The AI makes strange decisions in OOTP 18, hopefully fixed for OOTP19?

[joejccva71]

So I'm doing a Dodgers save and playing only as GM so I can let Dave Roberts do the lineups,rotations and strategies. I like doing this so I can evaluate personnel, not just players.

Now I'm pretty sure Dave Roberts is a decent manager. He's probably not top 5 even though it's close, but he's definitely top 10.

With that being said, I don't know if it's his decision making or if it's the AI of the game that is doing this.

Let me give you an example.

Current lineup roster vs RHP's: - (22 games played so far)

1. Kevin Kiermaier (CF): .330 avg - .360 obp - 7 hr - 19 rbi
2. Corey Seager (SS): .281 avg - .352 obp - 1 hr - 14 rbi
3. Justin Turner (3B): .277 avg - .346 obp - 5 hr - 18 rbi
4. Yasiel Puig (RF): .255 avg - .320 obp - 5 hr - 17 rbi
5. Yasmani Grandal (C): .308 avg - .416 obp - 4 hr - 11 rbi
6. Adrian Gonzalez (1B): .303 avg - .379 obp - 3 hr - 20 rbi
7. Logan Forsythe (2B): .307 avg - .365 obp - 3 hr - 10 rbi
8. O'Koyea Dickson (LF): .244 avg - .326 obp - 0 hr - 0 rbi
9. Pitcher

This is the standard lineup that the manager has set as a "default", with their current stats. But the manager will also make changes before each game.

Keep in mind that Franklin Gutierrez (LF) is on an absolute tear right now batting .385 and he's leading the team in HR's with 11. But he's not in the lineup, he's 100% rested and ready to go.

But the manager wants to play Dickson instead. Dickson does have better batting ratings than Gutierrez but Dickson's avg is .244 and hasn't been playing very well. I'm about to play the Phillies against Buchholz (RHP) who aren't playing very well.

Here is the lineup that the AI manager (Dave Roberts) set before the game:

1. Kevin Kiermaier (CF): .330 avg - .360 obp - 7 hr - 19 rbi
2. Corey Seager (SS): .281 avg - .352 obp - 1 hr - 14 rbi
3. Justin Turner (3B): .277 avg - .346 obp - 5 hr - 18 rbi - (in a slump)
4. Yasiel Puig (RF): .255 avg - .320 obp - 5 hr - 17 rbi
5. Adrian Gonzalez (1B): .303 avg - .379 obp - 3 hr - 20 rbi
6. Logan Forsythe (2B): .307 avg - .365 obp - 3 hr - 10 rbi
7. O'Koyea Dickson (LF): .244 avg - .326 obp - 0 hr - 0 rbi
8. Austin Barnes (C): .273 avg - .385 obp - 0 hr - 3 rbi
9. Pitcher

So here are my problems. This doesn't make a lick of sense.

1. If Justin Turner is "in a slump" according to the game, then WHY is he batting 3rd????

2. Yasiel Puig hasn't had a hit in 4 straight games putting up a straight 4 goose eggs, but he's batting cleanup? Um, why? His avg is .255.

3. Yasmani Grandal just came off a 2 HR game, but now he's on the bench and Barnes is playing this game? Keep in mind i have all depth chart set to "play sub only if tired". Another head scratcher.

4. Franklin Guiterrez is considered "HOT" right now but he's also on the bench and Dickson is playing instead who has a bad avg and is not playing well. Dickson has better batter ratings, but Guiterrez is HOT atm.

The AI seems to make some serious boneheaded decisions and this is one of the very things that I hope drastically improves with OOTP 19. Of course if you guys think I'm wrong with my assertations of this lineup and why the AI would do this, please feel free to critique.

[Lukas Berger]

First, I think you're expecting the ai to do something even real managers don't do, which is chop and change the lineup in a wholesale way based on very small sample size performances.

I've never heard of a manger moving his cleanup hitter down the lineup after 22 games while he's still 3rd on the team in HR's because he's hitting .255. Or move a player out of the 3 spot because he's "in a slump" even though he's hitting a very solidly healthy .277/.346 through only 22 games.

Maybe it happens in rare cases, but most managers tend to value stability a lot more than that. They don't change things around quickly based on small sample sizes. Doing so can have a pretty adverse effect on player morale in real life.

As far as playing a player because he's just had a 2-HR game, I'm not really sure how much that happens in reality. It's certainly not something that I think you could count on a real manager doing. Some might be more inclined to do it than others.

As far as Gutierrez, since he has more than twice as many HR's than anyone else on the team, I'd say the ai is indeed playing him appropriately. For any one game, players are going to be rested, bench players are going to play at times to keep them fresh. You really can't get worked up over whether a player sits or starts in any one or two games, as long as his overall playing time is reasonable. Neither real-life managers nor the ai are going to go to great lengths to fully maximize their lineup efficiency for every game of a 162 game season. They generally like to keep things stable, but they will play backups on occasion to get them some playing time. If the ai was always min-maxing it would actually be a big problem.

That being said, if you want the ai to take stats and current performance into account more, it can and will do that, but you need to tell it to do so.

What are your ai evaluation settings?

Go to 'league settings' > 'ai settings. Then take a look at 'player evaluation ai settings'. The default is heavily weighted toward ratings, and only takes current stats into account for 20%.

For what you're wanting to do, I'd guess you're probably going to want to put a heavy, heavy emphasis on current stats. Probably switch from the default of 65% ratings weight to 65% current stats weight and adjust the other settings accordingly.

You could even go 100% current stats as that sounds like what you really want to do, but I would not recommend that. It's there if you want to try it though.

Traditional and sabermetric lineups will also make a difference. Sabermetric is probably more likely to shift the ai in the direction you want it to go, but try both.

Also, take a look at the individual manager settings, and tweak those as well if you want the managers to behave differently than they are.

Long story short, there are lots of customization options in the game, that you can use to tailor make the experience into something that works better for you.

Just because something isn't working how you currently want doesn't mean the game is broken. There are so many different customization options that cater to so many different playing styles and expectations. Which just means you have to take the time to look around, to make some adjustments and experiment a bit to find the sweet spot that works for you.

[r0nster]

well said Lucus. If I might add a question for you on this. How much tweaking at one time or season ?? Would you prefer to make a league and use it as a tester then afterwards if its to your liking then input the settings into the game you care about ???

[Lukas Berger]

Originally Posted by r0nster

well said Lucus. If I might add a question for you on this. How much tweaking at one time or season ?? Would you prefer to make a league and use it as a tester then afterwards if its to your liking then input the settings into the game you care about ???

For me, yes that's the way to go. I'd hate to screw up my league I'm currently playing by getting weird results when testing out settings!

As far as how much tweaking, generally less is more imo. Too much can have unintended consequences that screw things up in unexpected ways, especially if tweaking multiple options at the same time.

So basically I'd recommend not doing that. Make a change to one setting, test it to see how it works, then if you're happy with that, repeat the process with the next setting. Much less potential for craziness that way.

[NoOne]

scientific method!



always use a backup for testing settings, too. i guess some are more "an-al" than others, but i can't stand seeing a major hiccup in my league history due to me shooting myself in the foot.

especially for hte op... try different STats/ratings eval settings, refresh managers to match new settings and see what changes.

mlb lineups, in real life, have all sorts of things that have nothign to do with reality or winning.. i.e. seniority is a real factor... and i don't speak of situations of small samples and bloated stats for the younger player.

so, every year in the mlb multiple managers put a player in a less optimal slot simply do to concerns about ego and divas. the non-rational reasons are not limited to seniority, i would bet the house on it.

just look at victor martinez... it took 2+ years of futility before detroit was willing to bump him down in lineup.. he had no business being there in year 2 for sure. age, surgeries, etc - context was obvious by that point in time and a fairly large sample was formed where the %error wasn't large enough for him to be anythign but ~average or less offensively. (average= not good)

v martinez batted ~5th in lineup for about a year soley due to his name and contract size... (sample was good enough by then)

[Drstrangelove]

Frank Howard 1968

May 5 - May 11 Batted .130 with 1 homer, 1 RBI. Cold.
May 12 - May 18 Batted .542 with 10 homers and 17 RBIs. Piping, steaming hot. (I saw these btw.)

He should have been benched May 12- May 18?


Willie Mays MVP season in 1965:

1) May 24 - June 8 batted .103.
2) June 9 - July 8 batted .371
3) July 10 - July 31 batted .143 with 1 homer
4) August 1 - August 29 batted .374 with 17 homers

IRL they know players slump. They let them work out of the slump and wait until they're hot.

Of course, the reply might be: Well Mays does a lot more than hit. And you don't know when he's going to be hot, so you have to play him or he might be on the bench hitting nothing in August. But that's just as valid a thought process for a human as a computer.

[Juggernt]

Plus, it's Willie Mays. You play him always.