Does the AI actively try to get no hit?

[Dyzalot]

Originally Posted by TomVeal

It took me a while to come up with a real-world explanation for leaving in a starter who was being pummeled. I suggest that you look into whether your manager has a gambling problem and has placed heavy bets on San Diego.

Interestingly enough the manager (me) made two substitutions before that game in order to replace a couple of slumping regulars even though we won the first two games of the series. After that game the manager reinserted the regulars back into the lineup and we have since gone on to take that series 4-1 and now have a 3-0 lead in the World Series. I don't know what the explanations would be but I'd love to be at that post-game press conference.

[Dyzalot]

Originally Posted by TomVeal

BTW, there's no way to tell from the screen shots when Severino left the game, except that it was after the grand slam that put the Padres ahead 9-0. Not removing him after he gave up the first five runs isn't preposterous, and I could see letting him pitch with the bases loaded. Even though he wasn't doing well, the manager might have confidence in his ability to get out of the jam. (So far as I know, OOTP, unlike, say, APBA, doesn't downgrade a pitcher's ratings when he is shelled.) Did the AI manager take him out immediately after the grand slam? If so, gambling may not have influenced his decisions.

Yes the AI removed him right after the slam.

[Westheim]

Originally Posted by Dyzalot

But would you pinch hit for your best hitter with the backup catcher? Look at it as a whole. Also not sure what you mean by "pen fresh" since someone has to pitch. It is just the absurdity of it all. You'd never see a regular season game, never mind a playoff game, play out this way in the modern era.

I probably would not pinch-hit for Guerrero there, but the game was totally over by then. And it is not like Guerrero broke up the no-hitter when he had the chance.

And you are right in that somebody has to pitch. In this scenario it should be the least-impressive reliever or the former #5 starter to soak innings, regardless of the outcome, including taking at-bats. Does it matter whether you lose 9-0 or 14-0? 9-0 or 9-1?

I am pretty sure if we had tons of time we would find instances of relief pitchers batting in MLB playoff blowouts. It's not like it never happens in real life, nor that the game was on the line here, tied at two in the eighth...

The AI will sometimes do weird things. So do MLB managers. So do OOTP players. Two seasons ago in my dynasty I tried to nail down the title in Game 6. Up 11-6, I sent my closer into the bottom of the ninth and he crumbled. 11-8, two out, tying run at the plate, a left-handed pinch-hitter. I had my lefty specialist still available, brought him in, and he got the final out. Would the AI do this? Would any MLB manager do this? Can we be sure there is a right or a wrong move in this context. I think the guys on sports radio will tell you whether it was a right or wrong move the next afternoon...

Was sending the reliever to bat in the original scenario here a right or a wrong move? I say no. The game was outta hand and the AI protected the pen for Game 4.

Was pinch-hitting for Guerrero in the original scenario here a right or a wrong move? Weird for sure; wrong certainly not.

Originally Posted by Dyzalot

One last point. What you would do in that situation is irrelevant. You aren't a MLB manager and thus any data you provide doesn't apply when determining the range of possible decisions made by a simulated MLB manager.

As has been pointed out to you before in this thread, if you don't want any uneducated opinions, don't post in the first place. If all you want is ferocious agreement from everybody around you, please print out the box score, pin it to the wall, record yourself ranting about it, then replay the tape and vividly agree to the guy on tape - 'Damn yeah, that guy's right! How can they ...!?' - since it seems your own voice is the only one that matters to you in this context. Just keep in mind that you are not a MLB manager either and thus your opinion does not matter, either.

Also, play out your playoff games!

[phenom]

I think Markus should rewrite OOTP in its entirety. It's obviously horribly flawed based on this one game.

[BirdWatcher]

Originally Posted by phenom

I think Markus should rewrite OOTP in its entirety. It's obviously horribly flawed based on this one game.

I see what you did there.

[Matt Arnold]

My take is that it was a little early to throw in the towel pinch hitting for the top guy, but I don't think it's out of the question to do that in their last AB of the game.

[Dyzalot]

Originally Posted by Westheim

I probably would not pinch-hit for Guerrero there, but the game was totally over by then. And it is not like Guerrero broke up the no-hitter when he had the chance.

And you are right in that somebody has to pitch. In this scenario it should be the least-impressive reliever or the former #5 starter to soak innings, regardless of the outcome, including taking at-bats. Does it matter whether you lose 9-0 or 14-0? 9-0 or 9-1?

I am pretty sure if we had tons of time we would find instances of relief pitchers batting in MLB playoff blowouts. It's not like it never happens in real life, nor that the game was on the line here, tied at two in the eighth...

The AI will sometimes do weird things. So do MLB managers. So do OOTP players. Two seasons ago in my dynasty I tried to nail down the title in Game 6. Up 11-6, I sent my closer into the bottom of the ninth and he crumbled. 11-8, two out, tying run at the plate, a left-handed pinch-hitter. I had my lefty specialist still available, brought him in, and he got the final out. Would the AI do this? Would any MLB manager do this? Can we be sure there is a right or a wrong move in this context. I think the guys on sports radio will tell you whether it was a right or wrong move the next afternoon...

Was sending the reliever to bat in the original scenario here a right or a wrong move? I say no. The game was outta hand and the AI protected the pen for Game 4.

Was pinch-hitting for Guerrero in the original scenario here a right or a wrong move? Weird for sure; wrong certainly not.



As has been pointed out to you before in this thread, if you don't want any uneducated opinions, don't post in the first place. If all you want is ferocious agreement from everybody around you, please print out the box score, pin it to the wall, record yourself ranting about it, then replay the tape and vividly agree to the guy on tape - 'Damn yeah, that guy's right! How can they ...!?' - since it seems your own voice is the only one that matters to you in this context. Just keep in mind that you are not a MLB manager either and thus your opinion does not matter, either.

Also, play out your playoff games!

Why are you being obtuse? Since when has this forum turned into a place where we get called out for challenging uneducated opinions? Would you rather that the uneducated opinions go unchallenged so that new members see them and think they are true? That seems like a disservice to all the newer players myself. Sorry, but you don't get to dictate under what terms I am allowed to post. And it isn't my own voice that only matters on this issue. If you had read my posts then you would know what my point is. The point has nothing to do with how I would manage the game personally if I were into "mix-maxing". However, I am trying to play a simulation of modrrn MLB baseball. In order for that simulation to be accurate, the sim should model the behavior of MLB managers fairly well. My whole argument is that such a model is lacking. Your input or anyone else's input as to how they would manage is irrelevant to that problem unless the poster was or is a MLB manager. However, your input into providing data or opinion on how MLB managers actually manage is very welcome.

[Dyzalot]

Originally Posted by Matt Arnold

My take is that it was a little early to throw in the towel pinch hitting for the top guy, but I don't think it's out of the question to do that in their last AB of the game.

I agree. I'd also like you guys to look at the starting pitcher AI. It appears to me from just pure observation that the AI way over values pitch count as compared to performance when deciding to pull a starter. It seems like the AI is finE with a starter giving up 7+ runs in the first few innings as long as the pitch count stays under 70 or so. I don't think that is an accurate representation of how MLB managers manage in real lofe, especially for the playoffs. I feel like performance is valued much more by real MLB managers than the AI in OOTP currently does. A real MLB manager will stretch a starter over 100 pitches if pitching well in a shutout but the AI ill often take that guy out as if he's on a putch count. Vice versa a real MLB manager will likely pull his starter if he's already given up 5 runs in order to pinch hit for him at the next opportunity. This would be especially true in the playoffs where bullpens are deeper and there is more rest per week for the pitching staff. Just from the point of view of looking at the strategy settings logically, I have a hard time justifying why the AI left the starter in to give up 9 runs when the starter usage strategy is set to have a "quicker hook than average".

[Dyzalot]

Originally Posted by Westheim

Does it matter whether you lose 9-0 or 14-0? 9-0 or 9-1?

To a real life manager it matters because it will affect his life. The press and pressure from the fans will be different if you lose 9-2 or you get no-hit 9-0. So yes, if nothing more than for the morale of your team, it matters. Does it matter in OOTP? Nope. And that's my issue.

[Westheim]

Originally Posted by Dyzalot

I am trying to play a simulation of modrrn MLB baseball. In order for that simulation to be accurate, the sim should model the behavior of MLB managers fairly well. My whole argument is that such a model is lacking. Your input or anyone else's input as to how they would manage is irrelevant to that problem unless the poster was or is a MLB manager. However, your input into providing data or opinion on how MLB managers actually manage is very welcome.

Maybe you should print out the box score 30 times, mail them to each MLB club, and seek input from Bochy, Maddon, and Cora then. They are the only ones that seem to be able to help you.

Regardless of the input you desire, you are still comparing apples with oranges. Not sure how many MLB managers have played OOTP before, or how many OOTP forum members have managed MLB clubs before. Either way, an MLB game is not the same as a game in OOTP, and especially not if you sim it and complain about it after the fact.

[Dyzalot]

Originally Posted by Westheim

Maybe you should print out the box score 30 times, mail them to each MLB club, and seek input from Bochy, Maddon, and Cora then. They are the only ones that seem to be able to help you.

Regardless of the input you desire, you are still comparing apples with oranges. Not sure how many MLB managers have played OOTP before, or how many OOTP forum members have managed MLB clubs before. Either way, an MLB game is not the same as a game in OOTP, and especially not if you sim it and complain about it after the fact.

I have no idea what you are talking about. We don't need to talk to MLB players in order to model them correctly and neither do we need to talk to MLB managers to model them correctly. How many managers have played OOTP is as relevant to the game as how many baseball players have played OOTP, it isn't. I have no idea how you think a sim can be developed if you think you have to talk to the people you are simming in order to produce the software. That's just ridiculous and 20 years of OOTP proves you wrong.

[Dyzalot]

Originally Posted by TomVeal

Ah, ha! Reading the tiny print of the screen shots (not easy for eyes as old as mine), I learn that "C. Bellinger" is Cody's brother Cole, whose 2018 performance at Short-Season A wasn't spectacular (1-6, 5.43 ERA). It's good to see his improvement between now and 2025.

In case you were curious...

[TomVeal]

Originally Posted by Dyzalot

I agree. I'd also like you guys to look at the starting pitcher AI. It appears to me from just pure observation that the AI way over values pitch count as compared to performance when deciding to pull a starter. It seems like the AI is finE with a starter giving up 7+ runs in the first few innings as long as the pitch count stays under 70 or so. I don't think that is an accurate representation of how MLB managers manage in real lofe, especially for the playoffs. I feel like performance is valued much more by real MLB managers than the AI in OOTP currently does. A real MLB manager will stretch a starter over 100 pitches if pitching well in a shutout but the AI ill often take that guy out as if he's on a putch count. Vice versa a real MLB manager will likely pull his starter if he's already given up 5 runs in order to pinch hit for him at the next opportunity. This would be especially true in the playoffs where bullpens are deeper and there is more rest per week for the pitching staff. Just from the point of view of looking at the strategy settings logically, I have a hard time justifying why the AI left the starter in to give up 9 runs when the starter usage strategy is set to have a "quicker hook than average".

That is, I think, the nub of the issue that you identify. In real life, players abilities vary day by day for all kinds of reasons. In OOTP, they change only slowly over time. A pitcher who has given up five runs and loaded the bases is still just as capable in the OOTP as he was before he took the mound. His performance is no indicator of how well he'll pitch to the next batter. Therefore, the AI sees no reason to bring in a reliever.

In real life, a manager will figure that there's a reason for bad performance other than random bad luck. In the game in question, I'm pretty sure that most managers would have yanked Severino after he put men on base in his final inning. They would have figured that he was having an off day and not pitching up to his usual ability.

APBA, the first baseball sim that I ever played (is it still around?) had a fix for the unrealistic constancy of ability. If a pitcher was doing badly, his rating dropped; if brilliantly, his ratings increased. Sabremetricians will scoff, but the effect was to make players' decisions more like those of real managers.

[Dyzalot]

Originally Posted by TomVeal

That is, I think, the nub of the issue that you identify. In real life, players abilities vary day by day for all kinds of reasons. In OOTP, they change only slowly over time. A pitcher who has given up five runs and loaded the bases is still just as capable in the OOTP as he was before he took the mound. His performance is no indicator of how well he'll pitch to the next batter. Therefore, the AI sees no reason to bring in a reliever.

In real life, a manager will figure that there's a reason for bad performance other than random bad luck. In the game in question, I'm pretty sure that most managers would have yanked Severino after he put men on base in his final inning. They would have figured that he was having an off day and not pitching up to his usual ability.

APBA, the first baseball sim that I ever played (is it still around?) had a fix for the unrealistic constancy of ability. If a pitcher was doing badly, his rating dropped; if brilliantly, his ratings increased. Sabremetricians will scoff, but the effect was to make players' decisions more like those of real managers.

Yeah that would seem to be one of the problems in modeling real life behavior into a sim, even one as constrained as baseball. The model uses randomness to feel like the ups and downs, hot streaks and cold streaks that players have whereas in real life these can often be explained by a minor injury, maybe a recent failure in technique or even just the mental state of the player on a given day. Those kinds of things can be revealed to the manager through the poorer performance resulting from one of those factors. In OOTP the manager just attributes a poor outing by a pitcher or a few bad games by a hitter as just statistical variance because in truth, that's all it is in OOTP. So if you wanted to "mix/max" your gaming experience to have more of an edge over the AI, you ignore performance and rely on the underlying ratings to win out over time. However, if you are looking for more of a "role playing" experience then you would want to see the AI use performance as a metric to determine when a pitcher doesn't have his "stuff" or maybe when a hitter has picked up a bad habit in his swing technique and then try to address that by benching them, demoting them, etc.

I can look at this also from a gaming perspective and see the dilemma for the developers too. If you go the route of the "role playing" experience where you are trying to model all of the actions as close to real life but need to use randomness to do that then the player can take advantage of the AI by doing what is called "mix/maxing" in the RPG communities. So you may end up with people claiming the AI is too easy to beat if the AI sees the randomness as a clue about an underlying cause for a performance drop and the player ignores it because he understands the underlying mechanics of the game and is more interested in winning than reproducing an accurate representation of the MLB.

Maybe we are at the point in the development where difficulty levels are needed. Perhaps one of them could be a "role play" difficulty where the player would expect to adhere to certain standard MLB conventions in order to still have a competitive battle with the AI and the AI would act more in accordance with what would be expected if one were to watch modern MLB every day. I understand the difficulties involved in all of this. At its core OOTP is attempting to simulate human action. Even within the confines of a rule set such as baseball it still leaves what seems like an infinite set of possibilities at hand for the AI to deal with. I just get frustrated when I see things happen in OOTP that never happen in MLB, especially when it is on the scale of things like starting pitcher usage and the lack of a change in strategy for how a bullpen is used in the post season.

[Westheim]

Originally Posted by Dyzalot

I have no idea what you are talking about. We don't need to talk to MLB players in order to model them correctly and neither do we need to talk to MLB managers to model them correctly. How many managers have played OOTP is as relevant to the game as how many baseball players have played OOTP, it isn't. I have no idea how you think a sim can be developed if you think you have to talk to the people you are simming in order to produce the software. That's just ridiculous and 20 years of OOTP proves you wrong.

You already said before that input from common strangers as to the behavior of MLB managers is irrelevant. Therefore, all you can hope for is input from MLB managers. That is all I suggested to you.

And good luck with that.

Originally Posted by TomVeal

That is, I think, the nub of the issue that you identify. In real life, players abilities vary day by day for all kinds of reasons. In OOTP, they change only slowly over time. A pitcher who has given up five runs and loaded the bases is still just as capable in the OOTP as he was before he took the mound. His performance is no indicator of how well he'll pitch to the next batter.

This is not true at all. (Yes, the AI will often be too tardy to replace a falling pitcher; that is not the point I disagree with)

You can see the writing on the wall so often with pitchers, they are quite like an open book. If you played enough games you can even get a hunch for when a guy is about to come up with an injury. If a guy has a bad day, you can often see that right from the start in terms of what counts he gets in, what contact the opposition generates off him, and so on. It is not often that they stop sucking suddenly after getting into an early shredding.

Of course, for that you have to play out your games. Do you play out your games?

[Dyzalot]

Originally Posted by Westheim

You already said before that input from common strangers as to the behavior of MLB managers is irrelevant. Therefore, all you can hope for is input from MLB managers. That is all I suggested to you.

And good luck with that.

No I did not. Or if I did, please provide the quote where I said such a thing.

[Westheim]

Originally Posted by Dyzalot

No I did not. Or if I did, please provide the quote where I said such a thing.

Not word for word, but surely implied strongly here:

Originally Posted by Dyzalot

One last point. What you would do in that situation is irrelevant. You aren't a MLB manager and thus any data you provide doesn't apply when determining the range of possible decisions made by a simulated MLB manager.

And then there is that:

Originally Posted by Dyzalot

However, I am trying to play a simulation of modrrn MLB baseball. In order for that simulation to be accurate, the sim should model the behavior of MLB managers fairly well. My whole argument is that such a model is lacking. Your input or anyone else's input as to how they would manage is irrelevant to that problem unless the poster was or is a MLB manager. However, your input into providing data or opinion on how MLB managers actually manage is very welcome.

First, I agree that the AI could be better. All the AI in all the games can always be better. We will say that all the way up to the point where the AI is going to be better than us, just before runaway AI will kill us all. But that is not the point now...

Second. Nobody here can for sure tell you how an MLB manager would play this. I watched enough of Terry Collins managing the Mets to keenly guess that at some point he'd get the guy involved that already had 78 appearances on the year, because of course he would. Or maybe he would not. What the heck do I know? I am not in Terry Collins' head.

An MLB manager will probably handle this one by ear when it comes to the long man, and not all of them would make the same move, either. All I can tell you is how I - in my (surely not spotless) capacity as manager in the game - would handle the situation. That is all you can get on here, and then we can form what we consider to be consensus. That is all the boards can be good for.

...and while I will readily agree that it is totally not the same, because none of us has to answer a hungry corps of beat writers after a mismanaged game, as well as our own angry players, conventional wisdom such as 'three strikes, you're out' and so on is very much valid. Some of us have managed more games in OOTP than any actual MLB manager can handle actual MLB games in their lifetime. I have outlasted Connie Mack. I have seen ****. But if none of that matters when it comes to handling a blowout, because I or anybody else here is not an active or former MLB manager, then there is really nothing anybody here can tell you.

[Dyzalot]

You are taking me out of context. This comment:

One last point. What you would do in that situation is irrelevant. You aren't a MLB manager and thus any data you provide doesn't apply when determining the range of possible decisions made by a simulated MLB manager.

Was in response to him saying what he would do if he was the manager. It had nothing to do with any opinion he had on what a typical MLB manager might do. I don't even remember him making a point anywhere close to that.

Your input or anyone else's input as to how they would manage is irrelevant to that problem unless the poster was or is a MLB manager.

Yes, exactly the point! How he would manage the game is irrelevant data if one's goal is to most closely model the decision making of a modern MLB manager. Once again, I said nothing close to what you claimed and was only taking to task someone who was trying to claim that the way he might manage a game was in any way relevant to the point I was trying to illustrate in my OP.

Second. Nobody here can for sure tell you how an MLB manager would play this.

We have tons of data. I don't know what else you need since that is all this game is based upon, data. That statement you make there is the same as saying that nobody here can tell you for sure how a MLB player might hit a 92 MPH slider. That doesn't keep us from using the available data in order to model it as closely as possible in order to provide a realistic as possible experience when playing OOTP.

But if none of that matters when it comes to handling a blowout, because I or anybody else here is not an active or former MLB manager, then there is really nothing anybody here can tell you.

It doesn't matter if the goal is to model MLB managers as closely as possible. If you want to model players who play OOTP then your input into how you might play the situation becomes relevant. I'm guessing the goal of the development team is closer to the former than the latter.

And I don't have anywhere near your experience in managing games in OOTP because I think it is a huge unfair advantage to have over the AI seeing as how poorly the AI manages on a day to day basis. My ultimate goal in pointing out the flaws as I have been doing over the past ten years or so is with the hope that someday I will be able to manage individual games against the AI and have a worthy opponent.

[Westheim]

I don't know what else to tell you. You want specific opinion, but you seem to be asking the wrong people still after all this time. And since by now I have been called out for being irrelevant, obtuse, and taking unfair advantage of the AI, I will give up on this quixotic fight before you also call me out for my haircut.

[Dyzalot]

Originally Posted by Westheim

I don't know what else to tell you. You want specific opinion, but you seem to be asking the wrong people still after all this time. And since by now I have been called out for being irrelevant, obtuse, and taking unfair advantage of the AI, I will give up on this quixotic fight before you also call me out for my haircut.

Stop being so narcissistic and making this about you. How about maybe an apology for lying about what I said? And the only "specific opinions" I want are those that pertain to how MLB managers manage. Thayt would be like you complaining that Markus only wants "specific opinions" for his game because he bases the probabilities on how MLB players actually play. He doesn't consider opinions from people who would play differently. Why would he ignore some rando claiming he could steal home on CC Sebathia? Because that is irrelevant to the question of whether a MLB player could steal that base. Same thing applies in this thread with anyone trying to claim that how they manage is somehow relevant to producing a model of how MLB managers manage. I am open to anyone that wants to provide actual data to support a hypothesis that they may have about the habits of the typical MLB manager. I have yet to see such a hypothesis supported by data. In fact, I have yet to see anyone even try to attempt to answer the question I had in my OP. When was the last time a manager allowed his starting pitcher to give up 9 runs in a playoff game? Since you seem to have so many answers for me, how about you try to answer that one.