Making Life As Easy As Possible For AI

DaBears · 01-06-2019, 06:33 PM

Go Bears!!!

Interested in thoughts on how to make life as easy as you absolutely possible for the AI. There seems to be a lot on here about how the AI is this or that, areas it struggles in. Thought it might be helpful to try and wrap some of those ideas into one post.

For me, I have noted:
*Do not enable the 40-man roster, thus avoiding waivers (which the AI seems particularly awful at) - this does also disable expanded rosters, so there can be a price to pay

*Change Super 2 deadline to "No Super 2 Players Allowed" - the AI does not seem to have a clue on promoting players as far as timing, so this makes it simple.

Brad K · 01-06-2019, 07:30 PM

Something that surprised me is the lack of a "help AI" switch which has been present in other games I've played.

There's a comment in another thread about making AI as good as a human, but this is clearly impossible. There are always computer limitations that mean AI will always be some amount less good than the designers. And clearly it is impossible for designers to be the best ever in every area.

Despite complaints of AI cheating people don't complain about the trading difficulty switch so it seems the presentation is the key.

Since GMs are looking at the stats produced by players help that increases their performance is not appropriate since stats have to be consistent with reality. And you can't have every players performance drop by going from an AI team to a human one.

So it seems rather than increasing player performance the help should be in the form of helping AI select better players. This should be pretty easy with scouting on: just give AI higher scouting accuracy. Don't have details on how it could be done with real ratings on.

t-bone shuffle · 01-07-2019, 03:34 PM

Two approaches here:

For helping the AI, I've found that reducing the number of minor league affiliates to 4 (AAA, AA, A, Short A), 5 max really helps. I usually increase roster size at the next to lowest level (27 or so), and make certain that the lowest level has no roster size limit.

I also adjust the team market size/loyalty, etc to make certain that there is a proper disparity between the haves/not's, so that you'll see the have's in particular spend plenty.

These are just a couple of examples. There are certainly more/better that others might detail for you here.

2nd approach, assuming you are running an organization is to make it more difficult for yourself. I use a 1-10 rating scale for example. It makes assessing talent much more difficult. Use the hard/realistic injury setting, etc.

I also try to play what I'd term as "realistic". Meaning, for example, I'm very cautious with pitching prospects and injured pitchers (limiting pitch count/not rushing return from rehab/bringing prospect's along slowly). There are many other "realistic" options (trades - if it feels like you're taking advantage of the AI, don't do the deal), and in general not doing anything that doesn't "feel" right when dealing with the AI.

Just a couple of approaches for you to consider that might swing the advantage to the AI's side.

Brad K · 01-07-2019, 06:14 PM

Maybe a good way to help AI is to voluntarily restrict spending to well below budget. I pay no attention to budget, instead trying to have a payroll that is maybe 8 to 10 in a 20 or 24 team league. It will creep up as players become arbitration-eligible etc but when I get up to 5 or 6 in payroll I get rid of the higher-paid players so I can get back to 8 to 10. Sometimes lower.

DaBears · 01-07-2019, 10:51 PM

Quote:

Originally Posted by t-bone shuffle

I also adjust the team market size/loyalty, etc to make certain that there is a proper disparity between the haves/not's, so that you'll see the have's in particular spend plenty.

Curious, what level of variance do you use? I have teams like KC/TB/OAK at 3 for market size, SDP/MIN/CLE at 4 - on the other end, NYY at 12, BOS/LAD at 10, SFG/STL/PHL/CHC at 9.

t-bone shuffle · 01-08-2019, 11:21 AM

Quote:

Originally Posted by DaBears

Curious, what level of variance do you use? I have teams like KC/TB/OAK at 3 for market size, SDP/MIN/CLE at 4 - on the other end, NYY at 12, BOS/LAD at 10, SFG/STL/PHL/CHC at 9.

I will have the NYY at 16 - 18 for example. Some of it depends upon league size too. I currently have a 40 team MLB, so in my mind, market size range is even greater.

To answer you more specifically, I'll always have NYY at 15+. LAD, BOS at 12+. CHC, SF, NYM, PHL 10-ish (teams with large market, but don't always spend like it).

Next level is WAS, HOU, ATL, STL, TEX, CWS (market size + fan interest/loyalty). In this example a team like STL is certainly a smaller media market, but their fanbase strength is an equalizer in my mind.

Bottom feeder markets are minimal in media, fanbase, and stadium situation (OAK, TB). Oakland is not a small media market at all, but their building effects their fanbase, etc. In my mind, KC is a better/stronger market than TB or OAK, because they keep their ballpark updated and have a much more financially influencing fanbase.

So, I try to factor in as much as possible, and include my eyeball tests to establish the size number.

highandoutside · 01-08-2019, 04:14 PM

Things I do to make the AI a more competitive opponent:
- Strict salary cap for me, and great disparity between big and small markets As someone mentioned, I ignore my budget and stick to a cap well below what most other franchises are spending.
- I play stats-only, with the exception of defensive and speed ratings.
- I play with 25 man roster; computer gets 28. Without this the CPU will constantly have a tired bullpen.
- I place no limits on AI's minor league roster size. Mine are at 25
- This is subjective, but I try to make fair trades.

01-06-2019, 06:33 PM	#1
DaBears Major Leagues Join Date: Jun 2013 Posts: 468	Making Life As Easy As Possible For AI Go Bears!!! Interested in thoughts on how to make life as easy as you absolutely possible for the AI. There seems to be a lot on here about how the AI is this or that, areas it struggles in. Thought it might be helpful to try and wrap some of those ideas into one post. For me, I have noted: Do not enable the 40-man roster, thus avoiding waivers (which the AI seems particularly awful at) - this does also disable expanded rosters, so there can be a price to pay Change Super 2 deadline to "No Super 2 Players Allowed" - the AI does not seem to have a clue on promoting players as far as timing, so this makes it simple.

01-06-2019, 07:30 PM	#2
Brad K Hall Of Famer Join Date: May 2016 Location: St Petersburg Florida USA Posts: 5,420	Something that surprised me is the lack of a "help AI" switch which has been present in other games I've played. There's a comment in another thread about making AI as good as a human, but this is clearly impossible. There are always computer limitations that mean AI will always be some amount less good than the designers. And clearly it is impossible for designers to be the best ever in every area. Despite complaints of AI cheating people don't complain about the trading difficulty switch so it seems the presentation is the key. Since GMs are looking at the stats produced by players help that increases their performance is not appropriate since stats have to be consistent with reality. And you can't have every players performance drop by going from an AI team to a human one. So it seems rather than increasing player performance the help should be in the form of helping AI select better players. This should be pretty easy with scouting on: just give AI higher scouting accuracy. Don't have details on how it could be done with real ratings on.

01-07-2019, 03:34 PM	#3
t-bone shuffle All Star Reserve Join Date: Jan 2004 Location: The dull edge of the blade Posts: 867	Two approaches here: For helping the AI, I've found that reducing the number of minor league affiliates to 4 (AAA, AA, A, Short A), 5 max really helps. I usually increase roster size at the next to lowest level (27 or so), and make certain that the lowest level has no roster size limit. I also adjust the team market size/loyalty, etc to make certain that there is a proper disparity between the haves/not's, so that you'll see the have's in particular spend plenty. These are just a couple of examples. There are certainly more/better that others might detail for you here. 2nd approach, assuming you are running an organization is to make it more difficult for yourself. I use a 1-10 rating scale for example. It makes assessing talent much more difficult. Use the hard/realistic injury setting, etc. I also try to play what I'd term as "realistic". Meaning, for example, I'm very cautious with pitching prospects and injured pitchers (limiting pitch count/not rushing return from rehab/bringing prospect's along slowly). There are many other "realistic" options (trades - if it feels like you're taking advantage of the AI, don't do the deal), and in general not doing anything that doesn't "feel" right when dealing with the AI. Just a couple of approaches for you to consider that might swing the advantage to the AI's side.

01-07-2019, 06:14 PM	#4
Brad K Hall Of Famer Join Date: May 2016 Location: St Petersburg Florida USA Posts: 5,420	Maybe a good way to help AI is to voluntarily restrict spending to well below budget. I pay no attention to budget, instead trying to have a payroll that is maybe 8 to 10 in a 20 or 24 team league. It will creep up as players become arbitration-eligible etc but when I get up to 5 or 6 in payroll I get rid of the higher-paid players so I can get back to 8 to 10. Sometimes lower.

01-08-2019, 04:14 PM	#7
highandoutside All Star Reserve Join Date: Mar 2004 Posts: 631	Things I do to make the AI a more competitive opponent: - Strict salary cap for me, and great disparity between big and small markets As someone mentioned, I ignore my budget and stick to a cap well below what most other franchises are spending. - I play stats-only, with the exception of defensive and speed ratings. - I play with 25 man roster; computer gets 28. Without this the CPU will constantly have a tired bullpen. - I place no limits on AI's minor league roster size. Mine are at 25 - This is subjective, but I try to make fair trades.