Do we know AI managers criteria for transfer targets?

[Orion_]

I'm asking because I regularly in my save came across the situation where among all my youngsters just one or two of them have VERY BIG interest from other clubs, there are like 5-10 interested clubs and dozens of offers once the transfer window starts while other have none.

The question is about known criteria of AI managers because to be honest I have editor enabled to track some club stats, like reputation growth etc so I out of pure curiosity checked those players. They are not the ones with highest CA/PA, not any favourable personality or even their reputation isn't higher than other players.

Why AI managers are going nuts about some youngsters and others don't bring any interest even if you want to sell them even below their value?

Just to clarify offers for those players aren't some outstanding numbers so not that much overspending, these are not situation where my 100k valued youngsters gets 10mln offers, more like 10 offers 100-200k, but there are exceptional differences in interest among players that I don't understand.

[Platinum]

I dont work for SI but based on what SO staff have said they take into account:

- Their team needs (based on managers preferred formations, and strength and depth of each position)

- Their transfer budget and how much they think the player will cost (transfer fee and wage)

- Their scouts perceived CA of the player

- Their scouts perceived PA of the player

- The players form (current and long term)

- Whether think think the player will want to join

Their will be other criteria Im missing out too.

Edited May 15, 2021 by francis#17

[Orion_]

Thanks for replay! This does not explain why some mediocre youngsters have interest of like 10 big clubs from my top division. I don't think they all need him. I'd also argue about that 'team fit'. We've all seen a lot of players being bought for a lot of money only to rot in reserves of big clubs.

Another interesting thing is that AI perceive player's PA more accurate than we (the players) can. For example I had a player that had some interest from other clubs, very minor but still. So to get rid of those offers that stop my game progression I set his asking price to be like ~20 mln. And then in next player's advice report my assman told to raise his asking price to ~70mln. It turned out boy had like 170PA.

[Freakiie]

Considering it's youngsters you're talking about, I'm guessing they're highly rated by their scouts. Yes, knowing the PA they might not  look good to you, but the scouts don't know their PA so there's a good chance that these players have high PPA according to them. As for why the entire league is interested, I do feel there's some background mechanics going on to popularize these "talented" youngsters. There's definitely something about them that attracts the AI. You often see it with low rep foreign talent as well, most have 0 interest in them (even at times when they're really solid targets for their age and/or have very high PA) and the ones that have interest shown in them instantly have multiple European (sub)topclub chasing after them.

Try making a separate save where you take over a couple of the clubs that are interested in these youngsters and send their scouts to have a look at your entire youth team. I'm pretty sure that the youngsters that are drawing interest will have ok talent and a relatively low price according to the scouts.

Oh and one case of your assistant thinking you underpriced a guy is hardly a sign that the AI knows PA. For your one example I've seen plenty of cases where my coaches told me a player still had plenty of room to grow, only to have already hit their PA, or the tons of players my coaches considered to be future star players, while having PA far below that level.

[Orion_]

32 minutes ago, Freakiie said:

For your one example I've seen plenty of cases where my coaches told me a player still had plenty of room to grow, only to have already hit their PA, or the tons of players my coaches considered to be future star players, while having PA far below that level.

That's one of the reasons I like to play with editor enabled. Not for straightforward cheat but more like checking how AI is evaluating some things. These can be some important lessons. Like randomly checking some youngsters where both of them had 2 stars PA and one was ~50PA and second one was ~110. I realize scouts/HoYD takes into account also probability of players development, not straight PA but that showed me for example that it's always worth signing all players from intake. This costs pennies while you can otherwise release really nice player. Maybe not another Leo Messi but solid backup one.

[Nacaw]

Not going to happen, nor should it. Frankly, the less we know about the intricate details in AI player evaluations the better. If this was figured out, it would lead to all sorts of exploits. I understand that in some games you need to go under the hood and figure out how the game dissects things, in order to beat the game at the highest level. But this isn't the case for FM. I would advise you to accept things as they are, and that not every single decision has to make sense in order to create a good football manager game. Looking at things you aren't supposed to look at is likely just going to create more confusion, because you can never see exactly what the AI sees, and the AI logic will be different from your own.

 

Besides, the AI always operates the same way, a shortlist of players are generated for each transfer window (probably right after the prior one closes, because that's where you will often see new clubs being interested in your players) and these players are the ones the AI will try to buy. However, if the AI makes a sale, it will need to replace this player also. Humans are more often the opposite, we buy the player we want first, then try to sell the player he is replacing. This often leads to situations where the human is struggling to sell unwanted players, and we get a fair few complaint threads about "rigged transfers" every year because of it. You don't need to know the system down to single criteria, it's more than enough know it loosely. 

 

4 hours ago, Orion_ said:

Thanks for replay! This does not explain why some mediocre youngsters have interest of like 10 big clubs from my top division. I don't think they all need him. I'd also argue about that 'team fit'. We've all seen a lot of players being bought for a lot of money only to rot in reserves of big clubs.

Another interesting thing is that AI perceive player's PA more accurate than we (the players) can. For example I had a player that had some interest from other clubs, very minor but still. So to get rid of those offers that stop my game progression I set his asking price to be like ~20 mln. And then in next player's advice report my assman told to raise his asking price to ~70mln. It turned out boy had like 170PA.

That's such an absurd thing to say. The AI is much worse than humans at evaluating who to buy or not to buy. Oh my god so much worse. Play for 10 seasons and you will see it for sure. They are at the total mercy of scouts and scouts can be really poor in this game. It's perfectly reasonable that a combination of factors mentioned by francis created interest in some of your players and not in others, even if you cannot see why.  

Edited May 15, 2021 by Nacaw

[Orion_]

2 hours ago, Nacaw said:

The AI is much worse than humans at evaluating who to buy or not to buy.

I've never said they are good at it. I was referring to the points that previous user wrote.

I realize AI choice of transfer targets is usually poor because I see what my scouts/assman are suggesting me.

Also that's the whole point of the topic. Why AI is so bad at transfers and goes for my random mediocre youngster with a lot of offers from multiple clubs (so there must be something that brings the clubs attention) while they don't even batch an eye on better ones.

[Nacaw]

To quote you exactly, you said that the "AI perceive player's PA more accurate than we (the players) can.". That is simply not true, and if it did you would be seeing different transfers.

Part of why the AI is bad at transfers is to keep a realistic game world. If all EPL clubs only went out and bought 160+ PA players the game would get very stale. The AI probably has about the same success rate as managers IRL when buying players. The problem isn't really the AI, it's that the user has access to way too much, way too detailed information and is better at using it.

In your specific case, since you keep going back and referring to it,  the AI has gotten reports on your players that is different from your coaching reports. They know less about these players than you do. Another big thing to consider, is whether think think the player will want to join. If a youth player has recently signed a contract, they won't leave no matter what. The AI is simply trying to find the best young players that want to join, for a price they can afford. And that's how they landed at your mediocre youngster. If you insist on knowing exactly why, then I am not sure FM is the game for you, because this will never be possible to tell. 

[Orion_]

My point is why they will try to go for the same 2 players every single transfer window completely ignoring the rest. It's just not very realistic. I have 43 players in U-19 team, and yet AI is going only after those 2 who are as stated before very mediocre even in my bottom table team.