Questionable AI decisions

[jaysdailydose]

Originally Posted by MKG1734

And it is a big deal ...
I mean, is it just a game? Yes. But...when the game sells itself as being 'realistic' and the forum members claim it works properly ... and it doesn't ... its frustrating.

What I want is for the game to produce a similar and realistic feel to what someone would expect to happen in the game, today. Because that would then mimic a realistic opportunity to watch a win or loss as the game would be played today. This is essentially what OOTP sells itself as.

And taking the time up to read through dozens of threads / posts / and watch people critique one way of doing something vs. another (when, even changing these settings does NOTHING), only to watch the game perform 'wonky' (all the time, and not my word) is extremely frustrating.

So you want the game to make absolutely perfect decisions all of the time, because that's exactly what it sounds like? I've already told you several reasons why he could be choosing between three players all within literally 3 points of overall rating between them. Russell also has no stats from the prior season which is also going to throw in on the evaluation.

I mean, seriously, they are all rating between 45-48. If you're so pissed at how your manager is utilizing the players, then modify the lineups if you're able, and if you aren't, FIRE THE MANAGER.

One thing you need to realize is that OOTP does not recognize names. It is all numbers to the engine. Based on your terms of evaluation and what I assume you consider a quality manger -- that's the decision he made in THE OOTP WORLD.

The great thing about OOTP is that if you don't like how something is working out, you can almost always make a change.

Going off on a rant when the players are literally rated within 3 points of each other is, frankly, pretty ridiculous. This is nowhere near the horrific idiocy you are making it out to be, I've already come up with a bunch of reasons why the AI manager made that choice, you just don't want to accept it.

In that case, change it or fire the manager and move on with life. The game isn't absolute perfection, and neither is real baseball.

[ra7c7er]

Originally Posted by MKG1734

Keep in mind ... its not that I even CARE about a win or loss. I really don't even care.

What I want is for the game to produce a similar and realistic feel to what someone would expect to happen in the game, today. Because that would then mimic a realistic opportunity to watch a win or loss as the game would be played today. This is essentially what OOTP sells itself as.

And taking the time up to read through dozens of threads / posts / and watch people critique one way of doing something vs. another (when, even changing these settings does NOTHING), only to watch the game perform 'wonky' (all the time, and not my word) is extremely frustrating.

The game is NEVER going mimic real life. The best it's going to do is get within a reasonable margin.

You aren't a dev for the game and nobody posting here is either. Even when Markus and Matt post they aren't going to tell you the reasoning behind the game making the decisions it does.

Something in all the modifiers has the AI in your case setting the lineup the way it is. You seem to be forgetting that this is a simulation game and a RNG game. Everyone is going to experience different things. You have continued to point out things YOU wouldn't do and then claim the AI is wrong for doing those things. YOU aren't the AI and the AI isn't designed just for you. The AI makes it's decisions to put the best available team on the field and just because for whatever reason it isn't setting the lineup YOU want doesn't mean it's wrong.

You've pointed out some things that the devs should look at and they will. But dude you've got to stop nitpicking every little thing the game does that YOU wouldn't do and be claiming it's wrong. The game is making the decisions it's making because it's programmed to based on settings and ratings and all sorts of modifiers that are more complicated then I think you realize.

[Dyzalot]

Originally Posted by jaysdailydose

Really? That play essentially swung a World Series.

Schiraldi's ****ty pitching swung the series.

[Dyzalot]

Originally Posted by jaysdailydose

"Was there ever a situation where he should have moved?"

Yes, essentially every time Prince Fielder played and would no longer be hitting. In most of those cases, instead, you would see Don Kelly as the defensive replacement for Cabrera.

The proof that disputes almost all of this thread... ladies and gentlemen, may I present:

Jim Leyland

Attachment 399946

Go read through some of old Smokes' game logs... and watch his protege Ausmus sometime...

OOTP does just fine.

If OOTP does just fine then why do I have an example earlier in this thread of a 2nd baseman being pinch hit for and then the AI moving the third baseman to second, something that has been argued shouldn't really happen, and put a guy who plays second about as good at third where his overall rating is "1" because he has no experience?

The argument that moving in the infield doesn't really happen and thus it is fine that OOTP doesn't do it in order to produce the best defensive alignment kind of falls apart when I find an example of OOTP moving a guy in the infield and creating a worse defensive alignment.

[MKG1734]

One thing you need to realize is that OOTP does not recognize names

I don't want it to recognize names. I want it to recognize both the ratings and statistics as per the settings.. It is doing neither in this situation. Neither.

Russell also has no stats from the prior season which is also going to throw in on the evaluation

I've tried to explain this ... no matter which setting is used .. the lineup remains UNCHANGED. I mean going to 100 / 0 / 0 / 0 vs. 0 / 100 / 0 / 0 vs. 25 / 25 / 25 / 25 ... its doesn't matter. What this DOES do, after recalc is hit, is changed the headline overall / potential ratings. But the actual lineups do not change, whatsoever. This makes me wonder if the game actually takes this into account, at all, in lineup configuration. The evidence seems to point to no.

Going off on a rant when the players are literally rated within 3 points of each other is, frankly, pretty ridiculous

I'm confused bc I've been led to believe (thru all the posts claiming so) that the AI, when the AI evaluation settings are set a certain way, will defer to statistics vs. ratings. It certainly doesn't appear so.

I mean, seriously, they are all rating between 45-48

Certainly you've noticed that one of the three players with both the best (by a substantial amount when totaled) of the hitters, who also has the best statistics by an immense amount ... and that this is the player not playing?

So you want the game to make absolutely perfect decisions all of the time

Take the names away from the players as you should, and only go by ratings / stats, the game should understand what to do in this situation.

I've already come up with a bunch of reasons why the AI manager made that choice, you just don't want to accept it.

I want to accept that the game works properly under different evaluation settings ... but changing them from one end of the spectrum to another - along with everything inbetween - doesn't change anything.

That's why I'm asking ... what is the game reading here? It seems you believe the game is reading Baez's current 65 power vs. Russell's 50 power is leading to Baez's placement in the lineup ... even though every single other metric favors Russell (and by a large margin....including each statistic).

So, then, the game is also favoring Alcantara (55 current power) over Russell's 50 power ... even though every other metric favors Russell, as well?

If that is the case (a developer can probably answer this) ... wouldn't it seem that maybe, just maybe, the game is valuing that very slight extra power a little too much, in general? (50 vs 55) when the other underlying ratings suggest one player is clearly superior to the other?

Not to mention the statistics wildly favor one player. Why doesn't the stats only evaluation settings adjust this, then?

The AI makes it's decisions to put the best available team on the field and just because for whatever reason it isn't setting the lineup YOU want doesn't mean it's wrong.

I think any reasonable person can look at this and understand something is 'wonky.' And, I'll repeat: I've been led to believe (thru all the posts claiming so) that the AI, when the AI evaluation settings are set a certain way, will defer to statistics vs. ratings, etc. It certainly doesn't appear so.

[jaysdailydose]

Originally Posted by Dyzalot

If OOTP does just fine then why do I have an example earlier in this thread of a 2nd baseman being pinch hit for and then the AI moving the third baseman to second, something that has been argued shouldn't really happen, and put a guy who plays second about as good at third where his overall rating is "1" because ha has no experience?

The argument that moving in the infield doesn't really happen and thus it is fine that OOTP doesn't do it in order to produce the best defensive alignment kind of falls apart when I find an example of OOTP moving a guy in the infield and creating a worse defensive alignment.

No one said it shouldn't happen. They said there is no data to support Miguel Cabrera doing it, which there wasn't and isn't.

[Dyzalot]

Originally Posted by ra7c7er

Just because the AI is making moves you wouldn't doesn't mean the AI is "wrong"

----

Nearly every sports sim game (including console games) has a offensive AI focus and they always will. Because the AI isn't trying to stop points from being scored. It's trying to score one more point then it's opponent. The AI has to take way more into account when making moves then a human player does because the AI has to always try and remain competitive both offensive and defensively at all times.

A human player says "ok I can bring in my better defensive guy because I'm up by three" and he wins the game he says "I'm a genius my decision was right" but if it happens with the AI and then a pitcher blows up and they wind up losing the game the human player says the AI should have never made that move in the first place. The AI always is bound to take everything into consideration in player moves because the ultimate goal of the AI is to win.

The AI's number one goal should be to model real MLB baseball, not win, in my opinion. If it was always trying to win then it should be sac bunting a whole lot less. Instead it makes sense to see the sac bunting to some extent because real life managers still overuse the strategy.

[jaysdailydose]

Originally Posted by MKG1734

If that is the case (a developer can probably answer this) ... wouldn't it seem that maybe, just maybe, the game is valuing that very slight extra power a little too much, in general? (50 vs 55)?

The game, or your AI manager? Maybe Maddon is coded to want power bats up the gut...

Now that I know you're evaluating across all positions, I can say for certain that you are looking way too far into it. All three of these players are technically below average players currently on the 20-80 scale, and there are a bunch of reasons why they could be being deployed as they are.

Also, just because its the lineup right now doesn't mean your AI manager isn't about to make a change. Maybe he wanted to give Baez a real shot because he had a ton of potential, and the stats are just starting to drive his numbers down and he's about to be replaced.

You said you're in September, finish the season and see what happens.

It sounds to me like you'd be much better suited to fictional play because you cannot divorce yourself from the names...the second you hit "Sim Day" real life went out the window.

[jaysdailydose]

Originally Posted by Dyzalot

The AI's number one goal should be to model real MLB baseball, not win, in my opinion. If it was always trying to win then it should be sac bunting a whole lot less. Instead it makes sense to see the sac bunting to some extent because real life managers still overuse the strategy.

It does pretty accurately model MLB if you look at the end-of-season statistics compared to what actually happened. Simulate a season and compare the bunting stats to what actually happened.

What is happening is who you feel are the "wrong" people are doing things, or in certain situations you are thinking in absolutes.

Come on, I saw you post a screenshot last night because Cutch tried to go first to third on Puig... and acted like something was wrong with that and wasn't completely happenstance in real baseball.

Now, there have been times when both of you gents have posted legit issues, and they've been taken into consideration by the devs already.

[Dyzalot]

Originally Posted by jaysdailydose

No one said it shouldn't happen. They said there is no data to support Miguel Cabrera doing it, which there wasn't and isn't.

Originally Posted by RchW

^^This^^

Also the fact that to be realistic some questionable things must happen in-game. A perfectly tuned AI (not possible I know) would become boring over time. Human managers in real life rarely use optimal lineups or optimal substitution patterns because seniority contract value and player comfort at certain positions still dictate who plays and where on many teams.

It is an established fact that real life teams do not swap IF positions. Starting SS do not play 2B and 3B do not play 2B or vice versa in any significant way. Even backup IF tend to play one position predominantly. Last season only 5 of 30 backup SS and 2B played significant time at both positions. So in real life a starting SS will not be moved to another position and replaced with a better fielding backup even if that is theoretically optimal. It shouldn't happen in OOTP either.


See this linked post.

http://www.ootpdevelopments.com/boar...ml#post3818738

[MKG1734]

It sounds to me like you'd be much better suited to fictional play because you cannot divorce yourself from the names

Im not a 'fan' of any team, or any 'player.' You have to attempt to understand that, first, before reading my posts. I knew I should have posted Player A, Player B, Player C. I, frankly, don't look at the names and care, at all. My entire argument is based on ratings / statistics. If you read back, you'll notice that.

You said you're in September, finish the season and see what happens.

I did finish. See the post above about Baez K'ing in the wild card game.

The game, or your AI manager?

I can't answer.... that's something the developers would have to answer. Whichever it is, though, and if this is the case (that the AI / manager wants a exceptionally small increase in power [Player A 55 vs Player B 50]), then it would seem that it likely should be scaled back a bit for the game's sake.

the second you hit "Sim Day" real life went out the window.

I'm not looking for 'real life.' I'm looking for the game to mimic the settings that are set-up.

Maybe he wanted to give Baez a real shot because he had a ton of potential

Are you saying the game makes a decision like this as in it is programmed into the logic? To 'give a shot' to a player with high potential and then sit him when he struggles, as in it is almost a 'conscious' decision? That would be a remarkable programming feat.

The bottom line is:
Player A: 322 Plate App., .282 - .332 - .480
Player B: 203 Plate App., .228 - .266 - .394
Player C: 243 Plate App., .330 - .374 - .542

If the game is set up to play 'stats only' - why is Player B starting over either Player A or Player C? Its a legitimate question ... and understanding its answer will either help me understand what the game is doing or help me adjust a setting (but which setting?)

Further more (with all defensive ratings essentially the same):

Player A:
Con: 50/55
Gap: 70/70
HR: 55/55
Eye: 40/55
AvoidK: 45/45

Player B:
Con: 45/55
Gap: 50/60
HR: 65/75
Eye: 45/50
AvoidK: 30/35

Player C:
Con: 60/75
Gap: 70/75
HR: 50/65
Eye: 45/60
AvoidK: 55/60


If the game is set up to play 'ratings only' - why is Player B starting over either Player A or Player C? Its a legitimate question ... and understanding its answer will either help me understand what the game is doing or help me adjust a setting (but which setting?) And if the answer to the 'ratings only' question is that 55 power vs 50 power trumps all the other ratings, I'm asking...doesn't it seem like maybe something (AI or manager setting) should be scaled a bit?

[Dyzalot]

Originally Posted by jaysdailydose

It does pretty accurately model MLB if you look at the end-of-season statistics compared to what actually happened. Simulate a season and compare the bunting stats to what actually happened.

What is happening is who you feel are the "wrong" people are doing things, or in certain situations you are thinking in absolutes.

Come on, I saw you post a screenshot last night because Cutch tried to go first to third on Puig... and acted like something was wrong with that and wasn't completely happenstance in real baseball.

Now, there have been times when both of you gents have posted legit issues, and they've been taken into consideration by the devs already.

It gets the overall stats right but they way it gets there is not realistic.

As to the bolded, no, it is not common for an experienced, good runner to make the third out at third because he tested an excellent arm in right by going first to third twice in the span of three games. It should happen rarely. I see way too many third outs at third or runners going to third where a throw is made with two outs in this game. I posted the "Cutch" ones as more extreme examples of the problem.

[jaysdailydose]

Originally Posted by Dyzalot

It gets the overall stats right but they way it gets there is not realistic.

As to the bolded, no, it is not common for an experienced, good runner to make the third out at third because he tested an excellent arm in right by going first to third twice in the span of three games. It should happen rarely. I see way too many third outs at third or runners going to third where a throw is made with two outs in this game. I posted the "Cutch" ones as more extreme examples of the problem.

I know you've heard this before, but you're going to have to show a helluva lot more data then you saying "I see way too many."

I don't. Also, you could be set too aggressive on the bases and would see numbers above the league average.

[jaysdailydose]

Originally Posted by MKG1734

Are you saying the game makes a decision like this as in it is programmed into the logic? To 'give a shot' to a player with high potential and then sit him when he struggles, as in it is almost a 'conscious' decision? That would be a remarkable programming feat.

The bottom line is:
Player A: 322 Plate App., .282 - .332 - .480
Player B: 203 Plate App., .228 - .266 - .394
Player C: 243 Plate App., .330 - .374 - .542

If the game is set up to play 'stats only' - why is Player B starting over either Player A or Player C? Its a legitimate question ... and understanding its answer will either help me understand what the game is doing or help me adjust a setting (but which setting?)

If the game is set up to play 'ratings only' - why is Player B starting over either Player A or Player C? Its a legitimate question ... and understanding its answer will either help me understand what the game is doing or help me adjust a setting (but which setting?) And if the answer to the 'ratings only' question is that 55 power vs 50 power trumps all the other ratings, I'm asking...doesn't it seem like maybe something (AI or manager setting) should be scaled a bit?

It's curious, you've never once mentioned the HR numbers for any of these players.

Do you know how the 20-80 scale works? 65 is a plus-plus power hitter. A lot of managers don't care if a plus-plus power hitter hits .230

No one has said that trumps all of their other ratings. I said that power is one of a lot of possible reasons why your AI manager with his own preferences and ideas wants to play a completely legitimate SS over another one.

Also, one thing I find funny? You're bitching that Baez plays over these other guys, when he has 40 less appearances than Russell and over 120 less than Alcantara in your quote above. It sure doesn't appear like that has been the case at all.

[ra7c7er]

Originally Posted by Dyzalot

The AI's number one goal should be to model real MLB baseball, not win, in my opinion. If it was always trying to win then it should be sac bunting a whole lot less. Instead it makes sense to see the sac bunting to some extent because real life managers still overuse the strategy.

No it shouldn't. OOTP is a sim GAME not a simulation engine. If the point of a game wasn't to win then it wouldn't be a game. If you are looking for a good game simulation engine a quick google search will find a dozen of them. Some of them even free ware. You could always create your own game.

[Dyzalot]

Originally Posted by jaysdailydose

I know you've heard this before, but you're going to have to show a helluva lot more data then you saying "I see way too many."

I don't. Also, you could be set too aggressive on the bases and would see numbers above the league average.

You'll notice that I hadn't posted any more about that issue until just now because you brought it up, mainly due to the sheer work that would need to be done to prove this. I am not prepared to become an unpaid troubleshooter/tester for OOTP. What I've done in this thread is work enough. I play the game to relax, not for it to become another job.

[jaysdailydose]

Originally Posted by Dyzalot

You'll notice that I hadn't posted any more about that issue until just now because you brought it up, mainly due to the sheer work that would need to be done to prove this. I am not prepared to become an unpaid troubleshooter/tester for OOTP. What I've done in this thread is work enough. I play the game to relax, not for it to become another job.

All I've seen is one thing that might actually be a legitimate issue from you and a whole bunch of unfounded complaints with absolutely no data to support them at all.

It's a game, relax. When you see something really crazy, post a bug report. Most of the stuff you have posted doesn't come close to qualifying. No offense.

[Dyzalot]

Originally Posted by ra7c7er

No it shouldn't. OOTP is a sim GAME not a simulation engine. If the point of a game wasn't to win then it wouldn't be a game. If you are looking for a good game simulation engine a quick google search will find a dozen of them. Some of them even free ware. You could always create your own game.

And yet Markus uses it to make predictions about the upcoming season as well as highlighting testimonials of other people using it to make predictions about certain aspects of the future of the game. He isn't afraid to "talk up" OOTP's ability to sim real life MLB. I'm not expecting perfection. Just don't tell me "you can make it as realistic or unrealistic as you could want". No, sorry, I can't. At least not yet. Maybe someday the computer power and coding needed will be trivial. I accept that today is not that day.

[Dyzalot]

Originally Posted by jaysdailydose

All I've seen is one thing that might actually be a legitimate issue from you and a whole bunch of unfounded complaints with absolutely no data to support them at all.

It's a game, relax. When you see something really crazy, post a bug report. Most of the stuff you have posted doesn't come close to qualifying. No offense.

No data? Haven't I posted screen shots of everything?

[MKG1734]

Do you know how the 20-80 scale works?

With regard to the ratings, yes, I worked for a MLB team as a scout from 2006-2009. So... quite aware, yes.

Baez plays over these other guys, when he has 40 less appearances than Russell and over 120 less than Alcantara in your quote above. It sure doesn't appear like that has been the case at all.

I had Baez in MiLB longer than Russell and Alcantara, that is the reason for the less PA.

you've never once mentioned the HR numbers for any of these players.

Ive since finished the season ... here are the year-end stats (HR and AB/HR in parentheses)

Total:
Player A: 376 PA, .283 - .336 - .477 (15, 23/1), 2.2 WAR
Player B: 313 PA, .257 - .294 - .429 (12, 25/1), 1.5 WAR
Player C: 326 PA, .318 - .374 - .505 (7, 43/1), 3.3 WAR

vs. RHP:
Player A: 322 PA, .282 - .332 - .480 (13, 23/1)
Player B: 203 PA, .228 - .266 - .394 (8, 24/1)
Player C: 243 PA, .330 - .374 - .542 (7, 32/1)

Again ...

With "stats only" .... player B should not see the field vs RHP. So...why is he?

Does the AI Evaluation settings only adjust the headline overall / potential ratings and not matter to lineup settings?