Is anyone really underachieving on FM?

[El Payaso]

Started to think about, and noticed that I have had hundreds of careers on FM since 2007 and I have never been sacked or underachieved on my saves. With underachieving I mean that you would pick for example a team predicted to finish ninth and end up finishing below that or even relegated. Does this ever happen with human players? 

Edited May 4 by El Payaso

[Jonthedon26]

6 minutes ago, El Payaso said:

Started to think about, and noticed that I have had hundreds of careers on FM since 2007 and I have never been sacked or underachieved on my saves. With underachieving I mean that you would pick for example a team predicted to finish ninth and end up finishing below that or even relegated. Does this ever happen with human players? 

I can't say that I have experienced this. My first save on every version of FM is Birmingham City, and without fail I always get promoted to the Premier League within 2 seasons, and I have never been relegated back down again. The first season in the Premier League is sometimes difficult and a few times I have come close to getting relegated, but in the more recent versions I seem to do really well first season and massively overachieve. This year I got promoted to the Premier League in my second season, and came 8th in the first season in the Premier League. I would actually really enjoy the challenge of getting relegated again.

[Etebaer]

I have seen a lot of people underachieve over the years up to this very FM24.

Usually they will end releasing YT Vids and Twitch Streams/Vids quite early and pretend it never happened or do not talk about it at all.

I myself was recently sacked after promotion in a lower league save bcs the Board believed we could achieve serial promotion with none of the players having a contract bcs of the amateurstatus that would only change after the season had started so i lost the key players, was unable to hire against the more prominent league competition (bcs still low reputation) and i played with Daveincids realism patch and when the promotion did not happen i was made the sinner of the club and released of my duty at 3rd or 4th position in the league.

[Costav]

Since I started to use a more challenging way to play FM yes (only scouting and observed players searching an no visible attributes for players and staff, just the octagon of ability and to rely only on comparisons and written report when selecting/signing players), it happened a couple of times that I was underperforming compared to the initial predictions. Before this change in the way I play, never happened before.

Fun fact: every time I risked to be sacked I was coaching the team I am fan of! 

Edited May 2 by Costav

[NineCloudNine]

There’s a thread right now in the Tactics forum from a new player who can’t make anything work and is on the point of quitting.

The vast majority of players who struggle won’t even make it that far, they’ll just quit and  do/play something else.

The regulars here have persevered through multiple FMs, learned to navigate a mindblowingly complex UI, studied Rashidi tutorials on what ‘mentality’ means (clue: it’s not what you’d think it means), learned what to do when players throw tantrums, worked out how to buy all the wonderkids. And so on.

We’re the weird ones.

I trust SI to have a much better grasp of the experience of the playerbase than we do. They have actual data, research, patterns.

If you overachieve it’s because you’ve mastered the game and know all the ways to make it work right.

Well done, you can now /flex on the forums about how easy it all is. 

Edited May 2 by NineCloudNine

[Jonthedon26]

36 minutes ago, NineCloudNine said:

There’s a thread right now in the Tactics forum from a new player who can’t make anything work and is on the point of quitting.

The vast majority of players who struggle won’t even make it that far, they’ll just quit and  do/play something else.

The regulars here have persevered through multiple FMs, learned to navigate a mindblowingly complex UI, studied Rashidi tutorials on what ‘mentality’ means (clue: it’s not what you’d think it means), learned what to do when players throw tantrums, worked out how to buy all the wonderkids. And so on.

We’re the weird ones.

I trust SI to have a much better grasp of the experience of the playerbase than we do. They have actual data, research, patterns.

If you overachieve it’s because you’ve mastered the game and know all the ways to make it work right.

Well done, you can now /flex on the forums about how easy it all is. 

This is spot on!

I have played FM for years and years and years so I know it inside out, I know what I should and shouldn't do and how to deal with certain situations, which I am sure is the same for many experienced FM players. On the other hand, I simply can't get my head around some games that I myself am new to, such as the OOTP baseball series, and Hearts of Iron 4. I am simply terrible at those games and no matter what I have tried I just can't get seem to do well in them and I end up quitting and not playing them. 

[lied90]

I think the only ones underachieving or struggling are new players, or players who are extremely casual (buys FM once every 5 years or something).

With any game like this there will be a feeling of being overwhelmed by all the information, and you just don't know where to start or what is important. This is the case with lots of strategy games.

For veterans it's way too easy, so you have to play online or restrict yourself to playing with the monitor upside down while drunk and only sign players from Mongolia.

[Rodrigogc]

That's probably the balance SI should strive to achieve. Like some said, FM is hard to learn but easy to master. It's too hard for newcomers, but too easy for veterans. I hope they find some balance in the near future. I suppose SI can start by streamlining some features and reducing significantly the amount of information that newcomers have to deal with. That can be done also by improving massively the UI, that right now feels a bit messy. The tactic creator has to be overhauled, and the default skin should be more clean. In my opinion, FM08 has the best skin of all time, balancing the right amount of information with buttons on the screen that take you wherever you want in the game easily. Right now you have to click a lot of buttons to just find the prefered foot of the player, because it's hidden too much..

Reducing the overwhelming amount of interactions would do also in my opinion, it would save learning time for newcomers and also spare veterans who know exactly what they have to click. Less is more in this aspect. And then, improve AI decision making when squad building and developing youngsters would make it harder for veterans to overachieve. FM07's AI squad building is fantastic, and should be taken as a role model for future FM's. 

My experience, I used to underachieve more in past FM's but I feel the game is getting easier and easier. Maybe I became better, I don't know  - but seems like a common theme for experienced players these days. 

Edited May 2 by Rodrigogc

[bpecanha]

Me! I've always played without using attribute masking and relying on player search, but in my current save I've been trying to avoid that, only signing players who are fully scouted (or trusting on my judgement if they're not fully scouted).

Started with Club Brugge, predicted to finish 2nd in the league, but we were actually in 8th. I held my job until the January window because we were doing alright in the Conference League (I almost lost a match to Lincoln Red Imps in the group stage though). After the transfer window I went 5 matches unbeaten, but lost one and they sacked me. New manager actually got knocked out of the UECL in the Round of 16 (where I left) and finished the league in 11th, but he managed to win the league next season.

Then I went to Belgium 3rd Division with R Knokke FC, but the team was terrible. Once we lost 3 or 4-0 and I ironically praised the team because I was angry. They reacted well. For me that was the end and I've resigned after 4 months with my team in 18th (last place), although our xG said we should be 2nd (!!!).

After that I went to FC Blau-Weiss Linz in Austria and only now things are starting to look better. Got in a relegation battle when I arrived, lost 10 games in a row, but actually managed to save the club in the final day. Last season I finally overachieved in this save and finished 6th with a club predicted to be last (we don't have money, so signing good players and real gems without using player search is really hard).

Edited May 3 by bpecanha

[GreenTriangle]

14 hours ago, NineCloudNine said:

If you overachieve it’s because you’ve mastered the game and know all the ways to make it work right.

Really? To see, how simple it is ! Let me tell you something. Let's assume you are an expert. Let's assume that 99% of your decisions are correct, although, beware, in the game you don't have enough data/info for 99% ! Let's assume that a match has 16 variables (11 players with the appropriate instructions plus 5 other variables).

What is the probability of controlling the 16 variables without any mistake ? 0.99^16*100 = 85.1%. That's just for one match. If you play 10 matches, you will see 0.851^10*100 = 19.7%.

But what happens if your decisions are "correct" in proportion to only 95% ? 0.95^16*100 = 44.1% for one match and 0.03% for 10 matches.

The difference between a good player and an "expert" is the difference between 95% and 99%. And a "99%" can still be wrong in 8 out of 10 matches.

[Costav]

15 hours ago, Rodrigogc said:

I suppose SI can start by streamlining some features and reducing significantly the amount of information that newcomers have to deal with. That can be done also by improving massively the UI, that right now feels a bit messy. The tactic creator has to be overhauled, and the default skin should be more clean

Basically, you want to play Fifa! 

[Costav]

3 hours ago, GreenTriangle said:

Really? To see, how simple it is ! Let me tell you something. Let's assume you are an expert. Let's assume that 99% of your decisions are correct, although, beware, in the game you don't have enough data/info for 99% ! Let's assume that a match has 16 variables (11 players with the appropriate instructions plus 5 other variables).

What is the probability of controlling the 16 variables without any mistake ? 0.99^16*100 = 85.1%. That's just for one match. If you play 10 matches, you will see 0.851^10*100 = 19.7%.

But what happens if your decisions are "correct" in proportion to only 95% ? 0.95^16*100 = 44.1% for one match and 0.03% for 10 matches.

The difference between a good player and an "expert" is the difference between 95% and 99%. And a "99%" can still be wrong in 8 out of 10 matches.

I am sorry, but I have the feeling that your assumptions are totally wrong. You based your calculations on things that are not even known (ME functioning, luckily, has not been revealed by any player/youtuber/video-money-maker). 

Beside the number of variables, your main point is that ME and the probability of overachieving is based on a scenario where everybody play with the same players, the same attributes, the same tactics (that's what you are stating: the only variables are the players and "their control", having the control of Messi the same "difficulty coefficient" of controlling a third-tier league striker), letting out differences in players' ability, and all interactions that are more complex than a simple probabilistic calculation.

You may answer that, based on what I am saying, probability of success should be even lower. However, you are missing a very important point: AI does not master variables control (using your words). That's why (as in real life), you can win against better team and lose facing a "worse" team.

That's why I believe your assumptions are waaay too simplistic.  

Said that, the vast majority of experienced players are stating that overachieving is possible and easy.
If you would like to put this in statistical terms to see what overachieving is influenced by, you may want to use a linear regression model where y (the independent variable) is "good perform/overachieving" and some explanatory variables such as experience in playing FM, attention to the detail (highly-correlated to experience, but let's skip this part), tactic used, ability of your team, board goals etc...and an error term, where every non-visible effects are included. That's the point! We are discussing about the fact that, most probably (but we cannot be sure about it), SI made the game slightly easier over the years (for several reasons). This could be only captured in the error term, and is not visible to any of us.  

[drewdavis]

The players I think are really difficult 

[Svenc]

1 hour ago, Costav said:

Beside the number of variables, your main point is that ME and the probability of overachieving is based on a scenario where everybody play with the same players, the same attributes, the same tactics (that's what you are stating: the only variables are the players and "their control", having the control of Messi the same "difficulty coefficient" of controlling a third-tier league striker), letting out differences in players' ability, and all interactions that are more complex than a simple probabilistic calculation.

More importantly, the model is based on the assumption that human managers as well as AI managers were seeing "eye to eye". Kind of a bit in like football management. That's blatantly not the case. For a start, unless SI were able to code AI able to read (and influence) the second by second match play the way a better human player can, there's always gonna be a gap. Generally, AI management is the main factor in why most experienced players are eventually experiencing what they experience. 


There's two possible solutions to make that gap smaller:

- Improve AI management. 

- Limit input. Every option a player has is one where he can gain an edge over AI, eventually. This goes in particular for options influencing the second by second match play, e.g. individual player positioning with possession / without, general decision making with the ball, etc. The more micro control or rather influence you have, the wider the possible gap in between AI and player. AI is never gonna be as creative as a human player too. Limiting input naturally makes it harder for players who really struggle to do anything genuinelly "bad" too as a side-effect. The performance gap between AI and players becomes smaller across the board. Not arguing this to be the best route, mind.


Long-term squad management naturally plays a big role as well, but that can be filed under point 1: improve AI management.

 

 

Edited May 3 by Svenc

[GreenTriangle]

5 hours ago, Costav said:

 This could be only captured in the error term, and is not visible to any of us.  

And if it's not visible, doesn't exist? Are you saying that you can't be wrong because you don't see that "error term" ? You can make mistakes all the more if you don't know what to avoid.
No, I base my calculations precisely on the fact that you use a number of units ("players") defined by more than 40 parameters each, units that act in groups of 11, each unit being strongly influenced by a set of dozens of parameters (" instructions" and "team parameters"). These units act within an event (match) in which they meet another set of units managed by the AI. That AI management is simplistic ? That's correct. Does this mean that the so-called "experience" of the human player can generate exactly the best possible response ? How do you prove this ?

Quote

 

4 hours ago, Svenc said:

This goes in particular for optins influencing the second by second match play, e.g. individual player positioning with possession / without, general decision making with the ball, etc.

Of course, the idea is that an expert "sees" what is not working according to his plan and corrects it, and thanks to his experience, these corrections are more effective than the initial state. It is also fair to say that AI, although it can self-correct some actions depending on the evolution of the event, is not at all effective when it does so. So, indeed, the human player has an advantage. But from here to affirming that an expert can always evaluate the direction and intensity of trends is an extremely long way.

Edited May 3 by GreenTriangle

[deltasfd11]

Em 21/04/2024 em 07:04, MichaelNevo disse:

There can be multiple reasons this happens, but the most obvious one would be your goalkeeper cracks under pressure too easy. The reason for that can most probably be found within his hidden stats which make up his personality. I'd maybe look there first, find someone with a bit more grit and determination.

You would automatically feel that the game is screwing you, but chances are it's actually a flaw within your team, work out the flaw and make the change and success will be yours.

the engine makes sure you are sucessfull most of the time even if you suck you will win a fair amount of games

[Costav]

1 hour ago, GreenTriangle said:

And if it's not visible, doesn't exist? Are you saying that you can't be wrong because you don't see that "error term" ?

I am sorry, I thought that you had a bit of familiarity with statistics, given the fact that you used it to explain your point. The error term in a regression model is something that you cannot see/explain when conceiving the model, like the functioning of the ME. Nobody said that does not exist. Nevermind.
 

My answer was based on the model that you made up based on assumptions that, although they seem to you very clear and reasonable, are completely out of sense.

A short example is the following: 

10 hours ago, GreenTriangle said:

What is the probability of controlling the 16 variables without any mistake ? 0.99^16*100 = 85.1%. That's just for one match. If you play 10 matches, you will see 0.851^10*100 = 19.7%.

So, if you play with only 11 players "without any mistakes" instead of making all 5 substitutions: 0.99^11*100 == 89.53% > 85.1%. If you play 10 matches (using your formula) == 33.1% of winning all the matches >19.7%. The trend is clear.

In a very absurd way, if you play with only 1 player (you control only one variable) you have almost 100% of possibilities to win 10 matches. Does it make sense?

 

Edited May 3 by Costav

[GreenTriangle]

35 minutes ago, Costav said:

Nobody said that does not exist.

***

So, if you play with only 11 players "without any mistakes" instead of making all 5 substitutions: 0.99^11*100 == 89.53% > 85.1%. If you play 10 matches (using your formula) == 33.1% of winning all the matches >19.7%. The trend is clear.

Aha, so you can't make mistakes simply because the model doesn't "see/explain" your errors/mistakes.
It would be interesting if you could explain to us how you can play FM with only one player. There is a limit in terms of what kind of hypotheses can be done.

On the other hand, logically speaking, why do you replace a player if, from the point of view of the correctness of the decision to use him, you were right 99% ? Or were you wrong, and in this case the risk of not replacing the players who are used based on a wrong evaluation is greater than that of replacing them ?

[GreenTriangle]

1 hour ago, deltasfd11 said:

the engine makes sure you are sucessfull most of the time even if you suck you will win a fair amount of games

You can win more matches if you use intensively, say, reloading and an editor. Of course, it is everyone's right to use whatever they like, because the purpose of the game is to generate fun for the player. But the question is... is the person who uses a lot reloading and editors "more expert" than the one who doesn't use them? According to the "results" theory, a player is the more expert the more he uses the two tools.

[Costav]

I am just using your example to demonstrate that what you are trying to explain makes no sense.

So, let's start again: @NineCloudNine says that "If you overachieve it’s because you’ve mastered the game and know all the ways to make it work right.", which implies two main things:
1. a person who has a lot of experience can be generally defined "a better performer" compared to somebody with less or experience at all.  
2. The more you have experience, the better you know how to make things work right.

Then you answered that the difference between a good player and a very experience player is resulting in 4% (based on what, exactly, we cannot know...) more in taking good decision, mentioning a model that you clearly made up and that is stated nowhere in the SI/FM universe. Moreover, you provide some calculation that make you say that an expert "can still be wrong in 8 out of 10 matches."

I try to show you that your assumptions are too simple (using examples as the competitiveness of a player, the interaction between choices during a match, among others) and that nobody, luckily, can know how exactly the ME works, adding also an example about linear regression model to describe the fact that, if the model was tested, is presumably correct to say that the more experience you have the better you play and perform in FM (again, logical). In addition, I use your model to show the limit of it (less players you "control", higher the chance to win) and you tell me that I am wrong, because i need to hypothetically replace a player, making an example that is perfectly in contrast with you model.

Your prediction model is static, you assume there is interaction only between controlled variables and % of making a good decision and using probabilistic calculus you say that over 10 matches, if you make 99% of correct decisions, you have only 20% of possibilities to make everything perfect. Being static, your model cannot include substitution decisions or similar (which are, by nature, based on the evolution of something that is happening), which are exactly what you've just brought as an example.

In addition to this, something else: you don't need to do everything perfectly to overperform. You just need to perform better than the AI. As mentioned earlier (jointly with @Svenc), AI does not take every correct decision.

You're confusing probability with likelihood.

Another example using you model.
A non-expert player (let's say 50% probability to control every variables, using your terminology) coaching Man City, in a match, will have 0.5^16*100=15.25% of possibility to be right. So, playing 8 matches, s/he will be wrong in  99.8% of the cases (because (0.5^16)^8matches*100=0.2%). Does it make sense? No.

Does your model include strength of the opponents? No.

Edited May 3 by Costav

[deltasfd11]

2 horas atrás, Costav disse:

I am just using your example to demonstrate that what you are trying to explain makes no sense.

So, let's start again: @NineCloudNine says that "If you overachieve it’s because you’ve mastered the game and know all the ways to make it work right.", which implies two main things:
1. a person who has a lot of experience can be generally defined "a better performer" compared to somebody with less or experience at all.  
2. The more you have experience, the better you know how to make things work right.

Then you answered that the difference between a good player and a very experience player is resulting in 4% (based on what, exactly, we cannot know...) more in taking good decision, mentioning a model that you clearly made up and that is stated nowhere in the SI/FM universe. Moreover, you provide some calculation that make you say that an expert "can still be wrong in 8 out of 10 matches."

I try to show you that your assumptions are too simple (using examples as the competitiveness of a player, the interaction between choices during a match, among others) and that nobody, luckily, can know how exactly the ME works, adding also an example about linear regression model to describe the fact that, if the model was tested, is presumably correct to say that the more experience you have the better you play and perform in FM (again, logical). In addition, I use your model to show the limit of it (less players you "control", higher the chance to win) and you tell me that I am wrong, because i need to hypothetically replace a player, making an example that is perfectly in contrast with you model.

Your prediction model is static, you assume there is interaction only between controlled variables and % of making a good decision and using probabilistic calculus you say that over 10 matches, if you make 99% of correct decisions, you have only 20% of possibilities to make everything perfect. Being static, your model cannot include substitution decisions or similar (which are, by nature, based on the evolution of something that is happening), which are exactly what you've just brought as an example.

In addition to this, something else: you don't need to do everything perfectly to overperform. You just need to perform better than the AI. As mentioned earlier (jointly with @Svenc), AI does not take every correct decision.

You're confusing probability with likelihood.

Another example using you model.
A non-expert player (let's say 50% probability to control every variables, using your terminology) coaching Man City, in a match, will have 0.5^16*100=15.25% of possibility to be right. So, playing 8 matches, s/he will be wrong in  99.8% of the cases (because (0.5^16)^8matches*100=0.2%). Does it make sense? No.

Does your model include strength of the opponents? No.

I understand what you say. but...

the issue is you assuming that FM is 100% accurate or even close to simulating performance from the coach. FM is just another game and its all about programming, programming cant simulate or cover or 100% simulate performance. 

this means that the game engine will naturally allow you to win a considerable amount of matches within your level, your team level. it will even allow you to beat big teams with that system of underdog position wich in FM ends up being favorable for you you can play without pressure and beat bigger teams. in real life it doesnt work like that...

it shows that FM systems allow you to win some matches.-

it doesnt matter what tatics you use, if you undestand the minimum on football  tatics you can have some sucess...

FM does allow you win a fair amount of matches to keep you playing the game.

every game does this software things...

anyone can have some sucess on FM, its fairly easy to win even if you are new. 

you just need to avoid the engine traps.

naturally the engine to keep you playing also makes you win games within a fair amount of sucess.

itas just a game...

they cant even make scouts find players within their own database. or the injury endless thing..

or the broken transfer system...

 

in the end anyone can have some sucess, the games allow it and makes it favorable..

never heard of anyone underachieving

[Spallo]

I used to do the same - argueing with people who clearly don’t understand the basics of what the seem to talking about.

@GreenTriangle clearly has no idea what you @Costav are saying 

so just leave at that, because you can’t argue with opinions who disguises as smth more. Maybe it’s the dunning Kruger effect - who knows 

[NineCloudNine]

I was wondering if someone who actually knows a bit about modelling would chip in .

Let me expand a little on the original statement of mine that triggered this debate:

On 02/05/2024 at 13:43, NineCloudNine said:

If you overachieve it’s because you’ve mastered the game and know all the ways to make it work right.

Firstly that's obviously hyperbole because "know all the right ways to make it work" implies a degree of omniscience that none of us have about how the ME works.

A better way to put it might be:

1. An experienced player knows which variables have the biggest effect on performance in FM and which can be ignored/delegated (eg, morale is vital, role stars can be ignored, training can be delegated)
2. An experienced player knows things that the AI doesn't know (for example, wonderkids are a great transfer policy)
3. An experienced player won't make rookie errors (like making promises you can't keep or clicking the 'wrong' option in a player interaction)

@GreenTriangle's model is actually an interesting starting point for understanding a real football match. There are dozens or even hundreds of variables, all available to both managers (and indeed players). Football is a low-scoring game, which adds a large amount of variation in outcomes. The margin of 'success' here is very small. Pep Guardiola, widely considered one of the greatest managers of all time, has a career win rate of 'only' 73% despite always having managed at clubs with title-winning ambitions and quality. So the small margin between a 'good' manager and an 'expert' one might indeed only be a few percent.

That model doesn't apply to FM though, because the expert player doesn't just know how to control variables, they know which ones can be entirely ignored and which are crucial. The margin over the AI is substantial. 'Success' here is also not a win/lose binary, but (as @Costav says) relative to the expectations of the club and the strength of the opposition.

Edited May 3 by NineCloudNine

[GreenTriangle]

3 hours ago, Costav said:

Does your model include strength of the opponents? No.

how do you quantify the strength of your opponents, considering the countless variables involved ? you can evaluate it with some approximation, taking into account the possible influences of morale, fatigue, players' interest in the match as well as the way in which you can assume that the AI "sees" you. how do you get a more precise quantification than that ?

[GreenTriangle]

1 hour ago, Spallo said:

I used to do the same - argueing with people who clearly don’t understand the basics of what the seem to talking about.

@GreenTriangle clearly has no idea what you @Costav are saying 

Are sure you have any idea ? Just one question : when you play FM you can control absolutely everything, or are there always parts of the game you don't know much about, so you can only evaluate them very roughly?

another question. you know, of course, what those hidden attributes are. let's assume that you use a certain method to visualize their numerical value. knowing those numerical values (as well as the level of visible attributes), can you predict the way a certain player will play in the next match? or can you just estimate, as a probable trend, the fact that that player will play well enough?

Edited May 3 by GreenTriangle

[NineCloudNine]

13 minutes ago, GreenTriangle said:

how do you quantify the strength of your opponents, considering the countless variables involved ? you can evaluate it with some approximation, taking into account the possible influences of morale, fatigue, players' interest in the match as well as the way in which you can assume that the AI "sees" you. how do you get a more precise quantification than that ?

11 minutes ago, GreenTriangle said:

Are sure you have any idea ? Just one question : when you play FM you can control absolutely everything, or are there always parts of the game you don't know much about, so you can only evaluate them very roughly?

Neither of these things are necessary. Football is far too complex to be modeled with the precision you are demanding here and the need for that precison is a strawman you've created. As @Costav said you are conflating probability and likelihood. Your initial response to me was also based on taking literally a comment that was obviously not intended to suggest that an experienced FM player has complete knowledge or control of every variable. I have expanded on that in my post above.

Edited May 3 by NineCloudNine

[Poison]

Well I've been sacked many times with any kind of team imaginable, title favorites where I've lost title, relegation fodders which i have not managed to save despite playing well, mid teams that didn't fare well (sacked within objectives). The last one is the most difficult, even if you are predicted 9th and finish 12th the board is not going to sack you probably.

[GreenTriangle]

20 minutes ago, NineCloudNine said:

Neither of these things are necessary. Football is far too complex to be modeled with the precision you are demanding here and the need for that precison is a strawman you've created. As @Costav said you are conflating probability and likelihood. Your initial response to me was also based on taking literally a comment that was obviously not intended to suggest that an experienced FM player has complete knowledge or control of every variable. I have expanded on that in my post above.

I presented a point of view. It's interesting that the reaction was strange, although what I was saying is that this game also involves a dose of unpredictability. Anyone could admit this without any problem. But they wanted to prove to me that in fact this game is totally predictable and that some real experts know absolutely everything. You have played FM, I assume, and you know that no one knows "absolutely everything" about this game. In fact, even an expert can make mistakes sometimes. Of course, an expert makes much less mistakes and he's able to quickly identify the mistake and also can correct that mistake much more effectively.

Regarding the subject of the topic: it is thought in terms of "results". If someone can present a list with as many positive results as possible, he is already an expert. Seriously ? Of course, the results also matter, but they are not the only aspect. Does the fact that someone has been relegated mean that he is a beginner? How about taking other aspects into account ? For example, the increased difficulty of the respective save. No very good player would say that simple relegation or dismissal is, by definition, a way of measuring another player's ability. Additional details are always needed, because in certain cases (and in certain contexts) it can be more difficult to manage a certain team than to win a trophy. Or ... do you not agree with this ?

[NineCloudNine]

20 minutes ago, GreenTriangle said:

But they wanted to prove to me that in fact this game is totally predictable and that some real experts know absolutely everything. You have played FM, I assume, and you know that no one knows "absolutely everything" about this game. In fact, even an expert can make mistakes sometimes. Of course, an expert makes much less mistakes and he's able to quickly identify the mistake and also can correct that mistake much more effectively.

No-one has sought to prove this. You have created a strawman, the doubled down on it. You took too literally the statement of mine that you quoted. I have expanded on this above. 'Success' in this context also does not mean winning every match, it means outperforming the expectations of the club the player is managing, which might well mean merely avoiding relegation - and no-one in this thread has said otherwise.

You are correct to point out that the number of variables involved in a football match is enormous and cannot possibly be fully controlled. You are also correct that this means the margin between bad, ok, good and great managers is surprisingly small statistically. However, these conditions do not apply to FM because of the disparity between what the player knows and what the AI knows. While much remains unknown or uncontrollable, the ability gap between an experienced FM player and their AI opponent is vast and can easily overwhelm the element of unpredictability and randonmess which remains.

Edited May 3 by NineCloudNine

[Svenc]

17 hours ago, NineCloudNine said:

 

@GreenTriangle's model is actually an interesting starting point for understanding a real football match. There are dozens or even hundreds of variables, all available to both managers (and indeed players). Football is a low-scoring game, which adds a large amount of variation in outcomes. The margin of 'success' here is very small. Pep Guardiola, widely considered one of the greatest managers of all time, has a career win rate of 'only' 73% despite always having managed at clubs with title-winning ambitions and quality. So the small margin between a 'good' manager and an 'expert' one might indeed only be a few percent.

That model doesn't apply to FM though, because the expert player doesn't just know how to control variables, they know which ones can be entirely ignored and which are crucial. The margin over the AI is substantial. 'Success' here is also not a win/lose binary, but (as @Costav says) relative to the expectations of the club and the strength of the opposition.

 

Absolutely agree with the latter. Just seen this was also discussed here, like me with the argument of ease of (match play) control at the center of things. Also, if SI coders were able to code an AI that can actually read the ME (whether for bugs or actual play) as well as the better/experienced FM players, they likely wouldn't be in the gaming business.  

Re: Pep, to me the general match simulation in FM has never been as "random" as football can be either. Sure, every year you have players ranting about how parking bus tactic AI teams would score with their first shot on target once in every blue moon. But I've never seen a side on this like Klopp's Dortmund in his last spell there. They were sitting at 18th (!) place come December, and everybody wondered WTF was going on, German pundits and TV "experts" even arguing about whether Klopp would be "finished". LFC staffers (luckily for the club) had figured that a big part of it was but rotten luck. They determined Klopp's Dortmund should be actually sitting ~4th place, rather than 18th.

Or how about Zidane in CR7's last season at Real? Zidane had already lost the title race to Barcelona come January, sitting 4th. A big part of the reason? All of his main strikeforce having the worst spell of their careers. CR7 averaged 7 shots per match alone (xG ~13) -- but only had scored 4 goals by that point, two of which penalties. Benzema et all didn't fare better, relative to their finishes. Whilst there's still funny headlines recorded for all eternity, I wish the Youtube video was still online showing all "so close" finishes in chronological order -- the only thing missing was the Benny Hill theme playing alongside to it. And at least public data seems to suggest that even Pep may once have been a bit of a victim of "chance". As every single season since he arrived at City, his side had by far the best data, including Expected Goals Difference, which clocks chance data over a season, rather than individual matches. Even in the season where Klopp would finally win it with Liverpool. 

Whilst "implementing" this may even keep the experienced players a bit on their toe every once in a while: Can you imagine this game's community if a Klopp/Dortmund or Zidane/Real happened to but one of the millions players ever?  

Edited May 4 by Svenc

[Footix]

Although I wouldn't say I'm underachieving, I'm not successful like most of you lot either. I sit in midtable, occasionally in the top 5. I get knocked out of cups in the earlier rounds, and I can never seem to find a really good transfer. I usually end up paying too much money for a player of similar calibre to those already in my squad.

I can't remember winning anything, except for the Championship one year. 

Season after season, year after year. England, Germany, France or Scotland, it's the same story. 

I don't know why I'm still playing. 

[phnompenhandy]

On 02/05/2024 at 19:27, Etebaer said:

Usually they will end releasing YT Vids and Twitch Streams/Vids quite early and pretend it never happened or do not talk about it at all.

I was following a fella on Twitch who was doggedly trying to gain promotion season after season and always fell short. It's a weak league where the champions go into a play-off league with winners of stronger leagues and the winner of that plays-off against the last team in the tier above. To be fair, the winners of his league have never been successfully promoted in real life, but FM is supposed to be easier! Anyway, he had more stamina than me, that's for sure. Never did get promoted after hundreds of hours of playing.

[Etebaer]

I had a lower league save a few FMs back when i was 7 or 8 seasons in german tier 4 but it was quite realistic as in that league you have only the champion having a promotion play off with another champion and it took 5 seasons to become champ and than 2 or 3 more to win the play offs...but i liked it as it felt precious to win.

[wearesporting]

I've been sacked. So I guess I do from time to time. Found it more likely on console oddly enough even though it should in theory be easier.

[Brother Ben]

It took me 20 odd seasons to get out of non-league but then I refuse to trial more than a few players and i'm not that keen on too many loans for realism sake.  My own "one hour's drive" rule made it super hard too.

When I eventually did get promoted though it was party time in my house, I took a screenshot of my promotion winning side and I'm very tempted to get it framed.