AI to AI trade logic is awful

PepeSylvia · 03-28-2023, 10:14 AM

Currently in December 2025 of a save and the AI to AI trade logic is insane.

Challenge mode, trading hard mode, neutral (as far as favors prospects/vets).

A few examples:

Brewers traded all of their young prospects (Turang, Mitchell, Frelick, Wiemer) for older unproductive players in 2023.

Bobby Witt Jr (78/80, still in Arb, coming off a 5.1 WAR season) was traded to Cleveland for a 17 year old currently 20/70 and a 20/20 minor leaguer.

Andrew Vaughn was coming off a 43 HR season in 2024 and was traded to Seattle for JP Crawford (who became a bench player for White Sox)

Jordan Walker traded after the 2023 season for Blake Snell. Walker was in AA with a .991 OPS and had 80 potential.

Bo Naylor traded for Starlin Castro in 2024, Naylor has solid ratings in this save.

Zac Gallen was traded for two mediocre prospects in 2024 coming off a sub 4 era year.

Anyone else experiencing this? A lot of teams are unrecognizable because teams are trading like there's no tomorrow and they have to make trades or they will implode. The only two remaining players from 2023 the Toronto Blue Jays have in their lineup are Vlad and Kirk, all of the others have been traded including Bichette, Chapman, Springer (dealt to Yankees), and Varsho.

CBLCardinals · 03-28-2023, 10:51 AM

Does real life MLB never have “awful” trades?
Where would Glasnow, Meadows, and Baz to TB for Chris Archer to PIT rank?

Also - are any of these players too expensive for a team to retain or are any in final contract season?

Probably a lot more variables occurring than just surface level.

wallewalls · 03-28-2023, 11:05 AM

Quote:

Originally Posted by CBLCardinals

Does real life MLB never have “awful” trades?
Where would Glasnow, Meadows, and Baz to TB for Chris Archer to PIT rank?

Also - are any of these players too expensive for a team to retain or are any in final contract season?

Probably a lot more variables occurring than just surface level.

I think the point is that trades like those happen maybe once a decade, and that's why they stand out. There shouldn't be multiple lopsided deals like that happening every year.

I think the AI trading frequency is way too high as well, which is what causes more of these lopsided deals to happen.

billyray1984 · 03-28-2023, 11:09 AM

I've also seen a lot of dubious trades between AI, so much so that I had to lower the trade frequency to very low because teams were basically emptying their prospect pools for bench guys or relievers. Since every AI team was doing it, it probably evens out in the end, but I don't remember 21, 22 or 23 being this crazy : it actually was the opposite where teams would hold on to dear life to any prospect they had and a vast majority of trades were AAAA guys, relievers or bench guys getting traded for one another. It's more realistic to have the AI deal prospects or big pieces since teams nowadays are more willing to go there (i.e. the Soto trade), but it's kind of extreme right now in my game.

As an aside I know that it's still early and that a lot of work was made with the trade AI (reputation/hard mode) and that maybe that made the trade logic a bit wonky. I'm not complaining.

PepeSylvia · 03-28-2023, 11:17 AM

Quote:

Originally Posted by CBLCardinals

Does real life MLB never have “awful” trades?
Where would Glasnow, Meadows, and Baz to TB for Chris Archer to PIT rank?

Also - are any of these players too expensive for a team to retain or are any in final contract season?

Probably a lot more variables occurring than just surface level.

In my examples Gallen was closest to FA with 2 years of Arb left. The sheer volume of trades is off but the trades going through are also unrealistic. It almost seems like the AI is not taking future into account at all in making trades or something.

Guthrien · 03-28-2023, 12:27 PM

No, it's bad. Far worse than 23. It took me a while to see it. The teams right now will become warped because the AI does not value valuable players correctly.

Thundercrack · 03-28-2023, 01:49 PM

Seeing the same, all settings neutral, average, AI is consistently trading stars in their prime for next to nothing. Never saw anything like it in any version before. Something is definitely off, so much so that I’ve gone back to 23 until it’s fixed

slugga27 · 03-28-2023, 01:55 PM

What are folks using for AI Eval settings?

Sent from my iPhone using Tapatalk

Gussy · 03-28-2023, 04:15 PM

The AI eval settings is the key. I've played with it some this afternoon trying to find the right setting. I did almost never and very low with eval ratings of 45% ratings weight, 30% current year, 20% last year and 5% 2 years ago and had a very reasonable amount of trades, this was with favors prospects maxed. I had zero top prospects of any sort traded at all up to the deadline. When I changed from almost never to very low, with favors prospects one step from max, and 40/35/20/5 I got similar results. Reasonable amount of trades up to deadline, zero top prospects traded. Neither of those are the right settings for realism imo, because I'd like to see some top prospect movement at deadline, so I'll keep playing with it, but I think the AI eval settings are the answer, it's just finding the right combination.

andyhdz · 03-28-2023, 04:47 PM

Quote:

Originally Posted by CBLCardinals

Does real life MLB never have “awful” trades?
Where would Glasnow, Meadows, and Baz to TB for Chris Archer to PIT rank?

Also - are any of these players too expensive for a team to retain or are any in final contract season?

Probably a lot more variables occurring than just surface level.

Exactly, when it comes to mlb truth is stranger than computer simulations.

CBLCardinals · 03-28-2023, 05:35 PM

Quote:

Originally Posted by Gussy

The AI eval settings is the key. I've played with it some this afternoon trying to find the right setting. I did almost never and very low with eval ratings of 45% ratings weight, 30% current year, 20% last year and 5% 2 years ago and had a very reasonable amount of trades, this was with favors prospects maxed. I had zero top prospects of any sort traded at all up to the deadline. When I changed from almost never to very low, with favors prospects one step from max, and 40/35/20/5 I got similar results. Reasonable amount of trades up to deadline, zero top prospects traded. Neither of those are the right settings for realism imo, because I'd like to see some top prospect movement at deadline, so I'll keep playing with it, but I think the AI eval settings are the answer, it's just finding the right combination.

Good info here - I suspect you are going down the right path on how the AI is valuing ratings vs stats with the different weights, along with other variables such as contract amount, team budget, budget space availability, etc

PepeSylvia · 03-28-2023, 06:24 PM

Quote:

Originally Posted by andyhdz

Exactly, when it comes to mlb truth is stranger than computer simulations.

There has undoubtedly been bad trades in real life, but I can't fathom a scenario where a team would trade a young player in their first year of arbitration after they just hit 43 HR for a 30 year old backup shortstop when they did not have an adequate replacement for the 43 HR player. That's exactly what happened to Vaughn in my current save.

You can't use "bad trades happen in real life" to excuse both the sheer volume of trades happening in the game or the nonsensical nature of most of them. If you haven't yet, Sim 3 years into a save and check it out for yourself. The problem isn't obvious until you are a few years in.

Pdubya64 · 03-28-2023, 06:30 PM

Quote:

Originally Posted by CBLCardinals

Good info here - I suspect you are going down the right path on how the AI is valuing ratings vs stats with the different weights, along with other variables such as contract amount, team budget, budget space availability, etc

Ditto Gussy, I am interested in what you turn up with your investigation.

SirMichaelJordan · 03-28-2023, 07:24 PM

These type of conversation are pointless without any data to explain what you are seeing. The game have way too many settings. For example, someone assume number of trades are too high but it’s actually on par with reality.

jtsly2000 · 03-28-2023, 07:45 PM

So disappointing to read all the problems with OOTP 24. I have been around since OOTP 2 but it looks like this years is a pass. Should not be having unrealistic roster management in the 24 version of the game!!

billyray1984 · 03-28-2023, 07:53 PM

I created a save with only the MLB, did not change any financials, blocked all trades until opening day, created a template with that save and did the following sims:

1. 25/25/25/25 Average trade frequency : from opening day to the trade deadline, the following trades were completed : 17 in Aprill, 10 in May, 17 in June and a staggering 181 in July.

In real life, in 2022, 75 trades were completed in July. I did not check the quality of the trades, but I did see a lot of teams retaining a lot of salaries and some head scratchers.

2. 25/25/25/25 Low trade frequency : from opening day to the trade deadline, the following trades were completed : 0 in April, 0 in May, 3 in June, 69 in July with the vast majority being in the last 5 days of July. Less teams were retaining salaries and those who did were clearly unloading bad contracts. Again I did not go with a fine comb, but did not see any major players get traded.

3. 25/25/25/25 Very Low trade frequency : from opening day to the trade deadline, the following trades were completed : 0 in April, 3 in May, 3 in June, 71 in July. Many teams were again retaining salaries. Christian Walker was traded fully retained for Jordan Balazovic. Not Arenado to the Cards bad, but still pretty bad since Balazovic was optioned to the minors.

This leads me to a couple of conclusions :

1. Obviously more data points would be needed.

2. 25/25/25/25, which I pretty much always used before in 21, 22 and 23 doesn't seem to sim well right now. I'll go back to the default setup or a more rating oriented player evaluation and maybe try again.

3. It seems that the AI REALLY likes to trade in July, which was not really the case in previous versions if I remember correctly. IRL, teams DO tend to trade a lot more in July than in other months, but 181 trades in July is a bit much.

4. Sortable trades for the league would be a godsend.

5. A lot of teams seem to have very little budget room in the first year, which might explain why so many trades involved retained salary.

PSUColonel · 03-28-2023, 07:54 PM

I am currently using 50/30/15/5 and setting the difficulty at 17 or 18 (I haven't decided yet) and I have trading frequency at the normal level. I leave the preference slider at neutral. I am also using Hard mode...not that most of this matters in AI to AI trades.

The only setting that can really control the AI trading is frequency, and how the AI evaluates players. In the past I have always been a proponent of higher ratings over stats, but with the new hard mode, I think it might be feasible to go 50/50. I would still like to see how the AI behaves with its trades as well as waivers and releases a bit more though. trading was a big reason I liked higher ratings in the past. 55/25/15/5 still might be the sweet spot for me though...I am still figuring it out a bit.

billyray1984 · 03-28-2023, 07:57 PM

Quote:

Originally Posted by jtsly2000

So disappointing to read all the problems with OOTP 24. I have been around since OOTP 2 but it looks like this years is a pass. Should not be having unrealistic roster management in the 24 version of the game!!

Every year there is some stuff to be ironed out and that's normal imho.

jtsly2000 · 03-28-2023, 08:15 PM

Quote:

Originally Posted by billyray1984

I created a save with only the MLB, did not change any financials, blocked all trades until opening day, created a template with that save and did the following sims:

1. 25/25/25/25 Average trade frequency : from opening day to the trade deadline, the following trades were completed : 17 in Aprill, 10 in May, 17 in June and a staggering 181 in July.

In real life, in 2022, 75 trades were completed in July. I did not check the quality of the trades, but I did see a lot of teams retaining a lot of salaries and some head scratchers.

2. 25/25/25/25 Low trade frequency : from opening day to the trade deadline, the following trades were completed : 0 in April, 0 in May, 3 in June, 69 in July with the vast majority being in the last 5 days of July. Less teams were retaining salaries and those who did were clearly unloading bad contracts. Again I did not go with a fine comb, but did not see any major players get traded.

3. 25/25/25/25 Very Low trade frequency : from opening day to the trade deadline, the following trades were completed : 0 in April, 3 in May, 3 in June, 71 in July. Many teams were again retaining salaries. Christian Walker was traded fully retained for Jordan Balazovic. Not Arenado to the Cards bad, but still pretty bad since Balazovic was optioned to the minors.

This leads me to a couple of conclusions :

1. Obviously more data points would be needed.

2. 25/25/25/25, which I pretty much always used before in 21, 22 and 23 doesn't seem to sim well right now. I'll go back to the default setup or a more rating oriented player evaluation and maybe try again.

3. It seems that the AI REALLY likes to trade in July, which was not really the case in previous versions if I remember correctly. IRL, teams DO tend to trade a lot more in July than in other months, but 181 trades in July is a bit much.

4. Sortable trades for the league would be a godsend.

5. A lot of teams seem to have very little budget room in the first year, which might explain why so many trades involved retained salary.

Wow! Great work. I commend you for rolling up your sleeves and producing a lot of useful data.

Gussy · 03-28-2023, 08:19 PM

Settings for trades:almost never, Ratings 40%, Current Year 30%, Previous Year 25%, 2 years ago 5%, max hard max favor prospects:

May: Cubs get Finnegan and Brandon House for McKinstry. Terrible trade for Nats.

Rangers trade Leody Tavares to Cubs for Ian Happ. Rangers currently lead AL West, probably not terrible, outside of all of the Cubs prospect depth is in the OF. Cubs are 10 games under .500 end of May.

June: Nothing really notable, a few contenders acquiring relievers from non contenders, no big prospects or stars moved.

July: Royals get Orioles to take Arnoldis Chapman by sending a better prospect for a worse prospect while both teams are in last. Interesting.

Padres got Nola from the Phillies for 2 mid level prospects. Phillies are 5 games under .500, Padres are in WC chase. Nola last year of deal but I'd think Phillies could do better than this.

Cubs trade Julian Merriweather and Jonathan Estrada to the Cards for a low level prospect. This wouldn't happen, Estrada is better than what they are getting back, but again, not earth shattering.

A's trade Seth Brown and low level spect to Braves for Orlando Arcia.

All in all not terrible. One really bad prospect trade, although AI only has House as a 3 star prospect, still made zero sense. Nola trade could have been better.

AI probably still valuing prospects a smidge too high, might need to move that down a notch and ratings down just a bit and statistics up just a bit.

Thoughts?

03-28-2023, 10:14 AM	#1
PepeSylvia Minors (Single A) Join Date: Mar 2019 Posts: 56	AI to AI trade logic is awful Currently in December 2025 of a save and the AI to AI trade logic is insane. Challenge mode, trading hard mode, neutral (as far as favors prospects/vets). A few examples: Brewers traded all of their young prospects (Turang, Mitchell, Frelick, Wiemer) for older unproductive players in 2023. Bobby Witt Jr (78/80, still in Arb, coming off a 5.1 WAR season) was traded to Cleveland for a 17 year old currently 20/70 and a 20/20 minor leaguer. Andrew Vaughn was coming off a 43 HR season in 2024 and was traded to Seattle for JP Crawford (who became a bench player for White Sox) Jordan Walker traded after the 2023 season for Blake Snell. Walker was in AA with a .991 OPS and had 80 potential. Bo Naylor traded for Starlin Castro in 2024, Naylor has solid ratings in this save. Zac Gallen was traded for two mediocre prospects in 2024 coming off a sub 4 era year. Anyone else experiencing this? A lot of teams are unrecognizable because teams are trading like there's no tomorrow and they have to make trades or they will implode. The only two remaining players from 2023 the Toronto Blue Jays have in their lineup are Vlad and Kirk, all of the others have been traded including Bichette, Chapman, Springer (dealt to Yankees), and Varsho.

03-28-2023, 11:09 AM	#4
billyray1984 Minors (Triple A) Join Date: Jun 2021 Posts: 216	I've also seen a lot of dubious trades between AI, so much so that I had to lower the trade frequency to very low because teams were basically emptying their prospect pools for bench guys or relievers. Since every AI team was doing it, it probably evens out in the end, but I don't remember 21, 22 or 23 being this crazy : it actually was the opposite where teams would hold on to dear life to any prospect they had and a vast majority of trades were AAAA guys, relievers or bench guys getting traded for one another. It's more realistic to have the AI deal prospects or big pieces since teams nowadays are more willing to go there (i.e. the Soto trade), but it's kind of extreme right now in my game. As an aside I know that it's still early and that a lot of work was made with the trade AI (reputation/hard mode) and that maybe that made the trade logic a bit wonky. I'm not complaining. Last edited by billyray1984; 03-28-2023 at 11:10 AM.

03-28-2023, 10:51 AM	#2
CBLCardinals OOTP Roster Team Join Date: Feb 2002 Posts: 1,773	Does real life MLB never have “awful” trades? Where would Glasnow, Meadows, and Baz to TB for Chris Archer to PIT rank? Also - are any of these players too expensive for a team to retain or are any in final contract season? Probably a lot more variables occurring than just surface level.

03-28-2023, 12:27 PM	#6
Guthrien Minors (Double A) Join Date: Mar 2011 Posts: 154	No, it's bad. Far worse than 23. It took me a while to see it. The teams right now will become warped because the AI does not value valuable players correctly.

03-28-2023, 01:49 PM	#7
Thundercrack All Star Reserve Join Date: Jul 2008 Posts: 851	Seeing the same, all settings neutral, average, AI is consistently trading stars in their prime for next to nothing. Never saw anything like it in any version before. Something is definitely off, so much so that I’ve gone back to 23 until it’s fixed

03-28-2023, 01:55 PM	#8
slugga27 All Star Starter Join Date: Jun 2011 Posts: 1,272	What are folks using for AI Eval settings? Sent from my iPhone using Tapatalk

03-28-2023, 04:15 PM	#9
Gussy Minors (Rookie Ball) Join Date: Mar 2017 Posts: 28	The AI eval settings is the key. I've played with it some this afternoon trying to find the right setting. I did almost never and very low with eval ratings of 45% ratings weight, 30% current year, 20% last year and 5% 2 years ago and had a very reasonable amount of trades, this was with favors prospects maxed. I had zero top prospects of any sort traded at all up to the deadline. When I changed from almost never to very low, with favors prospects one step from max, and 40/35/20/5 I got similar results. Reasonable amount of trades up to deadline, zero top prospects traded. Neither of those are the right settings for realism imo, because I'd like to see some top prospect movement at deadline, so I'll keep playing with it, but I think the AI eval settings are the answer, it's just finding the right combination.

03-28-2023, 07:24 PM	#14
SirMichaelJordan Hall Of Famer Join Date: Apr 2012 Posts: 2,701	These type of conversation are pointless without any data to explain what you are seeing. The game have way too many settings. For example, someone assume number of trades are too high but it’s actually on par with reality.

03-28-2023, 07:45 PM	#15
jtsly2000 Minors (Double A) Join Date: Mar 2002 Location: Canada Posts: 172	So disappointing to read all the problems with OOTP 24. I have been around since OOTP 2 but it looks like this years is a pass. Should not be having unrealistic roster management in the 24 version of the game!!

03-28-2023, 07:53 PM	#16
billyray1984 Minors (Triple A) Join Date: Jun 2021 Posts: 216	I created a save with only the MLB, did not change any financials, blocked all trades until opening day, created a template with that save and did the following sims: 1. 25/25/25/25 Average trade frequency : from opening day to the trade deadline, the following trades were completed : 17 in Aprill, 10 in May, 17 in June and a staggering 181 in July. In real life, in 2022, 75 trades were completed in July. I did not check the quality of the trades, but I did see a lot of teams retaining a lot of salaries and some head scratchers. 2. 25/25/25/25 Low trade frequency : from opening day to the trade deadline, the following trades were completed : 0 in April, 0 in May, 3 in June, 69 in July with the vast majority being in the last 5 days of July. Less teams were retaining salaries and those who did were clearly unloading bad contracts. Again I did not go with a fine comb, but did not see any major players get traded. 3. 25/25/25/25 Very Low trade frequency : from opening day to the trade deadline, the following trades were completed : 0 in April, 3 in May, 3 in June, 71 in July. Many teams were again retaining salaries. Christian Walker was traded fully retained for Jordan Balazovic. Not Arenado to the Cards bad, but still pretty bad since Balazovic was optioned to the minors. This leads me to a couple of conclusions : 1. Obviously more data points would be needed. 2. 25/25/25/25, which I pretty much always used before in 21, 22 and 23 doesn't seem to sim well right now. I'll go back to the default setup or a more rating oriented player evaluation and maybe try again. 3. It seems that the AI REALLY likes to trade in July, which was not really the case in previous versions if I remember correctly. IRL, teams DO tend to trade a lot more in July than in other months, but 181 trades in July is a bit much. 4. Sortable trades for the league would be a godsend. 5. A lot of teams seem to have very little budget room in the first year, which might explain why so many trades involved retained salary.

03-28-2023, 07:54 PM	#17
PSUColonel Hall Of Famer Join Date: Feb 2002 Posts: 13,088 Blog Entries: 37	I am currently using 50/30/15/5 and setting the difficulty at 17 or 18 (I haven't decided yet) and I have trading frequency at the normal level. I leave the preference slider at neutral. I am also using Hard mode...not that most of this matters in AI to AI trades. The only setting that can really control the AI trading is frequency, and how the AI evaluates players. In the past I have always been a proponent of higher ratings over stats, but with the new hard mode, I think it might be feasible to go 50/50. I would still like to see how the AI behaves with its trades as well as waivers and releases a bit more though. trading was a big reason I liked higher ratings in the past. 55/25/15/5 still might be the sweet spot for me though...I am still figuring it out a bit.

03-28-2023, 08:19 PM	#20
Gussy Minors (Rookie Ball) Join Date: Mar 2017 Posts: 28	Settings for trades:almost never, Ratings 40%, Current Year 30%, Previous Year 25%, 2 years ago 5%, max hard max favor prospects: May: Cubs get Finnegan and Brandon House for McKinstry. Terrible trade for Nats. Rangers trade Leody Tavares to Cubs for Ian Happ. Rangers currently lead AL West, probably not terrible, outside of all of the Cubs prospect depth is in the OF. Cubs are 10 games under .500 end of May. June: Nothing really notable, a few contenders acquiring relievers from non contenders, no big prospects or stars moved. July: Royals get Orioles to take Arnoldis Chapman by sending a better prospect for a worse prospect while both teams are in last. Interesting. Padres got Nola from the Phillies for 2 mid level prospects. Phillies are 5 games under .500, Padres are in WC chase. Nola last year of deal but I'd think Phillies could do better than this. Cubs trade Julian Merriweather and Jonathan Estrada to the Cards for a low level prospect. This wouldn't happen, Estrada is better than what they are getting back, but again, not earth shattering. A's trade Seth Brown and low level spect to Braves for Orlando Arcia. All in all not terrible. One really bad prospect trade, although AI only has House as a 3 star prospect, still made zero sense. Nola trade could have been better. AI probably still valuing prospects a smidge too high, might need to move that down a notch and ratings down just a bit and statistics up just a bit. Thoughts?