Questionable AI decisions

[MKG1734]

Baez is the better rated player to the scouts. The engine is going to still rely a bit on ratings, as none of these players have a lot of stats to work off

Sorry, my last post went up before I saw this. This makes some sense. But, I also wonder, if this is an accurate statement. Likely something the development would only know for sure.

[ra7c7er]

you do know it's not just Baez vs. Russel right? The game creates the best overall lineup and defense it can with all the available players. Overall for whatever reason it says Baez is part of the best overall lineup.

[MKG1734]

Originally Posted by ra7c7er

you do know it's not just Baez vs. Russel right? The game creates the best overall lineup and defense it can with all the available players. Overall for whatever reason it says Baez is part of the best overall lineup.

Yes, I'm very aware of this.

But, I'm still not seeing how it could be possible. Player C (Russell) is superior to Player B (Baez) in every aspect. Not to mention Player C is ALSO superior to Player A in every aspect. Every aspect meaning overall / potential, meaning total batting and fielding and baserunning ratings. Player C is superior to both in statistics. Its just very peculiar why the game would make such a decision. And I'm trying to figure out what is behind it to see if it can be adjusted.

The

for whatever reason

you state is what I am after.

[jaysdailydose]

Originally Posted by MKG1734

Sorry, my last post went up before I saw this. This makes some sense. But, I also wonder, if this is an accurate statement. Likely something the development would only know for sure.

That point has absolutely been researched through previous versions. When there are not adequate stats for it to evaluate, it does go under the hood for info.

With 100% scouting on, though, I really have no idea. I'd probably have to see your entire settings for the league.

The ratings you showed me earlier do not indicate a far superior player, however, just for the record. They were still very similar players.

[jaysdailydose]

I'm going to take an educated guess and say that you have a setting (or a coach/manager, etc) set to "Favor Tools" over Ability, or something along those lines -- or that the team strategy or some option has them favoring power hitters.

With what you're showing, I don't want you to think I don't understand what you're saying -- I totally do, and I wouldn't still be posting if I didn't want to help.

[MKG1734]

Originally Posted by jaysdailydose

That point has absolutely been researched through previous versions. Baez is the better rated player to the scouts. The engine is going to still rely a bit on ratings, as none of these players have a lot of stats to work off

With 100% scouting on, though, I really have no idea. I'd probably have to see your entire settings for the league.

The ratings you showed me earlier do not indicate a far superior player, however, just for the record. They were still very similar players.

Well ... I believe you.

the engine is going to still rely a bit on ratings, as none of these players have a lot of stats to work off

I simply figured that 100% current-year stats would negate this point once the players had reached 300 PA each, with one clearly far superior (statistically) ... which made me think something was not working properly (and, honestly, still seems 'wonky' if it is working properly in this case)

[jaysdailydose]

Originally Posted by MKG1734

Well ... I believe you.



I simply figured that 100% current-year stats would negate this point once the players had reached 300 PA each, with one clearly far superior (statistically) ... which made me think something was not working properly (and, honestly, still seems 'wonky' if it is working properly in this case)

Do you mean that you are evaluating at 0/100/0/0 in some tests?

I'm really thinking a lot of this is because these guys all have under 600 at-bats for their career... but at the same time I understand how that would suck because you don't want to sim MLB ahead.

Almost all of us that play stats-only do so in fictional environments, to be honest.

[MKG1734]

Originally Posted by jaysdailydose

Do you mean that you are evaluating at 0/100/0/0 in some tests?

Yes .... thats what I've been saying ...

setting AI Evaluation to ANY setting does not have ANY effect on the lineup, at ALL!

Ive tested:
100 rating / 0 current year / 0 last year / 0 2 year ago
0 rating / 100 current year / 0 last year / 0 2 year ago
25 / 25 / 25 / 25

and an entire array of different settings aside from the above

NONE of them change the lineups used. The lineups remain the same under EACH setting I've tested. Which seems like the AI Evaluation setting means, essentially, nothing when it comes to the lineups used.

Is this intended? Do the evaluation settings play zero role in lineup construction by the AI? Its an honest question bc I don't know ...

[Dyzalot]

Deleted

[jaysdailydose]

Originally Posted by MKG1734

Yes .... thats what I've been saying ...

setting AI Evaluation to ANY setting does not have ANY effect on the lineup, at ALL!

Ive tested:
100 rating / 0 current year / 0 last year / 0 2 year ago
0 rating / 100 current year / 0 last year / 0 2 year ago
25 / 25 / 25 / 25

and an entire array of different settings aside from the above

NONE of them change the lineups used. The lineups remain the same under EACH setting I've tested. Which seems like the AI Evaluation setting means, essentially, nothing when it comes to the lineups used.

Is this intended? Do the evaluation settings play zero role in lineup construction by the AI? Its an honest question bc I don't know ...

They absolutely do, but I honestly can't know if you've nerfed something with the constant changes, either...

This is one of the reasons if you're going to go stats only, you do it with fictional players. Most of this is based on the preconceived notions you have of these players...and I'd wager part of it is the game having to go off of ratings because you don't have the amount of stats into the system as a true stats-only player would.

Not saying that is a bad thing, either, but it does add layers of complexity to your question.

I still think its the fact that most managers would find a way to get a potential 40 homer bat into their lineup and the more "Punch and Judy" hitter would be the backup -- but that is just me.

Now that you know that potential is factored in, and that the game does go under-the-hood when it doesn't have enough stats to make educated decisions, doesn't that make it a bit more plausible that the AI saw a player that projected to be better when he came up, and is just now taking the ratings hit (as you said, now that they've all got 300+ PA?)

By all means, keep me posted as to how your team is deployed starting the next regular season... but also get a set of AI evaluation numbers and STICK to it. I mean, you could have simply forgotten to hit a rescout somewhere along the line and nerfed something, I can't know.

[MKG1734]

I still think its the fact that most managers would find a way to get a potential 40 homer bat into their lineup and the more "Punch and Judy" hitter would be the backup

For what its worth ... I just got back to the game and checked the manager's ratings in the editor:

Hitting Coach Focus: Neutral (does not favor power)
Overall Roster Strategy: Set all the way to Favor SPEED

So... now everything that has been 'guessed' has been incorrect.

The game now has been being played at 100% current year statistics ... with a manager that favors speed ...

and yet....

vs. RHP:
Player B: 203 PA, .228 - .266 - .394 (8, 24/1)
Player C: 243 PA, .330 - .374 - .542 (7, 32/1)
..
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

Player B is the player playing in the lineup. And, lets remember, Player C also has superior defensive ratings to Player B. The players also have EQUAL speed / stealing / baserunning ratings.

Now, what is going on here? I am more convinced than ever that the game is not working properly.

The information indicates that the AI Evaluation settings actually do nothing in terms of roster management. And, more odd than I even realized, the manager of the team highly DIS-favors the player that is actually playing.

Hell, I don't know ... maybe I'm the first person who has actually paid attention to this stuff and how it all interworks (or doesn't work, in this case) together.

[Dyzalot]

Two outs with a man on third and the AI intentionally walks the lefty hitting #9 hitter to pitch to the righty leadoff hitter with a lefty pitching.

[dunningrb]

Originally Posted by Dyzalot

Two outs with a man on third and the AI intentionally walks the lefty hitting #9 hitter to pitch to the righty leadoff hitter with a lefty pitching.

Strange decision. How did it play out?

Is Andrus terrible vs. lefties?

[Dyzalot]

Originally Posted by dunningrb

Strange decision. How did it play out?

Is Andrus terrible vs. lefties?

Andrus is better against lefties than righties and vice versa for Choo. It did not work out well for the AI controlled team as it led to a couple more runs being scored. I'd have to think that it is rare for the guy in the nine hole to get intentionally walked with two outs in order to face the leadoff guy and rarer still when you have a lefty pitching and are walking the lefty to face the righty.

[Dyzalot]

Playing with default modern day reliever usage settings. Using preset team strategy of "sabermetric". This happened twice in a four game stretch.

[madJ]

Originally Posted by Dyzalot

Playing with default modern day reliever usage settings. Using preset team strategy of "sabermetric". This happened twice in a four game stretch.

What does that prove? For all we know the Padres could just have had no lefty reliever in the bullpen. In that case those pitching changes are hardly questionable AI decisions.

[Dyzalot]

Originally Posted by madJ

What does that prove? For all we know the Padres could just have had no lefty reliever in the bullpen. In that case those pitching changes are hardly questionable AI decisions.

It is the ninth inning in a save situation. The closer is in with a three run lead. You don't take him out just because he gave up a lead off solo shot. You'd end up with some pretty pissed off closers.

[madJ]

Originally Posted by Dyzalot

It is the ninth inning in a save situation. The closer is in with a three run lead. You don't take him out just because he gave up a lead off solo shot. You'd end up with some pretty pissed off closers.

If I was a sabermetrical manager, I couldn't care less for the sentiments of my "closer". If he has a bad day, I would take him out of the game before he blows a save. Sure, you could argue the game should use the closer in the same wrong way closers are used in real life. In that case those decisions are bad indeed.

[Dyzalot]

Originally Posted by madJ

If I was a sabermetrical manager, I couldn't care less for the sentiments of my "closer". If he has a bad day, I would take him out of the game before he blows a save. Sure, you could argue the game should use the closer in the same wrong way closers are used in real life. In that case those decisions are bad indeed.

I don't know of any manager in the game that would take his closer out after one batter in the ninth when it is still a save situation, sabermetrician or not.

[madJ]

Originally Posted by Dyzalot

I don't know of any manager in the game that would take his closer out after one batter in the ninth when it is still a save situation, sabermetrician or not.

The problem is that all/most managers don't want to deal with the heat they would get when they hooked their closer quickly in a tight game. It is easier to deal with a blown save and a loss than to do the right decision and deal with everybody questioning your decision and the "closer" being pissed.

The sabermetrical strategy preset hooks relievers quickly. That's what you are seeing there. I don't see any questionable AI decision. I only see questionable human perception of the closer role.