How do you maximize OOTP's AI

[MKG1734]

Let me simplify this thread so that the substance of its posts and the player opinions can be stated/:

Player: Fyrestorm3
Suggestion(s): Reduce expanded (September) roster from 40 to 30
Reasoning: The AI does not handle the expanded roster configurations or in-game strategies well. For example, a team in the pennant race will start substandard players / use substandard players in-game that can affect/skew what the true outcome of a game should be.

Player: Sweed
Suggestion(s): Trade setting: Low frequency/favor prospects. MiLB use ghost players.
Reasoning: AI holds onto young talent and the AI is not forced into making sub-optimal roster moves in the minor leagues to deal with injuries

Player: The Wolf
Suggestion(s): Stats only
Reasoning: If a player is performing, he is valued and kept. If he isn't he gets benched/sent down/traded/cut

Player: OldFatGuy
Suggestion(s): Turn off 'everything' (coaching, scouting, player morale, player personality, options, Rule V Draft. Turn rating scale to 1-5 to mimic fog of scouting.
Reasoning: The AI handles each of these worse than a human player, therefore, the AI is handicapped. (AI won’t hire the best scouts, leaving some of the best available… same with coaches). Player morale/personality the AI makes poor decisions, i.e. trading players based solely on player morale/personality.

Player: Lukasberger: “Of course the AI doesn’t handle things as well as a high level player” …
Suggestion(s): None
Reasoning: None

Lukasberger, you have essentially mentioned this twice now. What option doesn't the AI handle as well as a human player? In your answer, consider that I am a human player that is not intent on 'gaming' the system but is playing an honest-effort game.

Furthermore, to expand on The Wolf's most recent comment:
I understand how to make the game 'harder' for myself. What this thread is attempting to ensure, though, is that the game AI is as "decision making error-free" as possible. I may not make the correct draft pick each and every time ... or sign an international leaguer when I should have. The development engine allows enough randomness that I will miss here and there. What I won't ever do, though, is make an incorrect 40-man roster move, incorrectly use my player options, hire a substandard scout/coach/trainer when a better one is available, etc. These are the things I want the AI to avoid, always. OldFatGuy states that the AI will make poor decisions with these options. Sweed doesn't seem to think so. Lukasberger, are these the options you mean that the AI won't handle as well as a human player? The Wolf, I understand your point about stats only, and agree that it can have a massive effect on both the AI and human player and will be using it to see if the AI uses it correctly. This is not about me, the human player. I'm not going to make mistakes with regard to the intricacies of roster management. Its impossible to always accurately gauge the development engine and that is a good thing. This is about the AI playing against the AI and ensuring that the AI games and rosters are managed correctly.

[Lukas Berger]

Originally Posted by MKG1734

Lukasberger, you have essentially mentioned this twice now. What option doesn't the AI handle as well as a human player? In your answer, consider that I am a human player that is not intent on 'gaming' the system but is playing an honest-effort game.

I must be explaining myself poorly, if so I apologize.

For "a human player that is not intent on 'gaming' the system but is playing an honest-effort game." the default ai settings should work just fine. They're tested, and should function very well.

To add a bit more of a challenge, if needed, go to stats only, at least in so far as turning off visible ratings. (I do disagree with the suggestion to not allow the ai to see the ratings. That will actually hurt the ai, not help it).

Also, set trading on hard or very hard.

Beyond that, don't set roster limits on the low minors or play with severely limited reserve rosters, as those are two things the ai struggles with.

[Grafton19]

Anyone here not set minor league roster limits or age limits?

[Lukas Berger]

As for some of the other suggestions, OFG's suggestions will actually hurt the ai, or at least won't provide enough benefit to turning it off. Having the rule 5 enabled will generally mean the player loses guys he's been able to stockpile (if any) to the ai, it actually improves ai teams rosters.

The ai won't make the decision to trade players solely on morale/personality. That's actually something a human is more likely to do

The intentional opaqueness of the personality system actually means that human managers won't have a full understanding of it and how it affects gameplay, while the ai probably does. So leaving it enabled probably actually provides the ai with a slight advantage.

Hiring the absolute best scouts and coaches doesn't provide nearly the value that people think it does, so the ai not grabbing them all up really doesn't matter that much, especially since the new coaching model doesn't clearly differentiate "best" coaches any longer.

[Leo_The_Lip]

To address the original question, I think that choosing to have a DH helps the AI because it still makes poor choices for defensive subs and lineup changes. Using the DH limits the number of times the AI has to make a choice.

Historical gamers won't like this, however it does seem to me that this helps out the AI at least as much as Stats Only.

[SirMichaelJordan]

Originally Posted by Grafton19

Anyone here not set minor league roster limits or age limits?

I set limits before (age) but the AI already does a good job of playing older guys at higher levels so its not really a good idea to limit them. It could've came in handy if I were seeing tons of 30 year olds in rookie ball but that just doesn't happen in my saves.

[Lukas Berger]

Originally Posted by OldFatGuy

Which of those does the AI handle better than or even as well than as the human player??? And if the answer is none, then how is having them on NOT an advantage? You can argue it may not be a huge advantage, and I'm not making that argument either way. The OP asked HOW TO MAXIMIZE THE AI. The way to maximize the AI is to REMOVE AS MANY ADVANTAGES as you can.

I address some of these on the next page.

But no one's criticizing you or attacking you here that I can see.

There's no problem with being critical of the game.

Your opinions are certainly valued, but sometimes some of us will disagree on things.

There's absolutely no need to take civil disagreement as a personal attack. Your views are just as valuable as mine, or anyone's. Please don't take anything being said as somehow personal, or feel you have to leave just because someone disagrees with you.

[SirMichaelJordan]

Originally Posted by OldFatGuy

Wasn't the question though how to MAXIMIZE the ai?

Coaching, I have never, NOT ONCE, played a game with coaching on where I eventually didn't end up with all of the best coaches at all levels. Another AI team here or there would also have a good staff, but mine would be the best in the league. If there's an advantage to having the best coaching staff, then doesn't having it on give the human player an advantage?

Scouting. Same thing. NEVER have I played with scouting on and not ended up with the best scout in the game, by ratings, across the board. It may take 5 seasons, it may take 10, but eventually, I will end up with the best scouts and coaches available. Does every AI team? No. Is this an advantage or not?

Sure, I could set up house rules, but I was answering the question at hand.

Player morale and player personality are jokes as far as trade decisions the AI makes for unhappy players demanding trades. Do I make trades that are less than optimal for my team just because a player is unhappy??? No. Is that an advantage or not?

Jeebus, I knew when I typed the response it was going to go over like a fart in a church because I dared to criticize the almighty OOTP in some way and didn't "qualify" my remarks by saying "OOTP's AI is the best available." It just gets so old and tiring.

Not one poster on here has named a single option that I listed where the AI handles it better than a human player. Can you? Can anyone?

Come on, here they are again.
Coaching
Scouting
Player Morale
Player Personality
Option years
Rule V Draft
Even financials (again I don't even turn this one off but if anyone thinks the AI handles financials better than the human player and doesn't lose quality players it wouldn't have to if it had handled them better than I'll eat my hat)

Which of those does the AI handle better than or even as well than as the human player??? And if the answer is none, then how is having them on NOT an advantage? You can argue it may not be a huge advantage, and I'm not making that argument either way. The OP asked HOW TO MAXIMIZE THE AI. The way to maximize the AI is to REMOVE AS MANY ADVANTAGES as you can.

I'll do everyone here the favor of making this my last post here because this crap just gets old. OOTP is the best baseball game on the market. I've said that at every site I've ever commented at. When I heard about this version adding the MLB license, I wanted to support that decision so badly I wanted to buy two copies and gift one here at the forums because of how many times I've received help here. Only to be told by Markus that he didn't approve of that idea as he wanted people purchasing the game instead of waiting to win it free. And then, within two weeks of taking down my post, what do I see???? At least TWO other giveaways, right here on the forums, of OOTP 16.

So I get it, I get the message loud and clear. And hopefully everyone else does as well so they can be saved themselves the frustration. Don't be critical of the game AT ALL and all is well. Otherwise, better carry a crap umbrella cause the turds will begin raining down on you.

Good bye

How do you know how good these features work if you never even used them? Getting the best scouting and coaching is in the past as that have been revamped for 16...

Players options for AI are fine in my game just as the Rule 5 draft (even though I have to use specific settings to get it active every year) players left off of the 40 were guys that were blocked by other guys whether it was a top prospect or a old vet with a big contract

[The Wolf]

Lukasberger: Having the AI see rating actually hurts it. If it can do that, it then keeps players with good ratings who never perform and gets rid of good performers with suboptimal ratings. The object here is to get the AI to be more competitive, and you can do that by having it keep the players who help it to win the most (those who perform better) rather than those who have better ratings but don't perform and who don't help it win.

Stats win games. Ratings don't.

[The Wolf]

Originally Posted by Leo_The_Lip

To address the original question, I think that choosing to have a DH helps the AI because it still makes poor choices for defensive subs and lineup changes. Using the DH limits the number of times the AI has to make a choice.

Historical gamers won't like this, however it does seem to me that this helps out the AI at least as much as Stats Only.

Tried it. You last statement is waaaaaaay off the mark.

I do recommend using the DH in fictional leagues. But historical? No way.

[Fyrestorm3]

You know what we can take away from the debates going on in this thread? Everyone has slightly different methods to improve the performance of the AI in their eyes. For Wolf, it's stats-only. OFG apparently turns off everything. I make minor roster rule changes, Leo turns the DH on, Lukas leaves things at factory default, etc.

I think the moral of the story here is that people's idea of a "better" AI can and does vary. The best thing to do is try out all the different suggestions and make your own conclusion as to what works for you.

[Lukas Berger]

Originally Posted by The Wolf

Lukasberger: Having the AI see rating actually hurts it. If it can do that, it then keeps players with good ratings who never perform and gets rid of good performers with suboptimal ratings. The object here is to get the AI to be more competitive, and you can do that by having it keep the players who help it to win the most (those who perform better) rather than those who have better ratings but don't perform and who don't help it win.

Stats win games. Ratings don't.

I think you're basing that statement on how you'd personally prefer things to be, rather than on how the game actually works.

The potential ratings in particular are crucial to the development of prospects in the future. Not being able to see them is a handicap. That's useful as a human wanting to level the playing field, but not useful if you want to improve the ai.

If you tweak the balance so the ai takes into account stats but doesn't ignore ratings, you actually get the optimal performance, ie not hanging onto old guys too long, but also being able to use the ratings to evaluate potential future performance for young players.

Allowing the ai to see ratings also allows it to more quickly see when an older player falls off a cliff, if you will, and the ai will react to that and ditch the player much more quickly with ratings taken into account than otherwise. That creates somewhat unrealistic gameplay in a way since the ai won't hang on to older more declining players as much as rl teams do, but it's unrealistic in the sense of being more efficient than reality, not less.

From a design perspective the ratings are what drives the game (wins games if you will). No matter how much you want to play stats only, the ratings are still used by the game in many ways, to assign correct statistical distributions, to influence the development of young players, and are still the foundation of gameplay. Not allowing the ai to take them into account will only handicap it, at least to some extent.

[Lukas Berger]

Originally Posted by Fyrestorm3

You know what we can take away from the debates going on in this thread? Everyone has slightly different methods to improve the performance of the AI in their eyes. For Wolf, it's stats-only. OFG apparently turns off everything. I make minor roster rule changes, Leo turns the DH on, Lukas leaves things at factory default, etc.

I think the moral of the story here is that people's idea of a "better" AI can and does vary. The best thing to do is try out all the different suggestions and make your own conclusion as to what works for you.

Absolutely!

[MKG1734]

lukasberger: These are excellent answers with regard to advising what the AI does and does not do/need and, I guarantee you, provide a ton of advice/feedback for both new players, moderate-term players (me), and veteran players.

[MKG1734]

lukasberger:

you say:

If you tweak the balance so the ai takes into account stats but doesn't ignore ratings, you actually get the optimal performance

If there is an exact setting, what is it? If not....what would you advise, based on the game's engine, is the 'range' that is 'most optimal'? (I assume you are talking about the base overall rating on rating/current year stats/previous year stats/two years ago stats)

[sandman2575]

Originally Posted by SirMichaelJordan

I set limits before (age) but the AI already does a good job of playing older guys at higher levels so its not really a good idea to limit them. It could've came in handy if I were seeing tons of 30 year olds in rookie ball but that just doesn't happen in my saves.

Interesting. I've set some age limits for certain minor leagues -- 23 y.o. max for R, 25 max for S-A, 26 max for A, no limits for AA or AAA. I also set roster limits -- AAA down to A, 25 players (13 batters, 12 pitchers). S-A, 30 players (15 b, 15 p). R, 40 players (20 b, 20 p).

I personally dislike having Rookie ball rosters set to 'unlimited' just because (for me) it starts to become an unmanageable mess down there. And to be honest, with R rosters capped at 40, I really haven't seen any egregious examples of AI releasing top prospects.

I'm of the mind that the way to make OOTP challenging is not to ask for miracles from the AI, but to make things challenging for me the player. FWIW, I play with Scouting on Low Accuracy, based on AI evaluation (not pure ratings), 20/45/20/15, Ratings on 1-10 scale, Injuries and Fatigue on High, Trading on Very Hard / Heavily Favor Prospects.

I find the game plenty challenging with these settings!

[Lukas Berger]

Originally Posted by MKG1734

lukasberger:

you say:

If there is an exact setting, what is it? If not....what would you advise, based on the game's engine, is the 'range' that is 'most optimal'? (I assume you are talking about the base overall rating on rating/current year stats/previous year stats/two years ago stats)

This is one of those things that everyone has a slightly different opinion on. Ask 10 of us and you'll get 10 different answers!

I wasn't advocating one specific setting so much as I was simply advocating not blocking the ai out from seeing ratings entirely.

I'll usually recommend staying with the defaults since working fairly closely with Markus over the years, I've seen how much effort Markus puts into testing them, and running test leagues to get the defaults looking realistic, matching up with rl data.

But some folks feel that other settings work better and feel more realistic for them, and I wouldn't criticize that.

A lot of it depends on just what your vision for the game is, and even what your personal opinion is on how heavily recent data should be valued over older data (the weight you put on current, 1 year, 2 year old stats etc).

In the end, I'd suggest listening to everyone's opinion on this and seeing what reasoning most closely matches your own feelings.

Then try those settings but be willing to tweak them and even try different settings from time to time to see how they work. Then you'll come up with the answer that's best for you.

[SirMichaelJordan]

Originally Posted by The Wolf

Lukasberger: Having the AI see rating actually hurts it. If it can do that, it then keeps players with good ratings who never perform and gets rid of good performers with suboptimal ratings. The object here is to get the AI to be more competitive, and you can do that by having it keep the players who help it to win the most (those who perform better) rather than those who have better ratings but don't perform and who don't help it win.



Stats win games. Ratings don't.

Yea, Ive had a good number of teams overachieve because their players put up decent numbers despite their actual ratings and was rewarded for their accomplishments by being valued higher than what they would have been.


Also to comment on the AI evaluation default settings, this can be confusing because the "defaults" in the MLB quickstart are totally different than the default when you start a non mlb quickstart game.

[The Wolf]

My experience with this game (since 6.5) makes me disagree completely with Lukasberger.

But everyone should make up their own mind, and find their own way. I do strongly recommend trying a LOT of different ways before you make up your mind, though.

[ScottM816]

Originally Posted by The Wolf

My experience with this game (since 6.5) makes me disagree completely with Lukasberger.

But everyone should make up their own mind, and find their own way. I do strongly recommend trying a LOT of different ways before you make up your mind, though.

What do you do about the draft if you can't see ratings though? I'd rather not have to use feeder leagues. I guess i could always just have my scout draft for me.