Is there really too much randomness in the game, or is it all a myth?

[Loversleaper]

Loversleaper,

I get your point and granted, if we were in the tactical discussions forum and looking ofr a tactic that works or improvements then you are right.

However, this thread initially was not about tactics so why would chopper99 have had to provide info on his tactics? No point in that.

Then, you say the other examples are worthless because there's no mention of the tactical setup. Now, I think that there are ciscumstances where mentality, passing and closing down settings ar just irrelevant, and we've seen some examples.

a 5-0-5 should not work. I admit it is the easiest way to get the ball through midfield quickly but if it was that easy I'm sure some teams would do that IRL. Now I'm eager to see RL examples if you know some.

It's the same for the 1-2-2-5 formation. If you play such a formation for 90 minutes you should be able to get away with it against a team 5 or 6 leagues lower than you, but not in the Prem or in Europe - no matter how you setup your players. I am still not satsfied that you can get anything out of any game in the league with such a formation. I'm not particulary worried about the lack of defenders, but there are no defensive mids whatsoever and you should not be able to close these gaps with the mentality sliders. If you can, then I guess I'll field my best players from now on, in their best individual position and do the rest with mentality settings.

I hope you get my point. There are certain formations that just should not work.

Answer one question? If there was some guy asking for advice in the tactics forums playing a 1-2-2-5, what would be your first advice? Look at his team's settings or use a different formation?

In the end, all you're saying is either something that you handle with care yourself ("how to counter the AI"-discussion) or proves the point that most people posting in this thread are making: You can't get your players to act like you want them in the ME unless you get a number of sliders correct. And I'm not talking about overall performance as this is of course reflected by the various settings. I'm talking about things like staying deep, closing down or maving the ball around quickly. You prove that point b saying that all us people who can't do it lack understanding of the tactical system or the ME. I don't think I'm stupid and I don't think I have no idea of this game whatsoever, and I'm sure most girls or guys posting here feel the same and are right.

Still, we're setting the sliders, are doing well, but all that without seeing our team play like we want them to. This game is not just about winning, but losing is even more frustrating when you have no clue why your players will nowhere near do what you want them to.

Well, then they should have posted a 'screenshot' of the slider itself and said this is the problem. The main point of the Thread was to show if a tactic is 'consistant' result-wise, and anybody that would make such a post would know if they make 'contradicting' settings in the tactic options they could do just that. The game is built up by calculating occurances in the game, and it won't read your mind. Maybe the weird formation part has a point but the test concluded that you can't win just on 'Team Potential' alone as claimed in this post, so I have been suggesting that before people make assumptions they know what they are talking about first and can make a 'guy like me' understand why. But you know what, I'll just let them get on with it...

[Nomis07]

The main point of the Thread was to show if a tactic is 'consistant' result-wise,

I think you've missed the point of the thread

FM and before that CM have always had a random element that negates tactics in certain scenarios, this is in order to maintain the simulation and try to generate a real football experience in which tactics aren't the be all and end all. The thread is about that random element and has nothing to do with tactics whatsoever.

[chopper99]

The main point of the Thread was to show if a tactic is 'consistant' result-wise,

No, it really wasn't. The point of the thread was to test the random element in the game, using the same game over and over again. The consistancy of the tactic had absolutely nothing to do with it, how could it when it was simply the same game played over and over again using the exact same tactic, team talks etc.

[Loversleaper]

So the point the Rupal and Chopper99 in the last couple of posting suggest that the out-come should be the same no matter where you put the settings as long as it is consistant?...

That means that everygame in the world has to follow a certain pattern or you will fail? I don't know where we are anymore and I can see why that you don't understand me. Forgive the confusing aspects I have brought forth. Good luck with the testing and I hope you get the satifaction you require...

[chopper99]

So the point the Rupal and Chopper99 in the last couple of posting suggest that the out-come should be the same no matter where you put the settings as long as it is consistant?...

No, what I was saying (and testing) is the claim made by some people that they can play a match and win 4-0, quit the game and play the same match with no changes at all and lose 4-0. This made them feel that tactics didn't matter as it was random whether you won or lost a game, and as a result made them feel like they had little control over results due to this random nature of replayed matches.

By re-playing the same game over and over (regardless of tactics, as if they're bad and I lose heavily then I should lose the majority of the matches in the test and vice verse) you should be able to see a pattern and the results should differ a little due to 'luck' and such being present in real football, but overall the results should not vary too wildly.

My initial tests showed that this was true for that test, more tests are needed.

[Loversleaper]

No, it really wasn't. The point of the thread was to test the random element in the game, using the same game over and over again. The consistancy of the tactic had absolutely nothing to do with it, how could it when it was simply the same game played over and over again using the exact same tactic, team talks etc.

You are right, tesitng a tactic if it is random or consistant are two entirely different aspects. Thanks for clearing that up for the Community, I understand your point a lot more now. Time for lunch...

[Rupal]

This whole thing has become confused and I suspect we've all contributed to it.

There's already been loads of discussion about the effect which an early event (eg hitting the post and the ball going in or not going in) can have on the outcome, elsewhere. We've had spreads of results quoted from 0-3 to 4-0 with identical setups and people have argued back and forth about how far this range is acceptable, whether there are too many things apart from tactics affecting the results, whether 'luck' plays too big a part and so on.

PaulC has assured us that results aren't predetermined and that how we react to what goes on on the pitch is a major factor in determining the final result.

What makes it very hard is that what we do when reacting to events unfolding on the pitch doesn't always appear to result in our players doing what we want them to do. It's the complexity of the alterations which you may need to make which causes the problem and the difficulty of 'reading' what's going on because the ME is quirky which makes it hard to decide what to do. I hope that the rethink about the tactical system, which we are told is being actively considered, will help matters.

[Squirmy Rooter 2.0]

Chopper,

I seem to recall (and this was years ago, so things might have changed since then), that there was a point just before a match where you could save the game and the random number generator for the match would already have been seeded and saved as well, meaning that unless you did anything differently, the match would play out identically.

I don't know if this is still the case since the new "match-flow" was added (or indeed if it ever was the case - I could have dreamt it!), but if so you might just need to take it into account when saving your "restore-point" for your tests, so that you don't get the match playing out perfectly identical each time.

[chopper99]

Chopper,

I seem to recall (and this was years ago, so things might have changed since then), that there was a point just before a match where you could save the game and the random number generator for the match would already have been seeded and saved as well, meaning that unless you did anything differently, the match would play out identically.

I don't know if this is still the case since the new "match-flow" was added (or indeed if it ever was the case - I could have dreamt it!), but if so you might just need to take it into account when saving your "restore-point" for your tests, so that you don't get the match playing out perfectly identical each time.

I have vague memories of the same thing now that you mention it. It's something I'll keep an eye out for

[Äktsjon Männ]

You know what. I just realised that someone just wrote that there are 20 different options...well since we are so far out there why not 22 then (the team mentality slider has 22 notches). I don't think I will waste my time anymore...

When the AI attacks, you defend.

When the AI defends, you attack.

When the AI plays normal you do to.

This is a basic way to play the AI and you missed this somewhere along the way so your response is "so what?". You are a big man....

This should not be the basic way to play the AI. If Man Utd were playing Leyton Orient (just as a random example) and Orient decided to suddenly go attacking would United switch to ultra-defensive or would they continue doing their thing and simply destroy the opposition in the progress because they're vastly superior? It's too easy for the AI (or human player for that matter, if they manage to get the settings right) to just suddenly turn things around completely and take the game to the opposition by playing great possession football. Part of this randomness argument comes from that as it's not only the results that vary but the nature of the game as well. Completely dominating possession and beating the opposition comfortably and then being outplayed in the same manner after a reload is a much bigger problem than winning 4-0 and losing 5-0 at the second attempt.

I would like to see more thorough analysis in future tests if someone wants to take this further, such as a comparison of possession percentages or chances created as well as other key statistics.

[Loversleaper]

This should not be the basic way to play the AI. If Man Utd were playing Leyton Orient (just as a random example) and Orient decided to suddenly go attacking would United switch to ultra-defensive or would they continue doing their thing and simply destroy the opposition in the progress because they're vastly superior? It's too easy for the AI (or human player for that matter' date=' if they manage to get the settings right) to just suddenly turn things around completely and take the game to the opposition by playing great possession football. Part of this randomness argument comes from that as it's not only the results that vary but the nature of the game as well. Completely dominating possession and beating the opposition comfortably and then being outplayed in the same manner after a reload is a much bigger problem than winning 4-0 and losing 5-0 at the second attempt.

I would like to see more thorough analysis in future tests if someone wants to take this further, such as a comparison of possession percentages or chances created as well as other key statistics.[/quote']

Why is it when I say 'basic' you present it as 'set in stone'. If attacking and defensive ideologies don't make up a football game then can you kindly tell me what does. Of course the game is taking players atributes into account when determining the success of the tactical ploy you might attempt and I have NEVER suggested otherwise.

With this statement you clearly haven't been following along and before this I felt like I had a monkey on my back explaining what it is I am trying to get accross and now thanks to you, I feel as I have now the planet of the apes on my back.

Fair enough that people want to make tests, but as anything else in this world you have to make things clear (as you youself requested in your post, so I don't know if I should thank you for beeing on my side...or...) so that people following knows the logistics behind the theories. Now I, and probably many others, would like to help explain/paticipate in these types of issues (let's not forget that I am on the FM gamers' side here, regardless) and can test things out playing games as I know for sure that some understand the ME better than others.

Maybe the game is too complex, but let's all remember that the game, in it's current state, is relatively new, so it might take some time to adjust to a new ME for a lot of FM Gamers. Maybe the SI hasn't been clear enough. Fair enough lets all gang up. If the testing/assumptions really does make sense then it would be easier to identify the problems for everybody involved unless we are all going to have to 'jump of the cliff' together because someone said so...

The settings are important because causing contradicting settings cause randomness and the average gamer should know why. As any game on earth the sequences in any game are more or less calculations (as in mathmatics) of current situations which gives certain values. The Game is not like the computer in the movie 'I, Robot', and if the calculation early in the game (lets say you go down by a goal) get tipped one way on the 'calculation scale' you will see that the effects will vary (maybe due to mentality, counter-attacking on or off, ect...). How would we know if the Tester is using counter-attacking + short passing on a wet pitch....you don't think this might cause som Statistic mayhem? Otherwise it simply isn't the correct path to show if things in this ME are functioning or not.

Now I suggest that PaulC and his team make two versions of the ME so that gamers have the choise of using the 'old ME' and then you guys can always come over to the Tactics Forum and get the the tactics that you won't have to tweak (I have some of the old ones still laying around, for example) then this wouldn't be an issue...

[BlackTiger]

Come on guys, this very useful thread is sliding dangerously towards childish bickering and arguing over who is right and wrong. We ALL have opinions on how the game is played and how it should be played and we dont need another decent thread closed because of immature behaviour.

No one is doubting that the tactical side of the game is immensely detailed and difficult to understand, however, with time, effort and a lot of patience you can design a suitable set of tactics that can do you good over a long career. At the moment for me personally, its not the tactical side of the game thats annoying, its the fact that im losing games i should be winning due to idiotic mistakes by defenders and strikers which are quite clearly bugs. Luckily PaulC has taken notice of a lot of PKM's uploaded so some (or hopefully all) of these should be fixed for 9.3.

Im confident we will all see a much improved game with the addition of the new patch.

[Calvincito]

Hey! I thought I'd contribute with some test results, I've undertaken today.

I'm managing Montrose in Scotland, and having advanced to division 1 in merely two years I'm not exactly favourite to win any of the matches. The game I've chosen to run the tests on is: Montrose - Partick Thistle.

The pre-match odds are: 5,50 - 3,60 - 1,50

I ran all the games with the exact same tactics and formation. Gave the exact same team talks, even in half time no matter the score.

I used my own tactics which has worked fine for me so far, having me situated mid-table.

Game 1:

Final score 1-3 (ht 0-0)

Shots: 7-13

On target: 2-5

CCC: 1-1

Pos: 48-52

Game 2:

Final score 1-1 (ht 0-1)

Shots: 9-12

On target: 3-1

CCC: 2-5

Pos: 45-55

Game 3:

Final score 0-1 (ht 0-1)

Shots: 7-9

On target: 1-6

CCC: 2-2

Pos: 46-54

Game 4:

Final score 1-4 (ht 0-2)

Shots: 12-5

On target: 5-4

CCC: 5-3

Pos: 54-46

Game 5:

Final score 1-2 (ht 0-1)

Shots: 5-11

On target: 3-4

CCC: 0-3

Pos: 48-52

Analysis: The results and stats are not differing wildly. Quite the opposite I'd say. The only game where I had posession and an opportunity to win was in the fourth game. This was also the game in which the opposition had the biggest lead at half time. Coincidental?

I have also taken this test further, playing the same game 5 times with a crucial difference. But before I give you those numbers, I'll let you reflect about this first string of games for a while.

[V50]

5 games is not enough for a statistical analysis, not even close. The bare minimum you could have is 20 but that would still be unreliable, since results with a random element to them will converge towards equalling out as you increase the number of observations (matches). I would imagine SI has a match engine tester that will do a thousand matches in a minute to check whether matches are within acceptable limits in terms of luck. If not it might explain half the problems with FM 2008

[Wayne\'o]

i think they need to change the way tactics work and scores lines are determined

[nymanr]

Made a quick test though i don't know what it is supposed to prove as you don't make changes in game.

Man Utd Vs. Sunderland (attendance 98100-99200) Moral: Superb All players in United____Fitness starting 11. 100% United

odds. (4-7) , 5-2 , 4-1

1. 2 - 1

Shots 14 - 9

On tar. 6 - 3

Off tar. 8 - 6

CCC 4 - 2

Pos. 51 - 49

2. 1 - 0

Shots 13 - 2

On tar. 8 - 0

Off tar. 5 - 2

CCC 4 - 1

Pos. 47 - 53

3. 2 - 0

Shots 11 - 5

On tar. 3 - 2

Off tar. 8 - 3

CCC 1 - 0

Pos. 48 - 52

4. 1 - 0

Shots 16 - 2

On tar. 5 - 0

Off tar. 11 - 2

CCC 6 - 0

Pos. 53 - 47

5. 3 - 0

Shots 19 - 2

On tar. 7 - 1

Off tar. 12 - 1

CCC 3 - 1

Pos. 56 - 44

6. 2 - 0

Shots 21 - 3

On tar. 9 - 1

Off tar. 12 - 2

CCC 4 - 0

Pos. 52 - 48

7. 2 - 0

Shots 17 - 8

On tar. 8 - 2

Off tar. 9 - 6

CCC 4 - 0

Pos. 57 - 43

8. 1 - 0

Shots 11 - 3

On tar. 6 - 1

Off tar. 5 - 2

CCC 3 - 0

Pos. 46 - 54

9. 2 - 0

Shots 14 - 6

On tar. 9 - 2

Off tar. 5 - 4

CCC 7 - 0

Pos. 56 - 44

10. 4 - 1

Shots 23 - 6

On tar. 13 - 4

Off tar. 10 - 2

CCC 5 - 2

Pos. 47 - 53

Only last game was approved by fans. All other where disappointments in confidence.

[wwfan]

Despite the '15 games isn't really enough for an analysis' claim that is bound to follow, I think a test that produces 14 consistent results out of 15 plays has considerable statistical weight. If the ME was overly random, I would expect to see, at most, a slight bias (perhaps 7-4-4). 14-1-0 suggests that the ME is extremely fair and the best team, if playing with a solid and logical tactic (which is completely different story), will invariably win.

[knap]

10 games with a tactic I'm testing

[knap]

Arsenal Chelsea game stats 1st time with tactic, part of 10 games above

[knap]

Same tactic same game replayed

[Smac]

This thread's aim is like trying to build a space station with a couple of chimps and a protractor. It ain't nevva gonna happen (not calling any of us chimps, just our methodology).

I'll state this point one more time. Apologies if this is somehow irrelevant or 'too obvious', but it seems to be entirely overlooked:

The amount of variation should itself vary depending on inputs.

Depending on the host of possible inputs. Some pairings very likely will produce wild results. Some will very likely produce extremely consistent results. Because we have no hypothesis and not even any criteria to suggest what kind of inputs should produce either the wild or consistent type results we're inevitably going to through up data set after data set 'proving' each case, which says exactly nothing.

wild

4-0

0-2

1-1

4-4

0-1

8-2

1-7

etc..

consistent

1-1

2-1

1-2

2-2

0-0

0-0

1-0

etc..

or

5-1

10-0

9-0

3-0

0-0

7-1

1-0

5-3

7-2

Additionally, even a wild set of results can be consistently wild, which would could again either tell us something or nothing depending on what we're testing. You'd expect two teams playing throttles-open football to produce consistently wild results if all else is fairly equal. But you might expect and predict a very lopsided set of results with the occasional anomaly or outlier under different circumstances. In theory, nearly every set of results you could achieve with a set of 20 replays should be possible to either:

A. Predict via inputs or

B. Ascribe to 'randomness'

depending on your hypothesis and methods. Again demonstrating that such a test really should tell us nothing at all unless we are quite clever about our criteria.

[Daley]

Can I just add that the problem with trying to understand how random the ME is when you are testing using an AI opponent is that you have absolutely no idea how the AI is changing it's tactics during the game and as we all know the tweaks and changes that we carry out in-game can have a major effect on the outcome of the game. Therefore (and especially with a top four side with a good manager) the results in this test are purely influenced by the AI and nothing else on a match to match basis. I think if you had both teams as human controlled then you could set the teams up exactly the same for each game & then just let the ME lpay it out without making any changes what so ever unless forced to. This would give you the clearest indication of how random the ME is. Just my opinion though.

[chopper99]

Smac - you're explanation is perfectly valid in the real world where anything can and does happen. I suppose the point of this test is to try and discover what has more importance in this, a game. Is it factors input by the user, or is it 'luck' or the so called 'Butterfly Effect'?

In my opinion, because this is a game (simulation or not) then the most important factors contributing to the outcome of a match should be those that the user has some degree of control over. Things such as squad (and therefore player attributes), moral, tactics, media interaction etc. So although a random element is needed, for the majority of the time it shouldn't be strong enough to override the things that the user has control over. Otherwise what's the point in getting a good squad together and working on your tactics?

I personally don't think the random element is too strong, but others have stated in other threads that they do. To me the only way to test this using the tools we have is to set everything before a match, save the game, and then replay that game over and over as many times as time allows. To eliminate the chance of the AI affecting the outcome of the game with their changes future tests should be done with two human controlled teams that both use the same tactics. In that situation the strongest team with the better moral should win the majority of the games and the results should have some sort of pattern to them. If that's not the case then people claiming that the game is far too random may have a point, if not then hopefully it will put the argument to bed.

I personally can't think of a better way to test this to try and prove that you will not get consistantly wide ranging results using the method described above. If you did though, would you not agree that it may highlight an issue?

[HUNT3R]

Chopper, I like this experiment. I have a question though. The same match, players, morale, team talks etc should play out quite similar. I agree with that. I agree with the 'butterfly effect'. However at half time, you're still influencing the game result by giving team talks. Even if it is the same talk, the half-time score won't always be. So at half-time you're still affecting morale and/or motivation.

If you do enough games though, we should see the games with HT scores, all mostly finish with similar FT results. So it might be easier to split your findings up into 3 groups according to HT score (leading, drawing, trailing).

That said though, the FT results should still all be fairly close except for one or two 'freak' matches. What do you think?

[chopper99]

In the next test pre-match and half time team talks will not be given at all to reduce any variables as much as possible.

I agree that the final scores should be fairly similar, with only a few deviations.

[nymanr]

Now i'm just super confused. As i still wasn't getting any pattern/consistant paly from my team i scrapped the tactic i had and just set everything on "global". (Men, Cre, Pas, CloDow, Tac)

Settings where about the same as i had before except i was much looser now in defence. The players that where effected most by me switching to global where my FB:s.

But now all of a sudden my team was performing they way that i had wanted to all along.

Very fast movment with a very nice range of mixed passing and palyers using their creative freedom in a great way.

Suddenly i was scoring over 3 goals a game and some of them where moments of true brilliance...

What is the difference on the striker when i tell him individually to have CF 15 and Team CF 15?

This just ads more to my confusion. Is it my tactics and what is wrong with them if they perform so differently on individual & global as the settings for the players where about the same in both cases.

Or is it all tied down to creative freedom. This was the setting that changed the most for many players.

Is there something wrong with the individual settings? Strange how the difference Was this big.

[Ljuba82]

I have to say, I think this is a pointless exercise. As I said - look at the league tables at the end of the season. They reflect the engine in a way that can be compared to real life.

I was always wandering is there some trigger in game that direct your team to perform average in total (average for Stoke city is 19. place on table, while average for Chelsea is 2nd).

...and after this version of FM I realised that in ME there is some kind of trigger. If you are 3:0 up against Arsenal (playing with Man UTD) for example in first half...chances are 90% that Arsenal will pull back one goal or even more AT 2nd half.

The only way I could explain that is via new, motivation menu from ME screen. When you are 3:0 up...all opponent players are "fired-up".

...so it's not problem about league tables at the end. Overall performance is relatively accurate, the problem is that SINGLE games are not realistic. I can't stand the fact that sometimes my team is playing like Brazil, while several min later they are playing like non-league side.

In purpose of making realistic league tables at end of season, realistic number of scored goals and so on, Im very suspicious that there is some kind of trigger that directs your team to perform average comparing to their quality...

So let's say that I'm using best tactics (and to perfectly and immediately alter it to counteract AI tactics), with best 11 players, with best morale, fitness, team talks...

...and let's say that in first several games I'm overperforming by 20,30 or 40% (for example I have 30 pts from 15 league games, while expectations were to have only 20 pts).

I'm pretty sure that if I continue with same best setup (tactics, morale, teamtalk...), my percentage of overperforming will drop significantly for the rest of season.

[This field intentionally l]

I was always wandering is there some trigger in game that direct your team to perform average in total (average for Stoke city is 19. place on table, while average for Chelsea is 2nd).

...and after this version of FM I realised that in ME there is some kind of trigger. If you are 3:0 up against Arsenal (playing with Man UTD) for example in first half...chances are 90% that Arsenal will pull back one goal or even more AT 2nd half.

The only way I could explain that is via new, motivation menu from ME screen. When you are 3:0 up...all opponent players are "fired-up".

...so it's not problem about league tables at the end. Overall performance is relatively accurate, the problem is that SINGLE games are not realistic. I can't stand the fact that sometimes my team is playing like Brazil, while several min later they are playing like non-league side.

In purpose of making realistic league tables at end of season, realistic number of scored goals and so on, Im very suspicious that there is some kind of trigger that directs your team to perform average comparing to their quality...

So let's say that I'm using best tactics (and to perfectly and immediately alter it to counteract AI tactics), with best 11 players, with best morale, fitness, team talks...

...and let's say that in first several games I'm overperforming by 20,30 or 40% (for example I have 30 pts from 15 league games, while expectations were to have only 20 pts).

I'm pretty sure that if I continue with same best setup (tactics, morale, teamtalk...), my percentage of overperforming will drop significantly for the rest of season.

I have the *exact* same suspicions as you. The game is strangely un-random at times. Having tried to "cheat" and replay a Liverpool vs Aston Villa game until I won; I gave up after about 12 attempts. 11/12 were 0-0 or 1:1 draws. The only game I won was when AV had a man sent off after 3 minutes. When I went ahead some of the equalizers were laughable; my GK passing the ball 5 yards to an opposition striker and then standing in admiration while he scored. Another was the GK heading the ball into his own net. That was a doozy.

I'm not saying the game cheats; I'm saying some unobservable input to the ME sometimes has an egregious effect on the result.

[Smac]

(to the two posts above)

Call me a naive fanboy if you want, but I don't think there's any un-random-ifier to balance the league tables. The strong teams are chock full of better everything, and tend to win because of it. Reloading a game reloads the same morale, squad harmony, player conditions, and other factors that sometimes are quite predictive of a result. Doesn't mean the result is pre-calculated though.

That said, I have seen similar injuries when reloading before internationals...

[Rupal]

PaulC has made much of the fact that the final league tables are reflective of reality and that this demonstrates that things are working properly.

This doesn't altogether satisfy those of us who want good gameplay at the individual match level. We want to feel that our input has (for good or ill) the greatest effect on the result and pattern of play. We want to stand or fall by our own competence or incompetence. That's good gameplay for us.

If we believe that results are 'random' in the sense that 'outside' factors ('butterfly effect' et al ) outweigh what we are doing tactically, or if we feel that these factors mean that no matter what different tactics we try the result will remain the same, then this is not good gameplay for us.

Obviously, a certain element of randomness must be present, otherwise upsets could never occur in the game. But this factor shouldn't be allowed to overwhelm the player's input as frequently as some of us feel that it does. It's a hard balancing act for SI but the concerns which have been expressed need to be looked at.