/pasting from, lol. 20+ years ago i learned multiple languages. some old as hell like fortran77 and cobol, some more modern like java and c#. i never had all that "visual" crap... just a handy text editor and CLI compiler for me! i actually would use the visual xxxx studio stuff, nowadays, lol. it has benefits.

[Dyzalot]

Originally Posted by NoOne

take a moment to ponder:
100+ straight groundouts or flyouts, lol. what factor, if locked in by random seed before the ab occurs, could cause that result each time and be so much more important than how the AB played out after that point? what factor before an ab starts could be so strong it causes the result to be absolute and unavoidable?

Let me make sure I understand the test. You are simming, right? Therefore if a seed number is generated the moment a player steps up to bat then every "random" number needed for that AB is already generated and known. So if pitch by pitch then every decision even down to what pitch is thrown and what strategies to employ that involves some sort of randomness to it is already known and used in the same order for every test. Therefore it will play out exactly the same every time. If using "one pitch" mode then less variables but same process such that every time a "random" number is needed it will be known & used in same order in every test. Again, this is only supposing that this is how the RNG works in OOTP.

[NoOne]

Originally Posted by Dyzalot

Let me make sure I understand the test. You are simming, right? Therefore if a seed number is generated the moment a player steps up to bat then every "random" number needed for that AB is already generated and known. So if pitch by pitch then every decision even down to what pitch is thrown and what strategies to employ that involves some sort of randomness to it is already known and used in the same order for every test. Therefore it will play out exactly the same every time. If using "one pitch" mode then less variables but same process such that every time a "random" number is needed it will be known & used in same order in every test. Again, this is only supposing that this is how the RNG works in OOTP.

i was playing out a game, ab by ab. leave game and make a restore point before starting an AB. always loaded from restore, not the original save after it crashes (try leaving a game, make a backup, then when you try to re-enter the game you just left, the game crashes on you. so, i was never reloading a crashed game. it was always loaded from a backup each attempt.)

i have no idea what was going on. all i know is that nothing generated before the player does anything should result in 100+ consecutive outs. that's all that matters. if that is what was happening, they are not allowing each factor of an ab add up on it's own.

the problem is not that RNG are involved, the problem would be that the RNG is determining things in an absolute nature before anything of note occurs. other factors need to be considered before the final probability of a result is determined.

if all the random stuff is generated before the ab takes place, that would be fine as long as it doesn't determine the at bat fully.

simplified: ignoring pitch selection, wind, whatever... per pitch:

rating of XX and pitch thrown equates to a 25% chance of making contact.. etc... the end result is a bottom line percentage, which then can be used in the final RNG (maybe out of 100, 1-25 is a positive, while other returns are failures), which then another equation or rng would determin in play or not, then where in play or out of play. you could argue that it's rational the first pitch is done the moment you step up, but the next one should not be. it's broken down into individual quantifiable parts and should be calculated each time a pitch is thrown anew. chance of getting contact, chance of in play, where in play, chance of fielding error, and so on through the process of a play as it is determined in real time.

even if pitch selection is part of the predetermined stuff... there are enough other factors that the bottom line probabilities should be recalculated each pitch and law of independent results would reign in an decent simulation.

it's just putting variables into an equation and calculating a result. e.g. that 25% chance i used would be the result of what pitch was thrown, the ratings of the player, the count, the choosen option for that player. baseball is actually a fairly easy simulation to code - it just depends on how sophisticated you want to be about it. it's simple code you can learn at a community college class, but it would be time-consuming to do it well.

the bulk of the equation should have little to do with an RNG until a final aggregate probability is calculated. only at the end of all of that do you have a % you can make use of a RNG.

even if some %s remain constant due to a non-random seed, the equation should still be calculated as you hit '1' or '2' ont he keyboard with the possibility of different results. that is the law of independent results.

if could be an anti-cheat mechanism, too.... who knows. it would be easy to recognize the baheaviour.

[Dyzalot]

Maybe not explaining it properly. Let's assume it seeds the RNG at beginning of each AB. Let's also use a "D100" for simplicity's sake although I assume it is a larger "die". So AB starts, it seeds the RNG and you get 73, 42, 13, 89, 55, 49, 95, 65, etc. Those would be the numbers used for the "RNG" every time if you run the sim for the same AB over & over again. So the first time it needs a random number it uses 73, the next "random" decision uses 42 then 13 then 89, etc. So assuming you aren't changing any variables yourself due to different strategies in the AB or different pitch selections, you will get the exact same result every time because it will use the exact same "random" numbers every time and in the same order for each decision. It isn't that it is deciding the result of the AB before it starts it is just that it is deciding the exact "rolls" of the dice in advance for each time it needs to do so.

[NoOne]

Originally Posted by Dyzalot

Maybe not explaining it properly. Let's assume it seeds the RNG at beginning of each AB. Let's also use a "D100" for simplicity's sake although I assume it is a larger "die". So AB starts, it seeds the RNG and you get 73, 42, 13, 89, 55, 49, 95, 65, etc. Those would be the numbers used for the "RNG" every time if you run the sim for the same AB over & over again. So the first time it needs a random number it uses 73, the next "random" decision uses 42 then 13 then 89, etc. So assuming you aren't changing any variables yourself due to different strategies in the AB or different pitch selections, you will get the exact same result every time because it will use the exact same "random" numbers every time and in the same order for each decision. It isn't that it is deciding the result of the AB before it starts it is just that it is deciding the exact "rolls" of the dice in advance for each time it needs to do so.

i realized this should be highlighted again: from ootp 15, not 16. it could be different now and would be an easy change. i mentioned it once, but it should be carried over.

i see what you mean, now.

i witnessed it, so it must be the case:
they must be creating 1 random number for each probability caclulation/category to start an AB and applies to all pitches in that ab afterward (in ootp15)... which when it returns an extreme value, you will have a predetermined result that neither pitch selection nor strategy can affect. that would explain what i saw.

that is a very poor way to do things. that's flipping a coin once, when it should be flipped numerous times for each new event with previous events affect the future (markov chain) to acount for change of speed or working in and out on the plate etc.. this can have a major impact on results. at minimum it should occur per pitch, if they don't want to acknewledge the influence of the previous pitch and the sophistication required to do so.

when the rng for where it is hit at determines it's a point in the field that is 100% an out, like right at the first basement, then that ab is over with before you do anything - even if the third pitch thrown should more likely be hit in the air, or inside therefore is more likely to be pulled than any previous pitch. also, if a pitcher throws three exactly same pitches in the same location, the player will not pick up on this, because it's already determined to be a strike out or whatever, lol. if that still works that way... wow. how unsophisticated.

****edit: just realized that the out was all over the field.. most of the time it was a ground out to a similar spot on the field, but it did vary a bit, if i recall.... hmm i guess if the flyouts were in the same direction and an of is there, it would be the same results... i don't rember if it oscillated left/right.

hmmm... i'm thnking of testing some 2-pitch pitchers as SP, because i don't think changing speeds or moving the ball arund the plate has any real influence on results under these conditions. so a monster stuff/movement CL with stamina should be an amazing SP, too. one more reason among many why it would be a very very terrible practice in a baseball sim to do it this way.

if the rng for where in the field the ball goes directs it at the first basemen, there is nothign you can do about it. my rng for x meant it was always an out in-play, no matter what i choose to do during the ab... even when a new pitch and a new environment is in place due to what happened previously.

you might as well not worry about strategy at that point and just click 1 on one-pitch (the one where you hit it once, unless it's take a pitch) mode with a few take pitches mixed in to run up the count.

only when the value returned is near a borderline for success/failure will see different results each time you replay an ab.

luckily that only affects day-to-day players, which i don't do anymore. so as long as it looks accurate for players and league-wide, it's not too big of a deal, i guess.


edit: lol "a c c o u n t" is being filtered as a bad word. see the asterisks? that's not me. okay, now it isn't asterisked, but it was, i swear... but i don't curse swear.

[NoOne]

well a 2-pitch reliever (with 125/250 stamina) is a good starter as long a they have high control like a typical ace starter.

if it's 110/250 control with elevated stuff and movement of a high-caliber reliever, they are pretty crappy.

sean doolittle had a 3.41 era pitching 200ip and a 125era+.

so it seem control becomes a more important if you have fewer pitches, is what i would surmise, for a starter.

he was fifth in ERA for the AL. led the league in QSP at .688.

doolittle is a low mph guy, too. the worst stuff of the bunch i tested. and a lefty, which in general results in worse pitching stats.

aroldis chapman on the other hand with much lower control did not fair as well. netiehr did the other ones who also had lower control.

so, 2-pitches influences results, but if they have 70/80 control and are a high-end talent reliever, they will likely be a cy young candidate in your league with enough stamin.

i'm gonna run one more year and change all control equal to doolittle.


edit: looks like a 1/5 chance for a great season and 4/5 for a crappy one, lol. so my theory isn't right aobut 2-pitch pithers.

[drowssap]

I know Jason Kendall was mentioned but John Jaso batted leadoff for a while in Tampa Bay.

[Dyzalot]

Originally Posted by drowssap

I know Jason Kendall was mentioned but John Jaso batted leadoff for a while in Tampa Bay.

Craig Biggio was a prototypical lead off hitter and came up to the majors as a catcher. So he's another one.

[bwburke94]

In the divisional era, only Kendall and Jaso have led off 50+ times. Third place is Butch Wynegar at 48.

[joefromchicago]

I think the unstated assumption here is that a person will always beat the AI - at least over the long run. I don't think that's necessarily true. Although I don't usually play out my games, I create the lineups and pitching rotations, make all the substitutions, and set the strategies for my teams, and, when it comes to those sorts of things, I consider myself somewhat smarter than the average bear. Nevertheless, I've found the AI to be a formidable opponent. I think most of my success comes as a GM rather than a manager - in short, when I'm successful it's because I put together a team that can win despite my managing.

Rather than anecdotal evidence, I think the only way to test Realstar's theory is to pit two identical teams against each other - one managed by the AI and one managed by a person - and see if one wins significantly more than 50% of the time over a run of, say, 50 or 100 games. Only then can we determine if the AI is penalizing a player for managing (or over-managing) his team.

[RchW]

Originally Posted by joefromchicago

I think the unstated assumption here is that a person will always beat the AI - at least over the long run. I don't think that's necessarily true. Although I don't usually play out my games, I create the lineups and pitching rotations, make all the substitutions, and set the strategies for my teams, and, when it comes to those sorts of things, I consider myself somewhat smarter than the average bear. Nevertheless, I've found the AI to be a formidable opponent. I think most of my success comes as a GM rather than a manager - in short, when I'm successful it's because I put together a team that can win despite my managing.

Rather than anecdotal evidence, I think the only way to test Realstar's theory is to pit two identical teams against each other - one managed by the AI and one managed by a person - and see if one wins significantly more than 50% of the time over a run of, say, 50 or 100 games. Only then can we determine if the AI is penalizing a player for managing (or over-managing) his team.

I think the experiment has been done already on these forums and the results (more than a decade's worth) clearly suggest that playing out games is advantage human. Probably 75% of threads related to AI and game results seek help in making the AI better and the job of the human more difficult. The two most recent versions have improved measurably and there are many posts confirming that improvement. However we are not yet at the end of the rainbow.

I think the OP's more recent "conspiracy" or fantasy allegations are complete nonsense, supported by zero evidence and verging into "truther" like territory.

[NoOne]

Originally Posted by RchW

I think the experiment has been done already on these forums and the results (more than a decade's worth) clearly suggest that playing out games is advantage human. Probably 75% of threads related to AI and game results seek help in making the AI better and the job of the human more difficult. The two most recent versions have improved measurably and there are many posts confirming that improvement. However we are not yet at the end of the rainbow.

I think the OP's more recent "conspiracy" or fantasy allegations are complete nonsense, supported by zero evidence and verging into "truther" like territory.

you'd have to avoid doing certain things like the take pitch exploit. i've heard of players essentially wiping out a teams pitching staff in the first 2 games of the series (or even during the 2nd game), and the opposing team just gets obliterated every final game of the series. just as one example.

sometimes people conduct an expirement like that to prove themselves right... others properly employ relevant aspects of the scientific method.

without evidence to support your opinion, it is just as valid as teh OP's or anyone else's. sometimes it's not about opinon, but rather curiousity. curiousity is a virtue.

[RchW]

Originally Posted by NoOne

you'd have to avoid doing certain things like the take pitch exploit. i've heard of players essentially wiping out a teams pitching staff in the first 2 games of the series (or even during the 2nd game), and the opposing team just gets obliterated every final game of the series. just as one example.

sometimes people conduct an expirement like that to prove themselves right... others properly employ relevant aspects of the scientific method.

without evidence to support your opinion, it is just as valid as teh OP's or anyone else's. sometimes it's not about opinon, but rather curiousity. curiousity is a virtue.

I can show way more evidence than the OP. It's recorded right here and in thousands of seasons in games posters play. Read the threads.

Oh and I've yet to see an attachment or a screenshot supporting any recent claim made by the OP.

[NoOne]

Originally Posted by RchW

I can show way more evidence than the OP. It's recorded right here and in thousands of seasons in games posters play. Read the threads.

Oh and I've yet to see an attachment or a screenshot supporting any recent claim made by the OP.

i'm sure your approval weighs heavily on his mind.

[RchW]

Originally Posted by NoOne

i'm sure your approval weighs heavily on his mind.

You're the NoOne asking me for evidence. Why not ask for evidence from all claims made?

Edit: are you saying that playing out games is not advantageous to the human player?

[NoOne]

Originally Posted by RchW

You're the NoOne asking me for evidence. Why not ask for evidence from all claims made?

Edit: are you saying that playing out games is not advantageous to the human player?

At no point did i request any such thing. you've made something up to support something wholey in your head.

i made it clear i am not saying one thing or the other.

i don't have the hubris to assume one thing over the other with any confidence about the OP's hypothesis without objective experimentation. incredulity is irrelevant, if there is any. whether or not i think it's happening is not a logical argument against or for testing it. the only way to know for sure is to apply the scientific method and test it.

in a controlled testing environment, you will learn infinitely more than trying to mine data from these threads/blogs/whatever from leagues that are hardly equivalent environments with all sorts of significantly different settings and eccentricities of their owners that will cloud the results and more likely to lead you to believe things that simply don't exist. any one person's experience with ootp is significantly esoteric. no human can simply read over a few blogs and threads and be able to answer the op's original question with any confidence, whatsoever.

the op showed humility by saying he is not sure if it is true to start. rational people have said the same while believing the opposite of the OP. one exceptionally rational person even mentioned that the only way to know is to test it (not me, i suggested a general method with some qualifications that human mangers shouldn't use exploits in the hypothetical testing, which was inspired by that post).

also, the bold sentences are about as obnoxious as using capital letters to emphasize something. sometimes i feel like i have just finished an interesting article of news and then i read one of your replies, and it's as if i was teleported to a yahoo comments section for a politically themed news item. you have no interest in a discourse that contradicts your narrative about the game.