How do you maximize OOTP's AI

[endgame]

Originally Posted by jaysdailydose

I really don't get what you're trying to ask.

Obviously it makes them favor prospects more than they do on the "favor" setting... I would think this would be self-explanatory?

I think he is looking for the blurred line. What elements of the prospect make the tilt toward more or less. What specific trait, rating, etc... would make one attractive under heavy, but not under favor, for instance.

[jaysdailydose]

Originally Posted by endgame

I think he is looking for the blurred line. What elements of the prospect make the tilt toward more or less. What specific trait, rating, etc... would make one attractive under heavy, but not under favor, for instance.

If so, that is a question for the devs, because I don't really know of anyone who did any major work to find out in previous versions.

All I know is that you can see them weighting prospects more strongly on the highest setting than just favor. It definitely doesn't preclude you from trading a veteran or for a veteran either, as I know others will ask.

I don't have data except for my years of using the settings and my personal experience, though. sorry!

[RchW]

Originally Posted by endgame

I think he is looking for the blurred line. What elements of the prospect make the tilt toward more or less. What specific trait, rating, etc... would make one attractive under heavy, but not under favor, for instance.

Better prospects perhaps. I thought jdd's response was correct. It's impossible to parse the effect of these settings without the context that is only available in a particular league. At the top level if a proposed trade does not meet team needs then a negative response will have nothing to do with prospect favor settings. You could take it further and suggest that a certain mix of prospects could render both settings moot as all prospects or none may qualify at certain stages of each season and within the history of the league (ie low and high populations of prospects that meet needs)

One must respond to feedback received, not necessarily expect to know what form that feedback will be be for any particular league set up or current state..

[Okay]

Originally Posted by jaysdailydose

I've played with both, but the way it's felt in prior versions is that the 5 for ratings would allow "hype" to play into decisions a little bit. It felt like it would give thoughts of what a players perceived abilities were (ratings) while still keeping the main focus on the players production.

The best way to see the difference is to set up one test league, run four seasons, and then scout the league using each setting. Do it in commissioner mode and go look through the AI teams and see what changes they make off this change in evaluation. You'll get a clearer view of what it does than me explaining. You can do it with a very small league with no minors and cut out all the guff like baseball cards and the like so it sims ultra fast.

The biggest change in the difficulty for me has been turning coaches and scout ratings off. Then you have to actually see who scouted or coached well instead of it telling you who is a legend so you always have the best and know to 95% trust the info you receive.

My first experience with 0/67/22/11 not looking hot. It is, as I said, a historical fantasy league. It starts in 1936. After the war years, attendance increases like crazy (a thing I verified at baseball reference) and teams have *tons* of money. Anyway, I simmed at various rates of speed, through 1949.

Trading seems okay (it's set on average frequency).

But I noticed some terrible AI decisions (v16), including teams flat out dropping rookie of the years and other really good players with about two years ML stats that both played well and also have incredible potential, for example, Yogi Berra, Carl Furillo, and some others. (And again, money is not an issue for teams.)

I don't know if this is due to the AI ratings setting. But I've not seen this before in many similar leagues I've done before (all pre-v16). Maybe it's a v16 thing, but I've turned off managers and owner's expectations. The only thing I can think of is to restart at 1936 and use a more 'traditional' ratings setting and see what happens

[jaysdailydose]

Originally Posted by Okay

My first experience with 0/67/22/11 not looking hot. It is, as I said, a historical fantasy league. It starts in 1936. After the war years, attendance increases like crazy (a thing I verified at baseball reference) and teams have *tons* of money. Anyway, I simmed at various rates of speed, through 1949.

Trading seems okay (it's set on average frequency).

But I noticed some terrible AI decisions (v16), including teams flat out dropping rookie of the years and other really good players with about two years ML stats that both played well and also have incredible potential, for example, Yogi Berra, Carl Furillo, and some others. (And again, money is not an issue for teams.)

I don't know if this is due to the AI ratings setting. But I've not seen this before in many similar leagues I've done before (all pre-v16). Maybe it's a v16 thing, but I've turned off managers and owner's expectations. The only thing I can think of is to restart at 1936 and use a more 'traditional' ratings setting and see what happens

In a vacuum, some of those things sound bad. If you get the time, though, can you post the situation around Berra, for instance?

He's played well, but is there a capable replacement on the roster? I mean, if they cut him without having anyone to replace his production -- or didn't spend that money elsewhere -- I can totally understand your frustration.

You've said money and production weren't the issue, and at that point, I would look at players around him, and I would also look at attitude and morale. Obviously, we would never expect Yogi to be a total jerk, but he could be in your universe...

I would just like more information before I say "that's bad" if you don't mind? Thanks.

[RchW]

Originally Posted by jaysdailydose

In a vacuum, some of those things sound bad. If you get the time, though, can you post the situation around Berra, for instance?

He's played well, but is there a capable replacement on the roster? I mean, if they cut him without having anyone to replace his production -- or didn't spend that money elsewhere -- I can totally understand your frustration.

You've said money and production weren't the issue, and at that point, I would look at players around him, and I would also look at attitude and morale. Obviously, we would never expect Yogi to be a total jerk, but he could be in your universe...

I would just like more information before I say "that's bad" if you don't mind? Thanks.

in real life DiMaggio thought Yogi was a malingerer.

[jaysdailydose]

Originally Posted by RchW

in real life DiMaggio thought Yogi was a malingerer.

Yeah, his teammates might find him too condescending for his own good.

[MKG1734]

I have a thought, maybe some OOTP veterans or even developers could chime in.

I've been messing around with the OOTP settings a lot recently, attempting to test how the AI reacts using different settings ... in particular the AI Evaluation setting. There is a clear belief among some that 'Stats Only' is the way to go, and the way the AI plays at an optimum level. This may be true ... I haven't done enough research to answer this for myself, however.

My thought, though, is this:

What if there were essentially three different settings that you could set for the game to use.

Setting #1 for teams in "Win Now!" mode
Setting #2 for teams in "Neutral" mode
Setting #3 for teams in "Rebuilding" mode

Each setting is applied to the teams above, respectively.

The thinking being that, maybe teams that are set to "Win Now!" could use a setting (customizable by the player, as always) to evaluate players at 100% current season statistics, for example, to highlight their want for immediate help.

The Neutral setting would be a mix of stats / ratings (as always, customizable)

The "Rebuilding" setting would take more into account for ratings than in-season statistics.

Perhaps this would make the 'trade AI' better? Maybe it would also encourage "Win Now!" teams to play producing veterans over a strong-ratings younger player who isn't producing much despite the ratings potential ... while also having a rebuilding team play the strong-ratings / high-potential young player over a producing veteran.

The developers could probably answer to this more, but this seems like it maybe might help the AI produce better (more correct?) trading activity as well as help the AI correctly build their teams with regard to their "WinNow!" ... "Neutral" ... "Rebuilding" settings.

[jaysdailydose]

Originally Posted by MKG1734

I have a thought, maybe some OOTP veterans or even developers could chime in.

I've been messing around with the OOTP settings a lot recently, attempting to test how the AI reacts using different settings ... in particular the AI Evaluation setting. There is a clear belief among some that 'Stats Only' is the way to go, and the way the AI plays at an optimum level. This may be true ... I haven't done enough research to answer this for myself, however.

My thought, though, is this:

What if there were essentially three different settings that you could set for the game to use.

Setting #1 for teams in "Win Now!" mode
Setting #2 for teams in "Neutral" mode
Setting #3 for teams in "Rebuilding" mode

Each setting is applied to the teams above, respectively.

The thinking being that, maybe teams that are set to "Win Now!" could use a setting (customizable by the player, as always) to evaluate players at 100% current season statistics, for example, to highlight their want for immediate help.

The Neutral setting would be a mix of stats / ratings (as always, customizable)

The "Rebuilding" setting would take more into account for ratings than in-season statistics.

Perhaps this would make the 'trade AI' better? Maybe it would also encourage "Win Now!" teams to play producing veterans over a strong-ratings younger player ... while also having a rebuilding team play the strong-ratings young player over a producing veteran.

The developers could probably answer to this more, but this seems like it maybe might help the AI produce better (more correct?) trading activity as well as help the AI correctly build their teams with regard to their "WinNow!" ... "Neutral" ... "Rebuilding" settings.

But, it already does this... we just don't have sliders for it.

Do some tests and get some data down... I know I can feel a big difference in my games when I'm dealing with "Rebuilding" teams over "Win Now" teams.

"Stats-Only" is a playstyle. You do not have to be playing Stats-Only to take advantage of using AI Evaluation, and as most of the veterans here will tell you, whether you are playing with visible ratings OR stats-only, nearly all of us now use the AI Evaluation statistics set how we think they should be rated.

I honestly don't know any veteran posters at this point that don't use it, customized to their own playstyle. Again, you DON'T have to play stats-only to take advantage of this.

[MKG1734]

Originally Posted by jaysdailydose

But, it already does this... we just don't have sliders for it.

Do some tests and get some data down... I know I can feel a big difference in my games when I'm dealing with "Rebuilding" teams over "Win Now" teams.

"Stats-Only" is a playstyle. You do not have to be playing Stats-Only to take advantage of using AI Evaluation, and as most of the veterans here will tell you, whether you are playing with visible ratings OR stats-only, nearly all of us now use the AI Evaluation statistics set how we think they should be rated.

I honestly don't know any veteran posters at this point that don't use it, customized to their own playstyle. Again, you DON'T have to play stats-only to take advantage of this.

Forgive me if I used incorrect OOTP terminology ... when I said "stats only," I meant to imply the AI using current season stats weighted very heavily over ratings in the AI Evaluation settings.

I understand that the three current 'win-mode' settings may do this; however, it may produce better results for each individual player if you were able to customize this.

For example (just an example)...

If I wanted to, I could set the AI Evaluation for "Win Now!" teams to be 0% rating / 100% current season stats / 0% prior year stats

And If I wanted to, I could then also set the AI Evaluation for a "Rebuilding" team to 100% ratings / 0% current year stats / 0% prior year stats, etc.

...adding this as a customizable option could be very beneficial to players of OOTP

[jaysdailydose]

Originally Posted by MKG1734

Forgive me if I used incorrect OOTP terminology ... when I said "stats only," I meant to imply the AI using current season stats weighted very heavily over ratings in the AI Evaluation settings.

I understand that the three current 'win-mode' settings may do this; however, it may produce better results for each individual player if you were able to customize this.

For example (just an example)...

If I wanted to, I could set the AI Evaluation for "Win Now!" teams to be 0% rating / 100% current season stats / 0% prior year stats

And If I wanted to, I could then also set the AI Evaluation for a "Rebuilding" team to 100% ratings / 0% current year stats / 0% prior year stats, etc.

...adding this as a customizable option could be very beneficial to players of OOTP

I'm never against further customization.

I just wish people would do more testing first before we get customization added for issues like this ahead of stuff that truly needs to be patched or fixed.

You said it yourself, you haven't done enough research, yet you think we need more customization...

If people would test some settings and then come up with multiple examples of weird situations, we'd have more to break down and maybe go to the devs with if it actually WAS a problem...

Plus, you still have to have the ability for the AI to make a mistake and do something stupid otherwise it'll be OOTP Robot Baseball 2017.

Good post, though... more customization never hurts as long as it doesn't get pushed in front of other more pressing needs!

[Okay]

Originally Posted by jaysdailydose

In a vacuum, some of those things sound bad. If you get the time, though, can you post the situation around Berra, for instance?

He's played well, but is there a capable replacement on the roster? I mean, if they cut him without having anyone to replace his production -- or didn't spend that money elsewhere -- I can totally understand your frustration.

You've said money and production weren't the issue, and at that point, I would look at players around him, and I would also look at attitude and morale. Obviously, we would never expect Yogi to be a total jerk, but he could be in your universe...

I would just like more information before I say "that's bad" if you don't mind? Thanks.

Yeah I checked for any team situation that might prompt this. There was none apparent, and, like I said, there were multiple such really really bad (literally inexplicable) decisions of the AI to dump young, tested stars (it seemed to me always after their freshman or sophomore season). I can't be all that more specific, because I accidentally deleted the game folder when I went to create another version with default AI evaluation settings. But, I have no problems redoing the 1936-49 replay (real rookies enter through amateur draft) with both settings. At least I can now play closer attention to these kind of moves.

[SirMichaelJordan]

The game already does that as mention above. Rebuilding team would look to move vets for prospects so they are placing potential ratings over stats in this regard. Neutral is neutral and with win now, teams are looking for guys who are producing (high OVR) over guys with high potential ratings basically giving up their top prospects for needle movers.

[jaysdailydose]

Originally Posted by SirMichaelJordan

The game already does that as mention above. Rebuilding team would look to move vets for prospects so they are placing potential ratings over stats in this regard. Neutral is neutral and with win now, teams are looking for guys who are producing (high OVR) over guys with high potential ratings basically giving up their top prospects for needle movers.

To add on this, I think the constant changes that would happen with the evaluation settings only adds more steps in the process to give the CPU time to make an error.

[PSUColonel]

I use 40/30/20/10. I feel it strikes a good balance. Very hard trading...no preference. Use 20-80 scouting scale.

[PSUColonel]

Originally Posted by lukasberger

Yep, I think forcing the ai to go stats only, or stats heavy, actually greatly hurts the ai.

Imo the way to go for max difficulty is to let the ai have access to the ratings and turn them off for yourself.

or show only potentials

[Okay]

Originally Posted by PSUColonel

or show only potentials

I'm for showing neither current or potential ratings.

Plus, I mean if you show potentials (especially 20-80), that's giving you alotta info

[PSUColonel]

Well, that's why the scouting is imperfect. Real GMs certainly have scouting reports. And as I have said before, you still have the written reports anyway, the ratings are nothing more than a reflection of those.

[wsenkow]

First time post here.

I have agonized over whether or not to checkmark the option "Overall rating based on AI evaluation, not pure ratings", for quite some time. What I believe now is that for those that use a 0 (zero) in the first line of the AI Player Evaluation Options, (first line a Ratings Weight of the value 0 while the other 3 lines are statistical) ie 0/55/30/15 must have the option unchecked.

I used the following experiment to test this. As an MLB GM I took my starting pitchers from my active 25 roster and put them on waivers/designated for assignment. I filled those spots now with AAA starters with overall ratings of something like 22, 23, 24, etc. (I use the 20-80 scale) Although the pitchers aren't that good I did welcome the relative figures shown in the ratings.

When I 'checked' the option for Overall Rating Based On AI Evaluation, Not Pure Rating, the overall ratings for all those pitchers were now all at 20. This just seems so wrong.

So I guess my question is why would anyone who wishes to play stats-only throw a variable of 0 into a mathematical formula to determine player's rating by check marking that one option?