Decisions is a broken attribute and appears to have a negative effect on team performance

[Flohrinho]

In essence what the guys found out is the attribute weighting. The attributes with the highest cost are at the bottom, those with the lowest are on top. Not what he intended to find out but still interesting I would say.

[fmnatic]

2 minutes ago, Cougar2010 said:

Not particularly worrying IMO.

As Ed has stated above altering one attribute results in a knock effect on other attributes making them higher or lower to keep the required CA of the player.

What you can theorise from it is that the weighting for decisions is probably a little too high when compared to the effect it has within the ME.

If i understand the flaw in the test correctly. Bluesoul instead of  constructing a table of the value of each attribute towards team performance inadvertently created  a table of attribute weighting i.e. the ability point costs of attributes.

referring to the first table here: (Ergo bravery, heading, workrate have lower weighting while positioning, agility, decisions have a higher weighting.)

https://strikerless.com/2017/12/16/fm18-labs-we-need-to-talk-about-decisions/

[themadsheep2001]

Just now, Flohrinho said:

In essence what the guys found out is the attribute weighting. The attributes with the highest cost are at the bottom, those with the lowest are on top. Not what he intended to find out but still interesting I would say.

In any case, he's brought something to Si's attention, which is always good.

[Robioto]

4 minutes ago, Cougar2010 said:

Not particularly worrying IMO.

As Ed has stated above altering one attribute results in a knock effect on other attributes making them higher or lower to keep the required CA of the player.

What you can theorise from it is that the weighting for decisions is probably a little too high when compared to the effect it has within the ME.

I agree when it comes to the first set of results, but this 2nd set is slightly worrying: https://strikerless.com/2017/12/18/fm18-labs-the-final-decisions-results/

I'm not too bothered myself, I'll just keep playing the game as normal and monitor this thread out of interest more than anything to see if it gets proven/acknowledged either way.

I know more tests are needed, but it's not looking good with the evidence so far.

16 minutes ago, wicksyFM said:

I think i will put starting a save on hold until this potential issue is sorted

This is a bit of an over reaction!

[wicksyFM]

2 minutes ago, themadsheep2001 said:

It's not even proven to be an issue yet...

 

2 minutes ago, KlaaZ said:

So it's not even sure there is an issue (as stated above by several people, the way of data analyzing is far from perfect) but you'll still stop playing? You do realise that this might take months if not longer, just to conclude that there's nothing wrong?

I know. Silly comment by myself. To be honest im not even sure it is an issue. Just looking through an old save now. The teams and players that do well (win competions/awards) seem to have very good decisions amongst the players in the squads. The highest scorers, pass completion, average rating etc are all topped by players with very good decisions

[fmnatic]

3 minutes ago, Robioto said:

I agree when it comes to the first set of results, but this 2nd set is slightly worrying: https://strikerless.com/2017/12/18/fm18-labs-the-final-decisions-results/

 

@Robioto FYI @John_Yuki confirmed that other attributes were being nerfed in the newcastle tests. 

 

[Robioto]

Just now, fmnatic said:

FYI @John_Yuki confirmed that other attributes were being nerfed in the newcastle tests. 

 

Fair enough.

[themadsheep2001]

1 minute ago, wicksyFM said:

 

I know. Silly comment by myself. To be honest im not even sure it is an issue. Just looking through an old save now. The teams and players that do well (win competions/awards) seem to have very good decisions amongst the players in the squads. The highest scorers, pass completion, average rating etc are all topped by players with very good decisions

This is the thing. When I look back at my save, it mirrors this. I'm not saying there is nothing to look at. But if there is an overall impact in gameplay, it's not really showing through

[santy001]

One big thing is, when you start imposing situations on the game in which the game will never have happen it will start to look a bit weird. 20 work rate, 20 teamwork, 1 stamina players could throw up something interesting. As could 20 anticipation, 20 off the ball, 1 finishing, 1 composure strikers. 

There's definitely something to look at here, but do you advocate fixing the game that in a league where everything bar 1 attribute is fixed, the 1 attribute should then be the defining feature of the league?

Bear in mind that decisions alone determines how good he is at making good decisions with or without the ball. A good decision isn't the right decision or the best decision. It is moving to block a passing lane, or getting tight to your man, playing a short pass to retain possession etc. 

You can lose a game of football without doing anything wrong just by virtue of the opposition playing brilliantly, in an entirely normalised world (purely conjecture on my part here) I believe decisions, or lack thereof, essentially becomes flair. The players with low decisions become those more likely to do the unexpected, they're more likely to be in an unpredictable location. The higher decisions are more likely to make solid logical choices, without any nuance or subtlety to their play in comparison. 

[Cougar2010]

2 minutes ago, fmnatic said:

If i understand the flaw in the test correctly. Bluesoul instead of  constructing a table of the value of each attribute towards team performance inadvertently created  a table of attribute weighting i.e. the ability point costs of attributes.

referring to the first table here: (Ergo bravery, heading, workrate have lower weighting while positioning, agility, decisions have a higher weighting.)

https://strikerless.com/2017/12/16/fm18-labs-we-need-to-talk-about-decisions/

Each attribute has its own weighting/ratio to convert it to/from CA, the weighting also depends on the position the player plays and how natural he is in that position.

In a perfect world that weighting would reflect exactly how important the attribute was within the ME and every attribute would have a 100% balanced weighting.  To me though that is very close to impossible to achieve given the number of attributes, positions, how natural a player is & how all the attributes interact with one another within the ME so its to be expected that cases like this will appear.

 

[EdL]

I've looked at the Newcastle saves and it is the same issue, increasing decisions has pulled down all the other attributes for an example comparison

 

[Bluesoul vs. The Bearodactyl]

3 minutes ago, EdL said:

I've looked at the Newcastle saves and it is the same issue, increasing decisions has pulled down all the other attributes for an example comparison

 

Thanks @EdL, I am going to be revisiting the first experiment (with each attribute being tested in isolation) and have CA = Recommended CA. That should stop the attribute decay, or at least minimize it to whatever rounding error there may be by using an int for such a big part of the calculations.

Also, how are you getting player data out in text format? Can you share? It's nearly Christmas, you know.

Which would be more beneficial, increasing the CA and PA to the RCA, or the PA to such a point that the CA should not meet it in a season?

[EdL]

I also took a look at the match code and I couldn't find any obvious misuse of the decisions attribute, so I think it is really just a case of if you pull down most other attributes the affect of that is going to outweigh the positive of decisions being higher.

[Bluesoul vs. The Bearodactyl]

21 minutes ago, Flohrinho said:

In essence what the guys found out is the attribute weighting. The attributes with the highest cost are at the bottom, those with the lowest are on top. Not what he intended to find out but still interesting I would say.

This actually does match up rather precisely with what I'm seeing. This is good! I didn't know I was going to come away with this with a rather accurate set of attribute weights, but that's a bonus. There's more testing to come with this taken into account.

[phnompenhandy]

2 hours ago, sporadicsmiles said:

Can I make an observation as a scientist used to performing experiments where you try to isolate a single variable from a large pool of variables?

The problem is that both in the real world and in FM, it is often not possible to change a single thing in isolation from all other variables. In this case, the creator of the article has done a good job (and put a lot of effort) into changing systematically a single attribute. However, this methodology says nothing about how the attributes interact with one another. That is true of any experiment like this, and it is also true in the real world. I work in science, and we have to put a whole load of effort into proving that we are indeed isolating a single variable and measuring its effect only. For example, if I was going to try to show effect X is caused by variable Y by technique Z on system A, I would also run an experiment on system B - which is as closely related to system A as possible - that also contains variable Y, but should not exhibit effect X when probed by technique Z. That way, I can show that effect X is not caused by any of the innumerable other variables shared by A and B, and can conclude safely that it is variable Y causing effect X. This is, of course, still not perfect, but I hope it serves to illustrate a point.

On the experiment here, there is no such "blank" experiment. Such an experiment would have been to make 36 absolutely identical sides, and run the simulation X number of times. What you would want to see there is that the winning percentages are totally random, since there should be no effect if everything is the same. This would be a great baseline, and not so difficult to set up. I may be wrong on the next point, but it would also be very important to run this test many, many, many times, resetting the database each time, in order to be certain your statistics are not a matter of fluke. These two things are utterly vital in order to ensure that your method is actually suitable. As I also tend to err on the side of caution, I would also suggest performing an identical experiment to the one where decisions run from 1 to 20, but with a different value for each of the other attributes. I would in fact run one where they were all set to 1, and one where they are all set to 20. This should tell you how decisions interacts with all the other variables. To be even more cautious, I would run all of these experiments (including the original one you did on decisions) on Bravery, which caused the most wins. If you expect that bravery follows a linear scale (1 = very important, 20 = very important), then you should see a clear correlation between winning and bravery. After that, I think I would be convinced by any interpretation.

Just to be clear, I am not doubting the validity of the results, because results are what they are, and the author of that piece has done some sterling work. I do not, however, think we have enough data to make any firm conclusions from what has been found. It is an interesting effect, but much more work would be required to understand. If I had the time and the computer this Christmas, I might well set up these experiments, but sadly I do not. I would be happy to collaborate on the data analysis with anyone who does want to run them though!

 

Wonderful post in a superb thread. I like to fondly imagine this is exactly how SI conduct their mysterious 'soak tests' ;-)

 

It's quite plausible that the final verdict is that the test merely exposed the relative weightings of all the attributes; if that turns out to be the case and that DECISIONS is the most heavily weighted attribute, I'd say that that's perfectly valid. For me, DECISIONS is certainly the most important attribute for any player in any position - and I say that wistfully as one of those managers who only bring through academy kids and so never see high decisions!

[EdL]

Making sure the CA / RCA is balanced is probably a better test but you are still changing his CA which could have other affects in the game so its still not really accurate (and with the way the game works there isn't really a way to control it as you would need)

The best control might be making CA = PA so things remain consistent in the season (although can still decrease!)

You can get the text output by clicking print on a player profile in the FM menu.

[Bluesoul vs. The Bearodactyl]

Just now, EdL said:

Making sure the CA / RCA is balanced is probably a better test but you are still changing his CA which could have other affects in the game so its still not really accurate (and with the way the game works there isn't really a way to control it as you would need)

The best control might be making CA = PA so things remain consistent in the season (although can still decrease!)

You can get the text output by clicking print on a player profile in the FM menu.

Yes it's interesting, if I set the Bravery team's CA to RCA, it's only an increase of 6 points. If I do the same for Decisions, it's 37. I'm not sure if that is what's needed to level the playing field, but I'm going to try and see what happens.

CA is already equal to PA in the first set of tests.

I always thought that Print Screen button was for a screenshot! First time I've clicked it in my life. That's neat!

[Bluesoul vs. The Bearodactyl]

2 minutes ago, phnompenhandy said:

It's quite plausible that the final verdict is that the test merely exposed the relative weightings of all the attributes; if that turns out to be the case and that DECISIONS is the most heavily weighted attribute, I'd say that that's perfectly valid.

If this is the conclusion to be gleaned then I am happy to have done it. I think that's really valuable information as well.

[EdL]

Yep as I said decisions is very highly weighted hence the need for a big bump. Again please make sure you are testing with the league set to full detail - it will take a while to run a season

[Bluesoul vs. The Bearodactyl]

This is the way each of the tests have been run.

[Guest]

As Ed has mentioned that methodology means the test will not be running the full match engine, if you want to test that you need to chance the detail settings to ‘All’.

[edk77]

11 minutes ago, Barside said:

As Ed has mentioned that methodology means the test will not be running the full match engine, if you want to test that you need to chance the detail settings to ‘All’.

The match detail for premier division is set to "All" in his screenshot.

[enigmatic]

41 minutes ago, Bluesoul vs. The Bearodactyl said:

Yes it's interesting, if I set the Bravery team's CA to RCA, it's only an increase of 6 points. If I do the same for Decisions, it's 37. I'm not sure if that is what's needed to level the playing field, but I'm going to try and see what happens.

CA is already equal to PA in the first set of tests.

I always thought that Print Screen button was for a screenshot! First time I've clicked it in my life. That's neat!

If you're rerunning, I'd suggest reducing randomness further by giving players a slight buff in their technique, long shots and passing (to 14 or so) so they can pull off trickier stuff if they decide to do it, and set high composure, concentration, consistency and personality stats to reduce bad decisions which owe little to the player's decision making under normal circumstances, and probably determination and stamina for similar reasons. .

Also, I'd recommend the default manager plays a very passing oriented style, to put more emphasis on smart decision making over big hoofs downfield which often work well in FM but any low intelligence footballer can manage if not instructed not to do it. (Especially since I think the game will default to playing the more defensive variant of formations when playing with 100CA players than 137CA players)

[mkkadi]

51 minutes ago, EdL said:

Yep as I said decisions is very highly weighted hence the need for a big bump. Again please make sure you are testing with the league set to full detail - it will take a while to run a season

I thought it is CA RCA related as a researcher but there are some attributes which have more weight than decisions. The weights are related to the positions and I dont know what happens to the weights if every single positions ranked as 20. If there are attributes weighter than decisions, probably there are, should not they be worse than decisions. 

[Bluesoul vs. The Bearodactyl]

Sincere thanks to everyone that's suggested explanations and ideas to test the results. It appears that the discrepancy between CA and Recommended CA plays a major role in the data.

Further testing is ongoing, but once I got the attached chart, I wanted to go ahead and put it out there. If there were no correlation, the two lines would be constantly crossing each other and at differing distances.

That does not appear to be the case.

I added a note to the Strikerless article this morning, indicating that there are likely other variables at play than Decisions.

If anyone would like to contribute data for this new experiment, the Workshop file is here and the Labs page has been created here and will only ask for your league table in alphabetical order, which you can collect by going to League->Stages and click on the club name heading.

 

The Newcastle with Decisions 20 Experiment is also being re-run with the CA adjusted to recommended CA (And PA set to CA if the new CA is higher than the old PA), if we see a dramatic change for the better then we'll likely say that things are working as intended. If we're still seeing a 65% relegation rate, I'm going to say there is still either a major issue in the Match Engine or the test methodology.

Workshop File

Lab Page

[Fmplaya]

What I have learnt through this emotional experience is that for example if two players had the same CA (150) and player A had 20 decisions and player B had 1 decisions player B would have better attributes (except for decisions) as the Decisions attribute is heavily weighted which takes up more CA leaving the other attributes with lower totals as decisions takes up so much CA.

This could possibly lead to another problem for when developing youth. For example player A and B have the same PA but player A has 20 decisions and B equals 1 decisions, player B would have a lot better attributes overall than player A as decisions is so heavily weighted and it takes up more CA leaving less space space for development of other attributes meaning his attributes in general would be lower than player B. This be very annoying because when you develop players there abilities can't be reached to the full as the decision stat takes up so much CA.

Also if two players had exactly the same in all attributes except from decisions ( Player A=20 B=10) in terms of CA player a would be seen as a much better player because he can make "decisions" better than B. This would then lead into the player being available for a lot more in the transfer market as his CA would attract the AI managed teams when you can sign the exact same player but just is worse at making the "right decision"

To conclude I think the "Decisions" attribute is unclarified as well as broken. Maybe it would be better off making it CA free??? It would be great if anyone at SI could take a look at this

(Apologies for any grammatical mistakes or if any of it doesn't make sense I am only 13😝)

 

[phnompenhandy]

9 hours ago, Fmplaya said:

What I have learnt through this emotional experience is that for example if two players had the same CA (150) and player A had 20 decisions and player B had 1 decisions player B would have better attributes (except for decisions) as the Decisions attribute is heavily weighted which takes up more CA leaving the other attributes with lower totals as decisions takes up so much CA.

This could possibly lead to another problem for when developing youth. For example player A and B have the same PA but player A has 20 decisions and B equals 1 decisions, player B would have a lot better attributes overall than player A as decisions is so heavily weighted and it takes up more CA leaving less space space for development of other attributes meaning his attributes in general would be lower than player B. This be very annoying because when you develop players there abilities can't be reached to the full as the decision stat takes up so much CA.

Also if two players had exactly the same in all attributes except from decisions ( Player A=20 B=10) in terms of CA player a would be seen as a much better player because he can make "decisions" better than B. This would then lead into the player being available for a lot more in the transfer market as his CA would attract the AI managed teams when you can sign the exact same player but just is worse at making the "right decision"

To conclude I think the "Decisions" attribute is unclarified as well as broken. Maybe it would be better off making it CA free??? It would be great if anyone at SI could take a look at this

(Apologies for any grammatical mistakes or if any of it doesn't make sense I am only 13😝)

 

The way I see it, DECISIONS really is the most important attribute. Therefore, a player might have high CA and great physical and technical attributes, but if his Decisions attribute is low, he'll consistently mess up. So, despite all other appearances, he is NOT a great player! In future, when I'm making decisions (!) on which academy kids to keep and nurture, it won't be the Determination attribute I'll be looking at primarily - it'll be Decisions. I don't transfer in players but for the rest of you, set your scouts to focus on that attribute and always make it your high priority.

 

The only actual problem I see with this is that DECISIONS can't be trained. Now previously I'd have argued that it's an inherent sense that can improve with experience but can't be taught. But I've changed my opinion on observing the change is the exemplar of the talented young footballer who constantly made stupid decisions - Raheem Sterling. I think the genius of Pep is to improve his decision-making facility and turn his potential into true greatness. Maybe that's the exception to the rule that merely highlights what a unique manager Pep is, however.

[jam jameson]

9 ore fa, Fmplaya ha scritto:

What I have learnt through this emotional experience is that for example if two players had the same CA (150) and player A had 20 decisions and player B had 1 decisions player B would have better attributes (except for decisions) as the Decisions attribute is heavily weighted which takes up more CA leaving the other attributes with lower totals as decisions takes up so much CA.

This could possibly lead to another problem for when developing youth. For example player A and B have the same PA but player A has 20 decisions and B equals 1 decisions, player B would have a lot better attributes overall than player A as decisions is so heavily weighted and it takes up more CA leaving less space space for development of other attributes meaning his attributes in general would be lower than player B. This be very annoying because when you develop players there abilities can't be reached to the full as the decision stat takes up so much CA.

Also if two players had exactly the same in all attributes except from decisions ( Player A=20 B=10) in terms of CA player a would be seen as a much better player because he can make "decisions" better than B. This would then lead into the player being available for a lot more in the transfer market as his CA would attract the AI managed teams when you can sign the exact same player but just is worse at making the "right decision"

To conclude I think the "Decisions" attribute is unclarified as well as broken. Maybe it would be better off making it CA free??? It would be great if anyone at SI could take a look at this

(Apologies for any grammatical mistakes or if any of it doesn't make sense I am only 13😝)

 

This is an interesting view at the situation, and I have to say that makes sense with all said before, one of the most intriguing and interesting FM discussions ever...so if the outcome is that Decisions is outweighted, the reasonable consequences both on market values, young players and development are the above. Anyway I do not see this as a flaw, it’s a choice (reasonable) of SI, it’s important for the users being aware of how it runs, but i’d No complaints at all should this be the explanation. On the other end, should Decisions ‘play against’ as it seemed to suggest the experiment at the top of the thread, it would be absolutely game breaking

[jam jameson]

15 ore fa, EdL ha scritto:

I also took a look at the match code and I couldn't find any obvious misuse of the decisions attribute, so I think it is really just a case of if you pull down most other attributes the affect of that is going to outweigh the positive of decisions being higher.

This is in my opinion one of the most important answer by SI, because they only have the inner view of the code, and where any experiment may just provide evidence for deductions, they can check directly the source of it. Hence the reasonable conclusion that Decisions is overweighted and it’s the most important attribute on the pitch, even if it seems not relevant off the pitch, where other attributes sound much more relevant in helping a player fulfil it’s PA (work rate, determination, professional personality)

[Bootador]

 

8 hours ago, phnompenhandy said:

The way I see it, DECISIONS really is the most important attribute. Therefore, a player might have high CA and great physical and technical attributes, but if his Decisions attribute is low, he'll consistently mess up. So, despite all other appearances, he is NOT a great player! In future, when I'm making decisions (!) on which academy kids to keep and nurture, it won't be the Determination attribute I'll be looking at primarily - it'll be Decisions. I don't transfer in players but for the rest of you, set your scouts to focus on that attribute and always make it your high priority.

 

The only actual problem I see with this is that DECISIONS can't be trained. Now previously I'd have argued that it's an inherent sense that can improve with experience but can't be taught. But I've changed my opinion on observing the change is the exemplar of the talented young footballer who constantly made stupid decisions - Raheem Sterling. I think the genius of Pep is to improve his decision-making facility and turn his potential into true greatness. Maybe that's the exception to the rule that merely highlights what a unique manager Pep is, however.

Can't say I agree with that, if it's proved anything this test it's that decisions is an undesirable attribute compared to every other one desired by a role because of how much CA weight it has.

[phnompenhandy]

1 hour ago, Bootador said:

 

Can't say I agree with that, if it's proved anything this test it's that decisions is an undesirable attribute compared to every other one desired by a role because of how much CA weight it has.

I'm not seeing the logic here. By the same token, BRAVERY would be the most desirable attribute in your book - is that reasonable? It sounds like you're thinking about the game in terms of algorithms and how to 'game' the system instead of how footballers behave, which is what SI are striving to emulate.

[herne79]

10 hours ago, phnompenhandy said:

The way I see it, DECISIONS really is the most important attribute. Therefore, a player might have high CA and great physical and technical attributes, but if his Decisions attribute is low, he'll consistently mess up. So, despite all other appearances, he is NOT a great player! In future, when I'm making decisions (!) on which academy kids to keep and nurture, it won't be the Determination attribute I'll be looking at primarily - it'll be Decisions. I don't transfer in players but for the rest of you, set your scouts to focus on that attribute and always make it your high priority.

Dangerous to look at one attribute in isolation from all others.  If you go down that road then how do you feel about having an academy full of players with high Decisions but low Determination?  They might make nice decisions but they won't be bothered about carrying them out.  Or train very well.  Or a myriad of other things that Determination affects.

10 hours ago, phnompenhandy said:

The only actual problem I see with this is that DECISIONS can't be trained

Try and find training a role which does not have Decisions as part of it's attribute development.  There aren't very many.

1 hour ago, Bootador said:

Can't say I agree with that, if it's proved anything this test it's that decisions is an undesirable attribute compared to every other one desired by a role because of how much CA weight it has.

and @phnompenhandy, nothing's actually changed.  Decisions is (and always has been) a desirable attribute because of the impact it has across all manner of actions.  Just as Determination is.  Or Tackling for a defender, or Off the Ball for an attacker, or being able to cope with pressure, or behaving professionally.  And so on.  It helps produce balanced players who will not only be able to tackle, shoot or pass but also be able to decide when to tackle, shoot or pass.

[phnompenhandy]

Hey Herne, don't misattribute to me! I very much agree with that last point; it's what I've been trying to point out.

And I agree with your points above that - it is a question of balance and of course high DECISIONS in a lazy youth is wasted potential. The reason I say DECISIONS is correctly heavily weighted and the most important attribute is for the very reasons toy give - it is inherent in the success of every other attribute, probably more so than any of the others. A winger lacking bravery can be a star,, as can a striker with low workrate. But find me a great footballer who persistently makes poor decisions.

[Fmplaya]

1 hour ago, phnompenhandy said:

Hey Herne, don't misattribute to me! I very much agree with that last point; it's what I've been trying to point out.

And I agree with your points above that - it is a question of balance and of course high DECISIONS in a lazy youth is wasted potential. The reason I say DECISIONS is correctly heavily weighted and the most important attribute is for the very reasons toy give - it is inherent in the success of every other attribute, probably more so than any of the others. A winger lacking bravery can be a star,, as can a striker with low workrate. But find me a great footballer who persistently makes poor decisions.

Decisions becomes wasted when the player has no ability

[Rashidi]

I get the impression that some people seem to think that attributes have some value in an absolute sense, that one attribute is more important than others. Every attribute in the game plays a part at some point, and at that point some attributes may be more important than others. For example, when you are defending, the ball could be with the opposition team in their half and near their defender. Does that make anticipation more important for the player closest to the ball or furthest? Or is it equally important for everyone? Distance should play a part right? It does, but how does that get factored into possible outcomes?

Attributes are relative, and as such, some of these attributes could get weighted differently at certain times, and without knowing how the match code weights these, we need to be careful drawing conclusions based on absolute comparisons. So while experiments can be useful, we should be aware of how relative things can get in the game. 

[Vernum]

4 hours ago, herne79 said:

and @phnompenhandy, nothing's actually changed.  Decisions is (and always has been) a desirable attribute because of the impact it has across all manner of actions.  Just as Determination is.  Or Tackling for a defender, or Off the Ball for an attacker, or being able to cope with pressure, or behaving professionally.  And so on.  It helps produce balanced players who will not only be able to tackle, shoot or pass but also be able to decide when to tackle, shoot or pass.

I m not sure i understand how it is a desirable attribute when setting this to one and having other random attribute with less weight, higher gives better result as he has shown in his experiments. Maybe it would be desirable if it had much less weight.

[Svenc]

On 18.12.2017 at 14:31, themadsheep2001 said:

This is the thing. When I look back at my save, it mirrors this. I'm not saying there is nothing to look at. But if there is an overall impact in gameplay, it's not really showing through

That's the only way to draw any conclusions. Without adjusting the CA, all those points left will be reallocated elsewhere. F'r instance, nerfing a forward's finishing and composure will free a good amount of points the game then reallocates elsewhere, and that impacts massively (in particular if this reallocation ensures the forward is then consistently in decent position, perhaps by a significant boost iin dribbling, pace and movement related attributes).

I'd keep an eye on this thread and experiment as it may in parts defy another long-term myth. The game one of being "morale manager". It's been long acknowledged by the coders that any such "man management" and morale stuff mainly rubs off on decision making attributes internally. You could therefore word it perhaps a bit differntly too: Even in the worst possible "mental" conditions, contrary to popular FM myth, teams are not destined to lose (which some knew anyhow). Naturally these are all AI managed teams, which means they are given simple but logical means to proactively manage matches, rather than click continue.
 

[herne79]

2 hours ago, Vernum said:

I m not sure i understand how it is a desirable attribute when setting this to one and having other random attribute with less weight, higher gives better result as he has shown in his experiments. Maybe it would be desirable if it had much less weight.

Scroll to the very top of the page and read what SI have written about these experiments .

[hoppo1982]

On 12/19/2017 at 05:34, phnompenhandy said:

The way I see it, DECISIONS really is the most important attribute. Therefore, a player might have high CA and great physical and technical attributes, but if his Decisions attribute is low, he'll consistently mess up. So, despite all other appearances, he is NOT a great player! In future, when I'm making decisions (!) on which academy kids to keep and nurture, it won't be the Determination attribute I'll be looking at primarily - it'll be Decisions. I don't transfer in players but for the rest of you, set your scouts to focus on that attribute and always make it your high priority.

 

The only actual problem I see with this is that DECISIONS can't be trained. Now previously I'd have argued that it's an inherent sense that can improve with experience but can't be taught. But I've changed my opinion on observing the change is the exemplar of the talented young footballer who constantly made stupid decisions - Raheem Sterling. I think the genius of Pep is to improve his decision-making facility and turn his potential into true greatness. Maybe that's the exception to the rule that merely highlights what a unique manager Pep is, however.

I am not sure the change in Sterling is down to Pep improving his decision making tbh. Id put it down to Pep giving him a clearly defined role and removing a lot of the decisions the player has to make. Pep's methods are very structured to the point where players must remember what to do when in receipt of the ball in different areas of the pitch. This has removed a lot of the thinking for a player like Sterling. 

 

Training decision making is a difficult one to quantify though.

 

Mute point given almost all of us are in agreement that focusing on one specific attribute is foolish. 

[BlueAnderson]

3 hours ago, hoppo1982 said:

I am not sure the change in Sterling is down to Pep improving his decision making tbh. Id put it down to Pep giving him a clearly defined role and removing a lot of the decisions the player has to make. Pep's methods are very structured to the point where players must remember what to do when in receipt of the ball in different areas of the pitch. This has removed a lot of the thinking for a player like Sterling. 

 

Training decision making is a difficult one to quantify though.

 

Mute point given almost all of us are in agreement that focusing on one specific attribute is foolish. 

What role has Pep given Sterling for him to make less decisions or thinking? This is rubbish and really undermines Sterling’s ability as a player. 

[ktime156]

5 hours ago, BlueAnderson said:

What role has Pep given Sterling for him to make less decisions or thinking? This is rubbish and really undermines Sterling’s ability as a player. 

 I think that what @hoppo1982 was saying is that Pep drills his players on offense in the same way that many managers train players on defense. A lot of managers still allow high levels of freedom and give basic instructions on how they hope to break down a defense. They drill their players on defending. Pep trains his players very hard on where to be in relation to the ball and positioning of their teammates. In a sense, it can make certain actions robotic after a period of time. Much like learning to play piano. There's still skill levels to it and the player must still actively think and concentrate, but many of the hand movements would become so well trained that they almost become natural action. So in sense, the player is now focusing on fewer things than in the past allowing for more proficient play. 

[ktime156]

DECISIONS as defined by SI: The player's ability to make the correct choice both with and without the ball. 

In addition to the discussions up above, I'm also interested in how a higher Decisions rating may help define the amount of risky passes attempted within the reality of the ME. 

In the data within the experiment, the higher the decision rating equaled the lower the pass completion percentage and vice-versa. Other users have commented that in their saves, across a season, players with a higher Decisions rating had more key passes across a season. In my mind, this would make sense as attempting risky passes at a higher rate would both lead to the potential for a significantly lower pass completion percentage (if the entire team had high Decisions) and more key passes when those passes were successfully completed in the final third. I know that there are other variables at play but if this were agreeable, it could be informative on how some users design their tactical system going forward... For example, users wishing to construct a counter or overload/switch to the other flank style tactic would know to perhaps target midfielders with higher Decisions as a rating while placing players with lower Decisions on the overload side to maintain possession and make an eventual safe pass back to the midfielder. In turn, the midfielder would make more attempts at a risky pass to the desired flank. Same with a possession tactic. Depending on which area(s) they would like to attack from, they could know that placing the player(s) with a higher Decisions rating could increase risky passes from those positions. Or players with a lower Decisions rating would more likely maintain possession. 

Of course the question here is why a higher Decisions rating would possibly equal riskier decisions instead of better decisions. Does any of that make any sense? 

[Mons]

2 hours ago, ktime156 said:

DECISIONS as defined by SI: The player's ability to make the correct choice both with and without the ball. 

In addition to the discussions up above, I'm also interested in how a higher Decisions rating may help define the amount of risky passes attempted within the reality of the ME. 

In the data within the experiment, the higher the decision rating equaled the lower the pass completion percentage and vice-versa. Other users have commented that in their saves, across a season, players with a higher Decisions rating had more key passes across a season. In my mind, this would make sense as attempting risky passes at a higher rate would both lead to the potential for a significantly lower pass completion percentage (if the entire team had high Decisions) and more key passes when those passes were successfully completed in the final third. I know that there are other variables at play but if this were agreeable, it could be informative on how some users design their tactical system going forward... For example, users wishing to construct a counter or overload/switch to the other flank style tactic would know to perhaps target midfielders with higher Decisions as a rating while placing players with lower Decisions on the overload side to maintain possession and make an eventual safe pass back to the midfielder. In turn, the midfielder would make more attempts at a risky pass to the desired flank. Same with a possession tactic. Depending on which area(s) they would like to attack from, they could know that placing the player(s) with a higher Decisions rating could increase risky passes from those positions. Or players with a lower Decisions rating would more likely maintain possession. 

Of course the question here is why a higher Decisions rating would possibly equal riskier decisions instead of better decisions. Does any of that make any sense? 

It does, but only partially as you're conflating decisions with technical ability. A player with good decisions is capable of knowing his limits and not overstretching himself. A willingness to try risky passes is more determined by PPMs and tactical instructions rather than the decisions attribute.

 

Actually, I would argue that a player with poor decisions would try more risky passes with relatively little success, while a player with a good decisions attribute would be more likely to maintain possession unless he was in a good enough situation to try a risky pass depending on his awareness of his passing and creativity attributes. 

[ktime156]

@Mons Thanks for the reply! That's how I've understood Decisions to be coded in the past. However, with the experimentation taking place... with their being a near linear degression in that a team with the lowest Decisions had the highest pass completion percentage while the team with the highest Decisions had the lowest... and with other users reporting that in their saves that players with the highest Decisions ratings in their leagues did indeed have the most key passes over a season, I wondered if that pointed to anything that may actually be useful for users in the here and now. Of course there are other variables at play and to your point, I'd be interested now in seeing what tactical system the AI chooses for each team in the base 36 team experiment as it may give slight insight a small part of the AI programming. Perhaps by looking at how the AI chooses different tactics for different highly rated skills, it may give insight into how the ME has been coded to view said skills? That's the rabbit hole that my mind goes down anyhow. 

[Mons]

47 minutes ago, ktime156 said:

@Mons Thanks for the reply! That's how I've understood Decisions to be coded in the past. However, with the experimentation taking place... with their being a near linear degression in that a team with the lowest Decisions had the highest pass completion percentage while the team with the highest Decisions had the lowest... and with other users reporting that in their saves that players with the highest Decisions ratings in their leagues did indeed have the most key passes over a season, I wondered if that pointed to anything that may actually be useful for users in the here and now. Of course there are other variables at play and to your point, I'd be interested now in seeing what tactical system the AI chooses for each team in the base 36 team experiment as it may give slight insight a small part of the AI programming. Perhaps by looking at how the AI chooses different tactics for different highly rated skills, it may give insight into how the ME has been coded to view said skills? That's the rabbit hole that my mind goes down anyhow. 

Sorry mate, but you've completely lost me there...

[Rhobium]

I feel like there is a lot of missing the point here. The tests suggest that

• According to the CA attribute weighting, decisions is very important.
• But according to the match engine, decisions is less important.

This is why given many teams of equal player CA, the team with higher decisions performs worse: because the CA attribute weighting does not correspond to importance of attributes in the match engine.

Given that the AI and the player's advice relies on CA to judge a player's ability, and not individual attributes, this definitely needs improving.

[mrbrownsays]

1 hour ago, Rhobium said:

I feel like there is a lot of missing the point here. The tests suggest that

• According to the CA attribute weighting, decisions is very important.
• But according to the match engine, decisions is less important.

This is why given many teams of equal player CA, the team with higher decisions performs worse: because the CA attribute weighting does not correspond to importance of attributes in the match engine.

Given that the AI and the player's advice relies on CA to judge a player's ability, and not individual attributes, this definitely needs improving.

This may be a stupid question to ask, but are the attributes on a linear scale. ie is The difference between 5-10 in decisions the same as the difference in 10-15.

Within a given CA would it have the same impact on other attributes.

Does the match engine treat it as the same gap.

Does my question even make sense?

 

Is there an FAQ/readme on how the underneath of the game actually works as I haven't thought about it much before now but am finding it fascinating.

[Mr U Rosler]

2 hours ago, Rhobium said:

I feel like there is a lot of missing the point here. The tests suggest that

• According to the CA attribute weighting, decisions is very important.
• But according to the match engine, decisions is less important.

This is why given many teams of equal player CA, the team with higher decisions performs worse: because the CA attribute weighting does not correspond to importance of attributes in the match engine.

Given that the AI and the player's advice relies on CA to judge a player's ability, and not individual attributes, this definitely needs improving.

Correct.

Or in other words, if you were building the most effective player you could from scratch and had 150 'ability points' to spend, the implication is that you'd be better off not spending a lot on the decisions attribute due to the 'weighting' and instead spend those 'points' elsewhere.  

[PJHoutman88]

4 hours ago, Rhobium said:

I feel like there is a lot of missing the point here. The tests suggest that

• According to the CA attribute weighting, decisions is very important.
• But according to the match engine, decisions is less important.

This is why given many teams of equal player CA, the team with higher decisions performs worse: because the CA attribute weighting does not correspond to importance of attributes in the match engine.

Given that the AI and the player's advice relies on CA to judge a player's ability, and not individual attributes, this definitely needs improving.

Decisions informs all other attributes. Decision-making alone won't win you a game. If all other attributes are declining because of CA weighting, then having good decisions alone shouldn't compensate for those declines in my opinion..

[Vernum]

On 12/19/2017 at 23:06, herne79 said:

Scroll to the very top of the page and read what SI have written about these experiments .

I did read the first post and sig said that decision is heavily weighted like i mentioned in my post and asked to run the league in full details, he posted a screenshot of his second batch of test using full details and found similar result , decision seems to weight too heavily to be viable compared to other stats since raising it, bring down too many stats without giving enough benefit and end up with poor team performance for the 15 + decision team and best performance for the lowest decision team.