Effect of Ambition, Professionalism, Determination and Workrate on CA growth

The purpose of this experiment was to test the impact of Determination, Ambition, Professionalism and Workrate on CA growth. In the early years of the experiment, most of the CA growth would be due to "natural progression" and training, as players would get precious little first-team football. What inspired me to do this experiment are the experiments of others, and the conflicting conclusions particularly about the Determination and Workrate attributes and their impact on player growth.

I remember reading the results of a test for an older FM (2012 or maybe even 2010) which showed that players with higher Determination DID NOT grow their CA any faster, but that players with Workrate did, albeit only slightly. However, I've seen also a more recent experiment, using 2013 as a base, where Determination was shown to HAVE an effect on player growth, but Workrate was almost counter-productive. This discrepancy seemed to make no sense, so I decided to test it myself.

The Players:

There are five: Control Group, High Determination, High Professionalism, High Ambition and High Workrate.

Their birthdates are the same, their starting age was 14, their CA was 100, and PA 200. Each had 20 points in the M© position. I wanted to keep the position the same for all players, and M© seemed like the most versatile one, and the one that required the most versatile attributes (e.g., a DC with 10 for jumping is a bigger liability than an M© with 10 for jumping). All starting attributes (visible and invisible) were set to 10, with the following exceptions:

- the High <attribute> players had that attribute set to 20

- ALL players had Natural Fitness set to 20, and Injury Proneness set to 1. This was done to minimize the injuries to these players, and thus minimize an extraneous factors that could affect CA growth

Other setup notes:

- the team selected was Barcelona, which has 9,500 reputation and high-quality youth facilities. I wanted both high-rep team and good youth facilities, because these all contribute to CA growth from "natural progression" and training and I wanted to see good CA growth during the early years of the experiment, when first-team playing time would be limited. This is because first-team playing time is basically another extraneous factor, like injuries, that can throw off this type of experiment and that is hard to control for unless I were to actually make sure these youths don't play games. Additionally, Barcelona has a history of nurturing its own youth, which meant that there was a good chance these players would never be sold. This would, of course, eliminate yet another factor that could throw off results and be difficult to control for.

Data gathering:

At the end of each season, on July 1st, I used the FM Genie Scout to check on CA growth of players. I would also note injuries, first-team games played, and any other interesting factors. Although I spent most of the season on vacation, I would occasionally check in on players to see if they were being tutored, as that could also throw things off.

Here is the CA growth for these players, for the first six seasons:

CONT GRP 110 121 130 138 148 151

HIGH DET 115 126 137 150 160 164

HIGH PRO 116 126 139 147 162 169

HIGH WRK 112 118 126 134 140 145

HIGH AMB 115 122 133 144 159 163

So, over the first six seasons, it seems that the things which were not really a question for me - that AMBITION and PROFESSIONALISM are good for CA growth in training, and that PRO is better for training, but AMB helps CA growth from game time - seem to be vindicated. As long as first-team games were few, Ambition lagged behind Determination, but as soon as Ambition went out for a loan at Arsenal (which happened in Season 5), and got some games under his belt, his CA growth put him right next to Professionalism.

There were two big surprises, though:

- DETERMINATION was for a while the pace-setter for the group in terms of CA growth. In fact, at the end of Season 4, Determination was the leader with 150 CA - although he had played 14 games that season, with no one else in double-digits. This seemingly contradicts the belief that Determination does not matter for CA growth due to training/"natural growth";

- WORK RATE seemed to have a detrimental effect on CA growth; that is, Control Group, with all of his attributes at 10, grew FASTER than High Workrate, who was identical except for his high Workrate.

The Work Rate result was definitely surprising to see, and the only possible explanation I have is that unlike Determination, Professionalism and Ambition, Work Rate is a physical attribute that costs CA points. It is also an attribute that is particularly important for CMs, ad all of the tested players were CMs. So, the one player with very high Work Rate was already "using up" more CA, and this may have thrown the game off somehow. But, that is pure speculation by me, as I have no idea.

If there is interest, I can keep the experiment going, possibly until the players retire, to also see which, if any, of their hidden attributes change as they mature.

There is interest, so keep going if you can.

However, this is a pretty small sample (maybe you could put some youngsters in Real Madrid at the same time). And the test has to be repeated multiple times, but I guess you all know how empirical science works.

- DETERMINATION was for a while the pace-setter for the group in terms of CA growth. In fact, at the end of Season 4, Determination was the leader with 150 CA - although he had played 14 games that season, with no one else in double-digits. This seemingly contradicts the belief that Determination does not matter for CA growth due to training/"natural growth";

It doesn't matter, if you search the forum you'll see Riz who designed the training module say that determination doesn't matter for player development. The conflicting reports about determination come from people who won't accept the official line from SI and make out determination impacts player growth when it doesn't.

You didn't mention how you actually trained these players which is odd considering this is a training/ca growth experiment as this is a big factor. You also say you holidayed most of the season, which if you did means the assistant actually takes control of training and trains the players how he thinks they should be trained.

As long as first-team games were few, Ambition lagged behind Determination, but as soon as Ambition went out for a loan at Arsenal (which happened in Season 5), and got some games under his belt, his CA growth put him right next to Professionalism.

You treat Pro/Amb like they don't work hand in hand, how come? The player needs both to grow to his full potential. Also the reason why he improved more when he was on loan was probably because he was getting game time at a top club.

You've also missed one vital aspect of all the experiment, you've not shown from season 1 all the way through to season 6 the actual attribute changes which is a necessity really as some attributes cost more than others to learn, some are free etc. So depending on how these attributes have actually been increased and spread then the results would vary drastically because it would force some players to have used more CA naturally compared to others.

Sorry to seem critical but if you are doing an experiement you need to show in what context the changes occurred etc if not its a rather pointless on

Btw workrate is no more important for CM's than it is for defenders, strikers etc.

I'd really have to question your training methods though as the CA should be a lot higher after the 6 seasons at a club like Barca especially when they started at 100 CA. They should have hit nearly full PA by the 4th season.

This still holds true, not sure if you've read it before? But its not changed since this experiment;

http://www.thedugout.net/community/showthread.php?t=67502

Thanks for the responses. I'm going to run a few more seasons tonight and post the results. I am already seeing some changes in the "hidden" attributes that are not due to tutoring (these boys have not been tutored yet), so there'll be some data there as well.

It doesn't matter, if you search the forum you'll see Riz who designed the training module say that determination doesn't matter for player development. The conflicting reports about determination come from people who won't accept the official line from SI and make out determination impacts player growth when it doesn't.

I will look for the Riz quote, but the basic idea here is that there are several potential sources of CA growth: competitive game time, non-competitive game time, training (as far as I know, this is MINIMAL, but I still assume that a player with NO training would grow slower than that same player with training), "natural progression" due to age, and any other x-factors that SI has not seen fit to mention and players have not discovered.

EDIT: I dug up two posts by Riz Remes that have the word "determination" in them. In one of them, he does state that Determination, along with Professionalism, Nat Fit and Injury Proneness, determine when a player peaks. So, maybe Determination is not related to training per se, but it does seem to be somehow related to CA growth/decline

In my data, the players from the ages 14-18 played very little competitive, first-team football. So, their CA growth was due to all the other factors BUT that one. The CA growth for High Determination during this time was higher than Control Group, and comparable to Professionalism and Ambition (indeed, better than Ambition). That's the DATA. There could be different interpretations of it, and there could be different criticisms of my methods, but the data is what it is. I would be curious as to what PRECISELY was written by Riz, as the fact that (DET is not used by the training module) =/= (DET doesn't matter for CA growth due to non-competitive football).

You didn't mention how you actually trained these players which is odd considering this is a training/ca growth experiment as this is a big factor. You also say you holidayed most of the season, which if you did means the assistant actually takes control of training and trains the players how he thinks they should be trained.

As to my methodology, I went on holiday so as to keep the experiment time manageable. I think this is a pretty standard method of running this type of experiment - indeed, Cleon, the study you linked to (which is the one I mentioned in my first post, I just couldn't dig it up, and now I know why - it was on another website) also uses this method. It is true that if the assistant used DIFFERENT schedules for each of these players, that would introduce noise.

You've also missed one vital aspect of all the experiment, you've not shown from season 1 all the way through to season 6 the actual attribute changes which is a necessity really as some attributes cost more than others to learn, some are free etc. So depending on how these attributes have actually been increased and spread then the results would vary drastically because it would force some players to have used more CA naturally compared to others.

But, for players who are ALL MC's, the CA weight of different attributes should be identical. By making all of my test cases play the exact same position, I took this source of "noise" out of the experiment.

You treat Pro/Amb like they don't work hand in hand, how come? The player needs both to grow to his full potential. Also the reason why he improved more when he was on loan was probably because he was getting game time at a top club.

PRO and AMB certainly do work hand in hand. But I wanted to try and isolate their impact, so that's why I varied them one at a time.

Btw workrate is no more important for CM's than it is for defenders, strikers etc.

What I meant when here is that Workrate is MORE EXPENSIVE, in CA terms, for MCs than for any other position. My source for this statement is:

http://community.sigames.com/showthread.php/381779-Current-ability-cost-of-of-attributes-position-breakdown

I'd really have to question your training methods though as the CA should be a lot higher after the 6 seasons at a club like Barca especially when they started at 100 CA. They should have hit nearly full PA by the 4th season.

Again, the impact of MY training methods on CA is not what was being tested here. Impact of attributes on CA growth is what was being tested. I was in fact trying to minimize ANY source of difference among these players other than the attributes being tested.

Thanks again for the response, I wasn't expecting something this long and detailed, but I certainly appreciate it. If there are any other holes you can poke in what I write, please do, as I agree that my sample size here is VERY small, and further work is necessary.

OK, here are the results of season 7:

All players are now 21 years old

High Determination has decided to join Sevilla - this is my mistake, as I should've given him a contract sooner. I may still go back to an earlier save, sell all Barcelona MCs, and give the experimental kids big contracts. But, not at the moment.

High Workrate has been allowed to leave Barca when his contract ran out. That means that a CA 148 player, with 200 PA, has been let go, with no one signing him up. I re-signed him for the team.

Here are the raw numbers:

CONt GRP 153 3

HIGH DET 168 3

HIGH PRO 178 32

HIGH WRK 148 -

HIGH AMB 168 11

First number is CA as of July 1, 2019. Second number is total first-team games played. PROFESSIONALISM is clearly ahead, but he has been playing more games than anyone.

As to other attributes grown, here it is:

CONT GRP CON 13, VER 13

HIGH DET CON 12, VER 12

HIGH PRO CON 12, VER 12, IMP 11, TEM 11

HIGH WRK CON 13, VER 13, IMP 11

HIGH AMB CON 12, VER 12

So far, the 21-year-olds have improved consistency, versatility (most of it in the last year or so), important matches, and there has been a temperament improvement in one of them.

As to my methodology, I went on holiday so as to keep the experiment time manageable. I think this is a pretty standard method of running this type of experiment - indeed, Cleon, the study you linked to (which is the one I mentioned in my first post, I just couldn't dig it up, and now I know why - it was on another website) also uses this method. It is true that if the assistant used DIFFERENT schedules for each of these players, that would introduce noise.

The AI/Assistant tends to focus on individual attributes when the player is young rather than role training, so depending on which attributes the assistant worked on when you holidayed would mean you lost control of the experiment due to some attributes costing higher value to learn. Plus it means more emphasis was on this attribute meaning everyone might not have been trained the same way which then makes any comparison obsolete.

But, for players who are ALL MC's, the CA weight of different attributes should be identical. By making all of my test cases play the exact same position, I took this source of "noise" out of the experiment.

This is the thing though you didn't remove the noise. Even though you set the players up to be identical if player A saw an improvement in a more costly attribute than player B then all the information you presented while accurate wouldn't actually be a true reflection of attribute growth under controlled conditions as it would skew the weighting you had used. So it wouldn't actually show the effect of pro/amb. After all your end results are based on how much CA growth there has been and trying to break down what's influenced it.

What I meant when here is that Workrate is MORE EXPENSIVE, in CA terms, for MCs than for any other position. My source for this statement is:

http://community.sigames.com/showthr...tion-breakdown

This makes sense now you added some context to what you actually said

Again, the impact of MY training methods on CA is not what was being tested here. Impact of attributes on CA growth is what was being tested. I was in fact trying to minimize ANY source of difference among these players other than the attributes being tested.

But training actually impact's CA growth and isn't all just about redistributing the attributes, training actually gives/free's up CA as well, that was my point .

Before FM14 almost everyone thought training just redistributes the attributes but Riz clarified this and said training actually frees up CA too.

I know it seems like I am being negative but I'm not, I actually like the idea of your tests and training was the only reason I've been playing the game for the past 4 years so I've done a lot of my own test and had access to information on how training actually works and learning the ins and outs of it. I only give critical feedback so you can make your tests better

You mentioned you've seen some improvements of attributes that are hidden, I'll list all the ones that can change by tutoring if you like so you don't have to focus on them at any point if you decide to do some tutoring tests etc

Adaptability

Ambition

Determination

Loyalty

Pressure

Professional

Sportsmanship

Temperament

Controversy

All those can be changed via tutoring, the ones that can't be changed are;

Consistency

Dirtiness

Imp. Matches

Versatility

Have you thought about purchasing the ingame editor? It makes experiments like this a lot easier as you can manually change the attributes and it allows you to focus on CA growth more due to giving you total control over every attributes visible and hidden while doing your experiment.

If I can help in any way I'll try

OK, here are the results of season 7:

All players are now 21 years old

High Determination has decided to join Sevilla - this is my mistake, as I should've given him a contract sooner. I may still go back to an earlier save, sell all Barcelona MCs, and give the experimental kids big contracts. But, not at the moment.

High Workrate has been allowed to leave Barca when his contract ran out. That means that a CA 148 player, with 200 PA, has been let go, with no one signing him up. I re-signed him for the team.

Here are the raw numbers:

CONt GRP 153 3

HIGH DET 168 3

HIGH PRO 178 32

HIGH WRK 148 -

HIGH AMB 168 11

First number is CA as of July 1, 2019. Second number is total first-team games played. PROFESSIONALISM is clearly ahead, but he has been playing more games than anyone.

As to other attributes grown, here it is:

CONT GRP CON 13, VER 13

HIGH DET CON 12, VER 12

HIGH PRO CON 12, VER 12, IMP 11, TEM 11

HIGH WRK CON 13, VER 13, IMP 11

HIGH AMB CON 12, VER 12

So far, the 21-year-olds have improved consistency, versatility (most of it in the last year or so), important matches, and there has been a temperament improvement in one of them.

You're beginning to see now how professional players who have game time improve quicker and at a higher rate compared to those without.

Are the professional players also the ones seeing more changes with hidden attributes?

Cleon, thanks again for taking the time to respond.

It is true that by no taking direct control of the team I pretty much allowed the Assistant to assign training. I thought, for no particular reason that I can name, that the Assistant would just train everyone on "Balanced-Medium" with no player-specific training. Oh well.

In any case, one thing that struck me when I read your posts is that the attributes that CAN'T be tutored - Consistency, Imp Matches, Dirtiness, Versatility - are precisely the attributes that are growing on their own, with no tutoring (as there has been no tutoring in my experiment so far). Well, except Dirtiness - that one hasn't changed.

In any case, I'll continue to run this experiment, even though Determination is at Sevilla, and Workrate is at Stuttgart; these are teams with enough reputation where the exercise might still make sense. I now *really* want to see which hidden attributes change by the time these boys retire, and which do not.

BTW, your Ajax thread has inspired my own Lechia Gdansk youth project. At some point, I hope to field a team that can take me through a season in the Polish Ekstraklasa consisting entirely of home-grown talent...

OK, here's the chart at the end of Season 8. This was the first season where ALL players got significant 1st-team playing time.

Here are the raw numbers:

CONT GRP 163 45

HIGH DET 177 52

HIGH PRO 190 57

HIGH WRK 161 30

HIGH AMB 176 49

As before, the first number is CA, the second number is competitive games played for their respective teams (although a few of these boys have started picking up caps). The biggest noise factor here is the fact that Determination now plies his trade at Sevilla, whereas Workrate is at Stuttgart. These are lower rep teams, but they play in big competitions and do give these players regular playing time.

Professionalism is ahead of the rest, Determination and Ambition are neck-and-neck in second, with Control Group and Workrate lagging behind.

I looked at the training done by the Assistant Manager to see if he was doing anything special - he's not, no Barcelona player has any specific training regime set. So, whatever variance there is in CA growth is not due to training.

Injuries were also not a big factor, with no player injured more than once, and no injury lasting beyond 4 weeks. Professionalism is the one who was out for 4 weeks, and he still recorded the highest CA growth.

What is surprising, though, is that the HIDDEN attributes seem to have DECREASED with some of these players. It is still the case that, other than a lone improvement in Temperament, only CONSISTENCY, IMPORTANT MATCHES and VERSATILITY have improved, but during Season 8 they seem to have decreased. One way I can explain it would be if these attributes were somehow related to a player's age, or the level of football he's playing. So, a talented 19-year old, who is likely to play mostly reserve/U-21 football, could play consistently at that level, but be inconsistent a level higher. Another way would be if these attributes were highly dynamic - so if these players played some poor matches at the senior level, or had erratic performances, then their consistency and important matches will have gone down. I'll continue to monitor this.

OK, here's the chart at the end of 10 seasons:

Professionalism has reached his peak at 200. No one else is close - Ambition and Determination are at least 10 points behind, and Workrate and Control Group are stuck at the 170s. No one has grown more than 4 CA. All the test players have had 30-50 first-team matches, and none have been injured for more than 4 weeks this season.

Hidden attributes have remained unchanged, except for Ambition gaining a point in Important Matches.

So, even though 10 years have passed, these players have gained NO POINTS in personality development. Sole exception is High Professionalism, who lost 1 point of Aggression, and gained 1 point of Temperament, presumably through being punished for a red card on the field. But, there is NO "natural" growth of personality so far.

But, for players who are ALL MC's, the CA weight of different attributes should be identical. By making all of my test cases play the exact same position, I took this source of "noise" out of the experiment.

I like that there are experiments on these things to get some numbers crunched out. Gives something to hold on to when thinking on these things .

One comment on the setup though: By making all the 5(?) players the same position (MC), are you not potentially creating noise by the fact that only 3 players can play in their favoured position at any one point? I.e. if your U19/U21 manager has a preferred personell these players will play more at youth level and therefor develop slightly faster independent of hidden attributes? (Say for example if your youth manager really rates the "high DET"-player and therefore uses him in every match, but does not rate the "high AMB" one, so he only plays when others are injured).

I understand your reasoning for selecting equal players, but could this not potentially cause these problems?

Raffen,

What you point out is definitely a source of error. There is no perfect way to resolve it and still be able to vacation my way through the game so as to make this experiment feasible. There are, however, at least three ways to mitigate it:

1) pick a position that is likely to be heavily used. I.e., some teams don't even have an AMC, or a DL, but the vast majority of teams have 2 MCs.

2) keep track of the number of games played (which I did), which at least gives some clue as to the distortion

3) as soon as your players get good enough for 1st team, SELL ALL MCs in the team. Which is what I did with Barca.

That said, two of these youngsters are now at other, albeit still high-rep, teams.

Here's a picture of HIGH PROFESSIONALISM - the first one of the bunch to hit 200 CA, at the age of 24. I did not interfere with his progression or set up any special training regimes for him. All he got was Balanced training, game time, and natural growth:

Here are the interesting bits:

- his physical attributes are beastly, but only Strength reached 20, whereas Jumping only got to 12. I've found before that Strength is the easiest physical attribute to train, whereas Jumping is the hardest.

- since he is only 24, it is most probable that he has reached his Physical peak already, but may be some way off his Mental peak, and possibly even his Technical peak. These different attribute groups do peak at different times, and he is living proof.

- both his Tackling and his Teamwork are quite low. Strange to see in a MC, especially the Tackling. Teamwork is also kind of strange, since he is at Barca.

- even at 200 CA, at the age of 24, his attributes are far from perfect - most are 16-18 or even lower. I'm guessing that as he ages, his distribution of attributes will change - maybe the same attribute set at 24 is simply more expensive than at 28 or 30.

Here is the situation at the end of Season 12. Players are now 26 years old. They have probably reached past the mid-points of their careers yet, as far as the MC position is concerned, they are not yet at their overall peak.

PROFESSIONALISM maintained his 200 CA for 1 season, but then tore his hamstring in the pre-season of his 12th year, and lost 5 CA as a result. Most of the others continued to show *some* growth and development, except AMBITION, who is fading for some reason.

All players continue to get 40-50 competitive games for their club sides every season.

Could you post a graph showing values of change between subsequent seasons? I always wanted to know how fast (and how long) they develop in their twenties.

I applaud your effort.

-edit-

Captions on the X axis on your last graph are suboptimal.

Are we certain that holidaying sees the Assistant take over training? I just did a very short test on my Porto save. I plan on doing a longer (but faster..) test on a save with less leagues over the weekend.

Anyway, I had the training set to Low - Team Cohesion. Maxed out Match Preparation. I also got a few players to train for positions they were orange/yellow (Maicon-SW, Herrera-ML, Mangala-WBL). Incidentally my tactic doesn't have any of these positions in it and the assistant was ordered to use my tactics. The result: I came back after 10 weeks with 10/63 players unhappy with their low level of training. Also Herrera had gone from yellow for M-L to the first shade of green. Not the most robust test in the world but it seems to me that maybe unless you hand responsibility for training over to another member of staff via the Staff Responsibilities screen then the ME defaults to whatever you have set whilst on holiday. Like I say, this was a very quick and dirty test so needs more work but I'd be leaning towards that theory for now.

It would be fascinating to find out whether the effects of DET, PRO & AMB are additive.....

Could you post a graph showing values of change between subsequent seasons? I always wanted to know how fast (and how long) they develop in their twenties.

Are you asking about ages later than 26? If so, I haven't ran the experiment that far yet... but I will.

Are we certain that holidaying sees the Assistant take over training? I just did a very short test on my Porto save. I plan on doing a longer (but faster..) test on a save with less leagues over the weekend.

Full agreement here. I also do not believe that the Assistant Manager issues his own instructions. I always thought he kept things at "Balanced-Medium" level in general team training, which is the default. This idea that Assistant Managers train players as they see fit was suggested by Cleon above, and I am reluctant to disagree with Cleon as he's done plenty of investigating of training in FM. However, I also have not seen any evidence that an Assistant Manager does ANY training work, in a player-controlled team.

Now, if you check your training screen, your coaches DO offer suggestions for training focus for particular players, and I can easily believe that AI-controlled managers listen to these instructions (that way, AI-controlled teams benefit from individual focus as well, as otherwise they would be heavily disadvantaged vis-a-vis human players), but I haven't seen that with player-controlled teams.

I'd like to see a graph showing values of change between seasons (e.g. the Professional player seems to have been improving by around 10 points yearly until 22, then it took him two years to get the remaining 10 points; other players might show us when the growth stops when it's not hitting PA).

It seems from the graph that their progress is pretty stable until they reach 22/23 and then gets slower.

I have some potentially great players that I managed to get when they were quite old (for developing) already (South Americans produce great talent and they don't release them before they're 18), and I'm fighting to make them reach their potential. I also have and had players that I failed to save from not reaching their potential.

zenzen,

Here's a chart that will hopefully show what you are asking about:

It shows that the test players experienced huge growth until they turned 19 (that's growth occuring WITHOUT significant 1st-team playing time). Between the age of 18 and 19, two of these players played 10+ 1st team games at Barca, whereas AMBITION played 18 matches for Arsenal, on loan. This lead to a HUGE growth spurt.

The growth then dropped off until about 21. Coincidentally, players seem to be unhappy about not being sent out on loan in that 18-21 age frame, so maybe at that time they NEED 1st team playing time, even if it is in lower divisions...

After 21, these players became regularly involved in Barcelona's 1st team (and, within about 1 year, two went to other clubs, but were always very involved in 1st team), and from 21 to about 24 they experienced a second burst of growth, which ended with HIGH PROFESSIONALISM achieving CA of 200 at the age of 24.

Since then, changes have been smaller, despite constant, heavy 1st team involvement, and despite the fact that plenty of PA remains for all players except HIGH PROFESSIONALISM. Again, the game hints at this when it gives us the hint that players after the age of 24 are unlikely to experience bursts of growth - bu the in-game hint limited this statement to "natural growth" - whereas here it seems even 1st-team playing time at Barcelona, Chelsea, and AC Milan does not produce drastic growth despite some of the players having 15-20 PA points left to fill...

I really appreciate the idea and the effort you're putting in this, however I'm afraid this research isn't really reliable unless you'll take a larger sample size. Completely equal conditions at all times for all players is just not possible no matter how you try. With a decent sample size these small differences do average each other out though, in a way making it easier for you as you don't have to care about external conditions.

That being said, your research does seem to confirm my personal experience that professionalism is the most important attribute in player development. Personally I've never noticed a significant difference in player development between fairly high and very high levels of ambition, but players with a lack of ambition do seem to do a lot worse. Ever considered doing it the other way around and see how lacking one of these attributes impacts their development?

so basically what you have achieved, regarding fulfilling potential, by the age of 24 that is it? with peaks during 18-19 and 21-24.

will be interesting to see if there will be any late bloomers, not sure this is implemented in the game though. also, the sample is very small.

keep up the experiment, cant wait to see what happens next.

This is what the data show so far, but I don't think that player growth HAS to slow down at 18. What seems to happen is that until the age of 18, players who train and are not injured, experience great bursts of "natural" growth. AFTER the age of 18 or so, either they become regularly involved in 1st team (at their own club, or perhaps on loan?) or their growth stagnates a bit.

I really appreciate the idea and the effort you're putting in this, however I'm afraid this research isn't really reliable unless you'll take a larger sample size. Completely equal conditions at all times for all players is just not possible no matter how you try. With a decent sample size these small differences do average each other out though, in a way making it easier for you as you don't have to care about external conditions.

That being said, your research does seem to confirm my personal experience that professionalism is the most important attribute in player development. Personally I've never noticed a significant difference in player development between fairly high and very high levels of ambition, but players with a lack of ambition do seem to do a lot worse. Ever considered doing it the other way around and see how lacking one of these attributes impacts their development?

Regarding sample size, there is no question that 5 players does not seem like much. But, from another perspective, if Professionalism affects CA growth, the question is, HOW OFTEN does it do that? If CA is updated, within the game, every 1 month (for example), then if I run the experiment for 12 years, that's 120 rolls of the dice per player, each roll being independently affected by attributes.

Now, a little tidbit: HIGH PROFESSIONALISM has the personality "Model Professional" - that is, high Professionalism ALONE is enough to make him a model pro. HIGH DETERMINATION has the personality "Driven" - that is, high Determination ALONE is enough to make him highly determined. HIGH AMBITION has the personality.... "Fairly Ambitious". Of course, if he'd had 1 less point in Loyalty, he'd be "Very Ambitious", so this is just a curious little bit of info, but still, it does seem like Ambition is less of a "stand-alone" attribute.

And yes, if I get around to it, my next test will involve more variance within each attribute. For example, HIGH WORKRATE was for a while doing the worst out of the group, but I just can't accept the conclusion that 20 Work Rate is a NEGATIVE attribute, all other things being equal. So, if the test shows that, e.g. a player with 1 Work Rate is really completely useless, that would at least vindicate that attribute somewhat.

Now, a little tidbit: HIGH PROFESSIONALISM has the personality "Model Professional" - that is, high Professionalism ALONE is enough to make him a model pro

No it's not, he needs between 10-20 temperament too.

HIGH DETERMINATION has the personality "Driven" - that is, high Determination ALONE is enough to make him highly determined.

Again its not.

This is the full list of what makes up personality types because judging by those 2 sentences you did you don't actually understand what makes up a personality type as you think high pro and high det alone are enough when they aren't;

Newgens

• Model Citizen Pro 18-20, Det 18-20, Pre 18-20, Amb 18-20, Loy 18-20, Spo 18-20, Tem 18-20
  
• Model Professional Pro 20, Tem 10-20
  
• Professional Pro 18-19, Tem 10-20
  
• Perfectionist Pro 18-20, Det 18-20, Amb 18-20, Tem 1-9
  
• Resolute Pro 15-20, Det 15-20, Pre 1-16 Spo 5-20
  
• Temperamental Tem 1-4, Pro 1-10
  
• Driven Det 20, Amb 10-20
  
• Determined Det 18-19, Amb 10-20
  
• Slack Pro 1, Det 1-9, Tem 5-20
  
• Casual Pro 2-4, Det 1-9, Tem 5-20
  
• Very Ambitious Amb 20, Loy 1-9, Det 1-17
  
• Amibitous Amb 16-19, Loy 1-9, Det 1-17
  
• Unambitious Amb 1-5, Loy 11-20
  
• Honest Spo 20, Det 1-9, Pro 5-20
  
• Sporting Spo 18-19, Det 1-9, Pro 5-20
  
• Easily Discouraged Det 1, Amb 1-9, Spo 1-17, Pro 5-20
  
• Low Determination Det 2-5, Amb 1-9, Spo 1-17, Pro 5-20
  
• Unsporting Spo 1, Det 11-20
  
• Realist Spo 2-4, Det 11-20
  
• Very Loyal Loy 20, Amb 6-7, Det 6-20
  
• Loyal Loy 18-19, Amb 6-7, Det 6-20
  
• Iron Willed Pre 20, Det 15-20, Spo 5-20
  
• Resilient Pre 17-19, Det 15-20, Spo 5-20
  
• Spineless Pre 1, Det 1-9, Pro 5-20, Spo 1-17
  
• Low Self-Belief Pre 2-3, Det 1-9, Pro 5-20, Spo 1-17
  
• Light-Hearted Pre 15-20, Spo 15-20, Tem 10-20, Pro 1-17
  
• Spirited Pre 15-20, Tem 10-20, Pro 11-17, Spo 1-14
  
• Jovial Pre 15-20, Tem 10-20, Pro 1-10, Spo 1-14
  
• Fairly Professional Pro 15-20, Det 1-14
  
• Fairly Determined Det 15-20, Pro 1-14, Spo 5-20, Pre 1-16
  
• Fairly Amibitous Amb 15-20, Pro 1-14, Det 1-14
  
• Fairly Loyal Loy 15-20, Pro 1-14, Det 1-14, Amb 6-14
  
• Fairly Sporting Spo 15-20, Pro & Det & Amb & Loy all 1-14
  
• Balanced Pro 1-14, Det 1-14, Amb 1-14, Loy & Spo 1-14
  
• Born Leader Inf 20 & Det 20. Age needs to be 23+ (if age requirement not met, he would be Driven).
  
• Devoted Loy 20 & Amb 6-7
  
• Leader Inf 19 or Inf 20 & Det < 20
  

None Newgen's

• Model Citizen Pro 18-20, Det 18-20, Pre 18-20, Amb 18-20, Loy 18-20, Spo 18-20, Tem 18-20
  
• Model Professional Pro 20, Tem 10-20
  
• Professional Pro 18-19, Tem 10-20
  
• Perfectionist Pro 18-20, Det 18-20, Amb 18-20, Tem 1-9
  
• Driven Det 20, Amb 10-20
  
• Determined Det 18-19, Amb 10-20
  
• Very Ambitious Amb 20, Loy 1-9, Det 1-17
  
• Amibitous Amb 16-19, Loy 1-9, Det 1-17
  
• Honest Spo 20, Det 1-9
  
• Sporting Spo 18-19, Det 1-9
  
• Very Loyal Loy 20, Amb 6-7
  
• Loyal Loy 18-19, Amb 6-7
  
• Iron Willed Pre 20, Det 15-20
  
• Resilient Pre 17-19, Det 15-20
  
• Light-Hearted Pre 15-20, Spo 15-20, Tem 10-20, Pro 1-17
  
• Spirited Pre 15-20, Tem 10-20, Pro 11-17, Spo 1-14
  
• Jovial Pre 15-20, Tem 10-20, Pro 1-10, Spo 1-14
  
• Resolute Pro 15-20, Det 15-20, Pre 1-16
  
• Fairly Professional Pro 15-20, Det 1-14
  
• Fairly Determined Det 15-20, Pro 1-14, Pre 1-16
  
• Fairly Amibitous Amb 15-20, Pro 1-14, Det 1-14
  
• Fairly Loyal Loy 15-20, Pro 1-14, Det 1-14, Amb 1-14
  
• Fairly Sporting Spo 15-20, Pro & Det & Amb & Loy all 1-14
  
• Balanced Pro 1-14, Det 1-14, Amb 1-14, Loy & Spo 1-14
  
• Born Leader Inf 20 & Det 20. Age needs to be 23+ (if age requirement not met, he would be Driven).
  
• Devoted Loy 20 & Amb 6-7
  
• Leader Inf 19 or Inf 20 & Det < 20
  

And workrate does not effect player growth at all.

In all your tests you've also ignored and not once mentioned event based development which can see an immediate increase in CA.

Regarding sample size, there is no question that 5 players does not seem like much.

Actually, you have 5 samples with a sample size of 1.

Now, a little tidbit: HIGH PROFESSIONALISM has the personality "Model Professional" - that is, high Professionalism ALONE is enough to make him a model pro. HIGH DETERMINATION has the personality "Driven" - that is, high Determination ALONE is enough to make him highly determined. HIGH AMBITION has the personality.... "Fairly Ambitious". Of course, if he'd had 1 less point in Loyalty, he'd be "Very Ambitious", so this is just a curious little bit of info, but still, it does seem like Ambition is less of a "stand-alone" attribute.

Well, that just shows that the personality description can be very deceptive and counter intuitive. A player with a 'better' personality description, doesn't necessarily have a better personality. Not sure whether this says much about the importance of the stand-alone value of the attribute though.

And yes, if I get around to it, my next test will involve more variance within each attribute. For example, HIGH WORKRATE was for a while doing the worst out of the group, but I just can't accept the conclusion that 20 Work Rate is a NEGATIVE attribute, all other things being equal. So, if the test shows that, e.g. a player with 1 Work Rate is really completely useless, that would at least vindicate that attribute somewhat.

I don't think there's anything that suggests that Workrate has a negative effect on development, the difference with the Control Group is so small it just seems to indicate it doesn't have an influence on development at all. But actually all the results are close enough that with your current sample size it's all within the standard deviation and it's still a possibility that actually none of the attributes matter, and it's all just coincidence.

Cleon,

Maybe I wasn't being clear again. I write within the context of the test I'm running. In this test, I started with players who had 10 for ALL attributes, which to me meant they were perfectly average, and indeed, had "Balanced" personalities. And then I started to make adjustments. And so, if the ONLY change I made was to adjust Professionalism from 10 to 20, he became a Model Pro. If the only change I made was to adjust Determination from 10 to 20, he becomes Driven. When I adjust Ambition from 10 to 20, he becomes "fairly ambitious". I also clearly wrote that this is nothing but a bit of trivia, since adjusting Loyalty from 10 to 9 would turn the "fairly ambitious" player to "very ambitious", and clearly a 1-point drop in Loyalty would have precisely 0 effect on a player's CA growth.

And workrate does not effect player growth at all.

The data indicate this as well, at least so far. However, I remembered reading a test where Work Rate had a small, but positive, impact on player growth (I even thought the Shrewnaldo test you linked to had that info, but it does not, so now I don't know where I read it). Maybe because a high work rate player gets involved in more game situations, thus giving him more opportunities for CA growth per minute spent on the field, or maybe because he applies himself more in training. Regardless, it's one of the factors I wanted to test. I am *very* surprised to find that Work Rate lagged behind Control Group for most of the test. Still, I'm not prepared to state that Work Rate has no effect on player growth, because for me, one test is not sufficient to conclude that.

In all your tests you've also ignored and not once mentioned event based development which can see an immediate increase in CA.

Yep, because, again, that is NOT what is being tested. To mention things like in-game events is to BEG THE QUESTION in a test like this, because for all we know, a Model Pro or a Driven player is MORE LIKELY to get himself involved in CA-raising in-game events. There is no way, that I know of, to control for in-game events in an experiment of this type.

Maybe I wasn't being clear again. I write within the context of the test I'm running. In this test, I started with players who had 10 for ALL attributes, which to me meant they were perfectly average, and indeed, had "Balanced" personalities. And then I started to make adjustments. And so, if the ONLY change I made was to adjust Professionalism from 10 to 20, he became a Model Pro. If the only change I made was to adjust Determination from 10 to 20, he becomes Driven. When I adjust Ambition from 10 to 20, he becomes "fairly ambitious". I also clearly wrote that this is nothing but a bit of trivia, since adjusting Loyalty from 10 to 9 would turn the "fairly ambitious" player to "very ambitious", and clearly a 1-point drop in Loyalty would have precisely 0 effect on a player's CA growth.

You was making out that all you needed to do was change Professionalism and made out that attribute was all that was needed. Now for everyone reading this thread who doesn't understand the training side of the game as well as I do then they'd believe this. But both myself and you know this wasn't as simple as this was it. lets look at exactly what you said shall we, so everyone is crystal clear about how it works, if not its threads like this that spread misinformation and this is how myths start;

You said;

Now, a little tidbit: HIGH PROFESSIONALISM has the personality "Model Professional" - that is, high Professionalism ALONE is enough to make him a model pro

This is why I corrected you because you wasn't clear and are making out that this attribute alone is enough to give you the model pro description when it isn't. Hence why I posted the guide so people can actually see what makes up the personality so its clear for all to see how it would change if they changed the values.

All your post did was further complicate things which are simple and easy to understand. Not many people understand the training/personality side of the game so when you post something that is incorrect I will pull you up on it to stop myths being started or misinformation from spreading. What's point of saying something is trivia when we can just deal in facts and how the game actually works? It saves you time and stops anyone else being confused by what you write hence the post I did with the personality values.

Yep, because, again, that is NOT what is being tested. To mention things like in-game events is to BEG THE QUESTION in a test like this, because for all we know, a Model Pro or a Driven player is MORE LIKELY to get himself involved in CA-raising in-game events. There is no way, that I know of, to control for in-game events in an experiment of this type.

You say you aren't testing this BUT and its a big but, its all linked and you cannot separate it. If you are testing CA growth then this cannot be discounted because its the only thing in game that instantly changes CA and is visible instantly after the event. Now if you've had a player who has been lucky enough to have this happen then the whole result of your comparison of players is flawed and has no accuracy at all.

And you are wrong again, someone more driven or more professional aren't more likely to involve themself in in game events at all, this would still be determined by his technical/mental/physical attributes regardless of his personality type.

You seem to want to test CA growth yet at the same time you dismiss important stuff which directly impacts it.

I need to clear something else up too;

Maybe because a high work rate player gets involved in more game situations, thus giving him more opportunities for CA growth per minute spent on the field, or maybe because he applies himself more in training.

The thing that affects training is professionalism it has nothing to do with workrate, ambition or anything like that. How well someone applies themself is down to professionalism and professionalism alone in terms of training and applying himself. It's sentences like the above that causes confusion for people because with all due respect we have people running test about training and showing proof of things yet don't understand the mechanics or the workings behind what they are trying to understand. So when this happens we get things like this, where you imply work rate could possibly be linked to applying himself in training, when I can state 1000000000000000000000000000000% this is not and never has been the case at all.

I know I'm being critical but its hard not to when you know 100% how something works.

Actually, you have 5 samples with a sample size of 1.

If I understand you correctly, if I had FIVE players per category, instead of just one, then it would be 5 samples with a sample size of 5 per sample? I would have liked that, but that would mean that these player could not all be in the SAME position, and at the SAME team. So, further sources of noise would be introduced.

Well, that just shows that the personality description can be very deceptive and counter intuitive. A player with a 'better' personality description, doesn't necessarily have a better personality. Not sure whether this says much about the importance of the stand-alone value of the attribute though.

It really doesn't say anything about the attribute's importance. It was just something I thought looked funny. But, in fact, all of the "fairly x" descriptions COULD have attribute x at 20, but not have other attributes within certain ranges. So yeah, the word "fairly", which to me suggests a 11-15 attribute range, is in fact MISLEADING, because it means 15-20 attribute range PLUS some other attributes being low.

I don't think there's anything that suggests that Workrate has a negative effect on development, the difference with the Control Group is so small it just seems to indicate it doesn't have an influence on development at all. But actually all the results are close enough that with your current sample size it's all within the standard deviation and it's still a possibility that actually none of the attributes matter, and it's all just coincidence.

The difference is very small, but it is also negative - whereas I expected it to be small, but positive, based on an earlier test I remember reading about, but cannot find now. As to your other point, I really HOPE that the attributes DO matter, because otherwise the hours I spent finding tutors, growing tutors, and then arranging for tutoring and obsessing over it would basically be wasted time... and the non-heavily-tutoring AI would have its final, metallic laugh at my expense...

It's incredibly difficult, or just plain impossible, to draw any definite conclusions from one sample set - particularly one where the results are not comparable given the diverse circumstances which these players are exposed to. There just isn't a proper control group.

You need a MASSIVE sample set to give a definable set of results and (apologies for the self promotion here) but I've tried to summarise those results in my tutoring guide on my blog - http://footballmanagerveteran.wordpress.com/2012/08/21/shrews-guide-to-tutoring-in-fm/

One of the key sources that I draw on is a "The Dugout" thread from Maestro Ugo where he used a holiday save that someone had produced with (I think) 200 years worth of results. Mugo extracted all this data and analysed it incredibly thoroughly using a number of mathematical formulae - a lot of which I didn't fully understand. You can find his thread here - http://www.thedugout.net/community/showthread.php?t=75695

The results from that are quite clear, though, and, due to the enormous sample data that he is able to draw from, removes the majority of the 'noise' created by external influences.

I hope that helps a little.

You was making out that all you needed to do was change Professionalism and made out that attribute was all that was needed.

Yep. If you start with base values of 10 for every attribute, as I did, and clearly stated so at the beginning of this thread, then ALL you need to do is change Professionalism from 10 to 20 and you get a Model Pro personality. But again, this IS just trivia stuff, so I'd rather not spend more time on it.

You say you aren't testing this BUT and its a big but, its all linked and you cannot separate it. If you are testing CA growth then this cannot be discounted because its the only thing in game that instantly changes CA and is visible instantly after the event. Now if you've had a player who has been lucky enough to have this happen then the whole result of your comparison of players is flawed and has no accuracy at all.

And you are wrong again, someone more driven or more professional aren't more likely to involve themself in in game events at all, this would still be determined by his technical/mental/physical attributes regardless of his personality type.

Cleon, if I had access to a Magical In-Game-Event Calculator (M.I.C. for short), I would certainly use it to (a) control for this factor, and (b) test whether certain personality types have more in-game events. But, SI has not seen fit to provide anyone, to my knowledge, with a M.I.C. So, in-game events are not something I can measure. They are part of the output I measure (that is, they are part of the CA growth I can measure), but I cannot isolate them. If the developers did provide you with a M.I.C., and permitted you to share it with others, then by all means let me know and I'll start using it.

But, even without a M.I.C. the experiment doesn't suffer that much. These players have had about 350 1st-team matches under their belts, EACH, so far. If one player had been lucky once, it would not make the "whole result" have "no accuracy". It would just be a single event. Due to the amount of matches played by the players, this is precisely the sort of event that should even itself out, OR, if personalities DO affect player behavior on the pitch (and, therefore DO affect how likely they are to have CA-changing events), then the frequency of CA-changing events would simply be one of the REASONS why, e.g., a Model Pro grows faster. I don't KNOW this to be the case, but it could be. If you are in touch with developers of the game or, better yet, if you've actually read/written the code for the game, and you KNOW, from these sources, that player personalities have NO IMPACT on the frequency of CA-changing events, then that's another story.

The thing that affects training is professionalism it has nothing to do with workrate, ambition or anything like that. How well someone applies themself is down to professionalism and professionalism alone in terms of training and applying himself. It's sentences like the above that causes confusion for people because with all due respect we have people running test about training and showing proof of things yet don't understand the mechanics or the workings behind what they are trying to understand. So when this happens we get things like this, where you imply work rate could possibly be linked to applying himself in training, when I can state 1000000000000000000000000000000% this is not and never has been the case at all.

I know I'm being critical but its hard not to when you know 100% how something works.

Cleon, CA growth due to training is only ONE possible source of CA growth. There are at least three others - natural growth, 1st-team game time, and the in-game events you just mentioned, and possibly others as well (such as non-competitive game time, or U-21 international game time and so forth). So, it may be perfectly true that Work Rate is NOT related to the training module, but that Work Rate DOES affect CA growth from other sources.

I run the test precisely because I don't understand the game as well as I would like to. If SI gave you access to code or other inside information, and based on that you KNOW how things work, and are permitted to share it, by all means do so.

Work rate does not affect player growth or CA growth at all. It's simple really

And yups I had close contact with the development team for 14 years (beta tester, visited SI Towers on many occassions, did some in house testing etc) and know all the ins and outs of how stuff work hence why I can categorically tell people they are wrong with some of their assumptions because I know 100% how it works.

I've posted how stuff works for many years and provided sources for all information on these forums. It appear the search isn't working properly still atm though but all the answers already exist if you do your own searching.

@ Cleon & Shrewnaldo. Big fan of your contributions here and elsewhere. Question for ye.... You've got a hot prospect you want to tutor to hone his mentality (let's say you're not bothered about teaching PPMs at that particular time). All else being equal, given the choice of tutoring him with a Fairly Ambitious/Ambitious/Professional/Model Professional tutor which do you pick? I'm in a bit of a quandary as to whether I should lean towards the various Ambitious personalities or the Professional ones for tutoring. And would you go out of your way to bring a Perfectionist/Model Citizen in?

Work rate does not affect player growth or CA growth at all. It's simple really

The thing that affects training is professionalism it has nothing to do with workrate, ambition or anything like that.

These are two VERY different statements. You've written before that Work Rate is not a factor within the training module. I'm guessing that's one of the bits of info you got directly from SI. However, that's a far cry from saying that Work Rate does not matter for player development overall. Is that also something SI told you?

And yups I had close contact with the development team for 14 years (beta tester, visited SI Towers on many occassions, did some in house testing etc) and know all the ins and outs of how stuff work hence why I can categorically tell people they are wrong with some of their assumptions because I know 100% how it works.

Could you be more specific? I understand of course if SI does not want you to share too much, but can you provide more details on what tests you ran etc.?

I've posted how stuff works for many years and provided sources for all information on these forums. It appear the search isn't working properly still atm though but all the answers already exist if you do your own searching.

I've done some searching - e.g. I tried to find the Riz Remes quote you mentioned that said that Determination DOESN'T matter for player development, and instead found a Riz Remes quote that said Determination WAS a factor in when a player peaks (i.e. Determination DOES matter in player development). So now I have a 6-year-old quote from a developer telling me that Determination does matter, and a more recent quote from a beta tester telling me that it doesn't. My own, very limited, data corresponds to the quote from the developer. I'm sure you can appreciate my dilemma.

I hope that helps a little.

Shrewnaldo, this does not help a little. This helps A LOT. I couldn't possibly replicate those results (I don't think I'd have the patience to create 500 of my own players). So, even though those players were regens, and thus their initial attributes were more random rather than controlled, the sheer number of them makes the analysis by Maestro Ugo much better than mine. His analysis also determines the effects of all the attributes, whereas mine only focused on four attributes.

One of Ugo's conclusions is that Determination and Work Rate DO NOT have an effect, but, at least with respect to Determination, this was a surprise to him, is a surprise to me, seemingly contradicts the Riz Remes quote I dug up, and seemingly contradicts my own results. The explanation here could be that Ugo only measured CA at TWO points - point of generation of the regen, and 13 years down the line. So, it is still possible that higher-Determination players build up CA FASTER, and then stay at their peak LONGER, than lower-Determination players. This, if true, could make the Riz Remes quote, my results, and the Maestro Ugo's analysis, simultaneously be accurate.

When I get time I'll try and post the link when I find it again but if you've been searching you'll know that posts between 2008-13 are missing when searching so it may take some time. I9 posted it on twitter around 2 years ago so it be easier to find the link that way. I'll try when I get the time.

When Riz spoke about it impacted player development he meant from a declining attribute perspective and not gaining CA one, which he clarifies in the quote I'm searching. Natural fitness, determination and professionalism are what stops a player declining fast and can keep him from deteriorating at a much slower rate. But it does not aid in player development at all.

CA speed of increase is linked to the quality and amount of first team football.

Shrewnaldo, this does not help a little. This helps A LOT. I couldn't possibly replicate those results (I don't think I'd have the patience to create 500 of my own players). So, even though those players were regens, and thus their initial attributes were more random rather than controlled, the sheer number of them makes the analysis by Maestro Ugo much better than mine. His analysis also determines the effects of all the attributes, whereas mine only focused on four attributes.

One of Ugo's conclusions is that Determination and Work Rate have an effect, but, at least with respect to Determination, this was a surprise to him, is a surprise to me, seemingly contradicts the Riz Remes quote I dug up, and seemingly contradicts my own results. The explanation here could be that Ugo only measured CA at TWO points - point of generation of the regen, and 13 years down the line. So, it is still possible that higher-Determination players build up CA FASTER, and then stay at their peak LONGER, than lower-Determination players. This, if true, could make the Riz Remes quote, my results, and the Maestro Ugo's analysis, simultaneously be accurate.

No, it wasn't. To quote from the thread:

Determination and work rate are presented as I personally expected them to have at least some influence, but alas - they do not. Work rate has a small zero order effect, but Determination has literally zero impact.

Or:

3) If you don't know what the hidden attributes are like (for example if the player's personality is "balanced"), our best bet is to look for anticipation, decision and creativity. Work rate and Determination have no effect what-so-ever.

Playing attributes will have some impact on how well a player develops in so much as the player will be of higher quality and therefore 'go further'. However, they are not weighted. i.e. High finishing is just as valuable for 'progression' as high work rate or high technique - and you should notice the negative impact of acceleration although it's worth pointing out the counter-argument that I made in post 6 which Mugo corroborated in post 8.

The only attributes which have a DIRECT influence over player development are Ambition and Professionalism.

If I understand you correctly, if I had FIVE players per category, instead of just one, then it would be 5 samples with a sample size of 5 per sample? I would have liked that, but that would mean that these player could not all be in the SAME position, and at the SAME team. So, further sources of noise would be introduced.

You're looking at it wrong. You try to limit the amount of noise, but you should look to limit the amount of influence that noise has on your results. You don't have complete control of the environment, so you will never be able to give all your samples the exact same conditions. With a decent sample size, the effects of the environment should average each other out though, with a single sample, even the most trivial amount of noise can have a noticeable effect on your results.

Actually, your current research method does introduce some noise. Since you have more players (5) than places in your squad (2 or 3 in most tactics). So your samples are actually competing and thus influencing each other.

If you create 5 clones, place them at 5 different clubs and repeat this for each of your groups, actually makes the effect of noise much smaller. Maybe there's something special at Barcelona that negatively influences the development of players with high work rate, but positively influences players with high determination. We don't know. However, if at Bayern, Arsenal, Juventus and Ajax determined players develop better than hard working players as well, even though they're all in a different environment, well that really starts to point in a certain direction.

The difference is very small, but it is also negative - whereas I expected it to be small, but positive

Run it again, if it doesn't have an influence, you have a 50% chance to get your prediction right this time

You really shouldn't pay attention to differences that small, unless you have a huge sample size, so who cares whether is positive or negative, it's insignificant.

If you do research, you better do it in a way that you can actually learn out of it, otherwise you're just wasting your time, or even risk become dumber instead of smarter. I hope you don't take my criticism on your research methods the wrong way, I salute your effort. Anyway, this thread does seem to have provided you with the answers you were looking for, so congratulations.

No, it wasn't. To quote from the thread

Yep, I mis-wrote that. I'm editing it now.

Playing attributes will have some impact on how well a player develops in so much as the player will be of higher quality and therefore 'go further'. However, they are not weighted. i.e. High finishing is just as valuable for 'progression' as high work rate or high technique - and you should notice the negative impact of acceleration although it's worth pointing out the counter-argument that I made in post 6 which Mugo corroborated in post 8.

I read through that again, thought about it, and for me, this actually raises more questions than it answers. For example, if you look at lance101's experiment I linked above:

http://community.sigames.com/showthread.php/381779-Current-ability-cost-of-of-attributes-position-breakdown

What you will see, which I find quite intuitive, is that the CA cost for Acceleration and Pace is the SAME (accounting for position), except for Goalkeepers, whom Master Ugo wasn't testing anyway. So, Pace should also have produced a negative correlation to player development in Master Ugo's experiment, but it did not. This is assuming that the reason why Acceleration had a seemingly NEGATIVE effect on CA growth is that high Acceleration takes up a lot of CA.

One possible explanation would be if Acceleration took up the same CA as Pace, but somehow provided LESS benefit for CA growth. But, this is again counter-intuitive for me, because Acceleration does not depend on Pace, whereas a player may never reach his maximum Pace if his Acceleration is poor.

So, given above, do you have ANY idea why Acceleration had the negative correlation with CA growth, and Pace did not?

Also, it surely cannot be the case that ALL attributes affect CA growth equally? I mean, Master Ugo's test also showed that Creativity, Decisions and Anticipation seem to be key attributes. My understanding is also that Technique is sort of a meta-technical attribute, in that Technique could conceivably be used in ANY situation where a ball is being struck - as contrasted with, say, Corners attribute which is useful for Corners only, and even then, is not a stand-alone ability.

OK, here are the final graphs. The players are about to turn 36, and one of them (CONTROL GROUP) is about to retire, so this seemed like a good ending point:

Here are the year-on-year changes in CA for the players

Here are the interesting points:

- only PROFESSIONALISM and DETERMINATION reached 200 CA. Professionalism stayed there the longest, and was able to recover from injury to reach 200 twice in his career; AMBITION got very close, reaching 199 CA.

- these were all MCs, plying their trade in the highest European leagues, and relatively immune from injury throughout their careers, with excellent Natural Fitness. The high water mark for their career came at the age of 31-32. Only Control Group reached his peak earlier

- AFTER the age 32, or from the 32nd to the 33rd birthday is when the decline began. It became far more pronounced after the age of 34. PROFESSIONALISM and DETERMINATION kept their CA better than the others, which is in line with the Riz Remes quote regarding the impact of Professionalism and Determination on when a player peaks and how fast he declines

I also kept track of the invisible attributes for players, not just the personality ones, but the ones like Versatility etc. that cannot be tutored. Here are the results:

- the attributes which did not change at all, for any player, during any of the 22 seasons of the test, are: Adaptability, Ambition, Controversy, Dirtiness, Injury Proneness, Loyalty, Pressure, Professionalism and Sportsmanship. That includes 7 out of the 8 attributes that make up a player's personality. This means that on- and off-the-field events that affect these attributes are probably quite rare, or maybe they don't occur while a manager is on holiday. I was hoping to find that, e.g., a player's Dirtiness might change in response to red cards, his Adaptability might change as he moves to different leagues, maybe his Pressure or Sportsmanship would change as a result of on-the-field events, but no dice. So, to change a player's personality (beyond Temperament), tutoring is THE way to go. At the same time, I've definitely seen the personalities of my players change, so again, maybe changes do happen when you are not holidaying.

- the ONLY personality attribute that changed was Temperament, apparently according to the familiar formula - when a player is punished for getting a red card, his Aggression drops, while his Temperament rises

- among the five additional hidden attributes, Dirtiness and Injury Proneness did not change, as noted above. Injury Proneness was a bit of a surprise, but perhaps it is controlled by Natural Fitness. The Natural Fitness for all players remained at 20, even at the ripe age of 36.

- Consistency, Important Matches, and Versatility showed the biggest changes. However, a large portion of these changes occurred late in the players' careers. As late as the age of 30, only Professionalism had Consistency at 15. All other attributes for all the players were 14 or lower. Important Matches grew the slowest (throughout the experiment, no one had higher than 12). Versatility and Consistency were about equal in terms of growth. By 36, two players had Consistency 16 and one had it at 17 - but by then they were well past their CA peak.

CONCLUSIONS:

- up until the age of 18 or so, players grow like crazy, even without 1st team playing time. Post-18, that first team playing time is needed, or the rate of growth will experience a dip. So, play your youngsters or loan them, but don't just keep them in the reserves

- even if you did just keep them in the reserves, up until the age 24 or so, growth still occurs in large bursts. AFTER 24, growth slows down dramatically; thus the in-game hint that growth due to "natural progression" slows is a bit misleading - ALL growth slows after 24

- pay attention to Scout reports regarding CONSISTENCY and IMPORTANT MATCHES. These CANNOT be tutored and they grow only slowly. When you look at a player's attributes, it is easy to "cheat" yourself into thinking that this is how good he is. This is misleading. An inconsistent player may conceivably NEVER be a reliable performer who lives up to the attributes you see every time you click on his attributes page. The IMPORTANT MATCHES attribute barely grew for the 22 seasons during which these players routinely played in Champions League finals and other big matches. So, a poor score there may perhaps improve a bit with time, but again, players with low IMPORTANT MATCHES stat may NEVER be reliable in these game situations.

Right, so that's that for this one. Next test might involve, as suggested above, more players in more teams, and at this point I'd probably give them more than one position. Maybe I'll go with AMR/L ST, MLRC and DLRC, so that conceivably a team can play all of the tested players at the same time.

Agreed - great work. I think some of the conflict within this thread is due to theory versus experiment. januszpa is actually reporting what happens. This may well conflict with the intent of the coders. For playing the game, probably only what the code actually does matters, although the difference essentially becomes a bug report.

If you wish to estimate the uncertainties (error bars) on such values, it is likely governed by Poisson statistics, one way or the other. I'd subtract the CA values from 200, add them all up (for each group of players) to get a total value, square root that, divide by the players within that group and that would be a very rough guess at the (one sigma) uncertainty. You'll find you need loads of players to increase accuracy.

I think I've found the test that first suggested to me that Work Rate might have something to do with CA growth - it is right there in the thread linked by Shrewnaldo:

http://www.thedugout.net/community/showthread.php?t=75695

If you look in Post 18 by mixs, his own test shows that Ambition is the most important, followed by Professionalism and... Work Rate. This is different from the result Maestro Ugo got, but I'm not stats-savvy enough to determine who's right and who's wrong

His test also shows that Decisions has a NEGATIVE impact, which is counter-intuitive, but then again Maestro Ugo's test also had some surprising results (Dirtiness being a factor, Acceleration being a factor due to its CA cost, but Pace NOT having the same effect despite having the SAME CA cost for all outfield players etc.)

So, for me, this shows that the matter is far from simple.

Question here, are all first team matches considered equal? Does a player's match rating effect his CA growth? If so that could possibly explain why work rate has a effect on raising CA. I've noticed during my loanee match reports of high work rate players it says something along the line of "had a good game with his work rate on display" something to that nature. Could his high work rate cause a small uptick in his match rating compared to someone with lower work rate and as a result increase his PA faster since he is doing on average better in matches?

- pay attention to Scout reports regarding CONSISTENCY and IMPORTANT MATCHES. These CANNOT be tutored and they grow only slowly.

don't know if the information on guidetofootballmanager is realiable but it suggests that consistency raises with good form (playing inconsistent performers against poor opposition to help develop) and if this is true i believe important matches comes from consistent good ratings on big games.

Question here, are all first team matches considered equal? Does a player's match rating effect his CA growth? If so that could possibly explain why work rate has a effect on raising CA. I've noticed during my loanee match reports of high work rate players it says something along the line of "had a good game with his work rate on display" something to that nature. Could his high work rate cause a small uptick in his match rating compared to someone with lower work rate and as a result increase his PA faster since he is doing on average better in matches?

That is exactly how I believed it would work. However, so far, Maestro Ugo's results militate AGAINST Work Rate being important. However, mixs's work shows that it can be. If you look at the linked thread, there were important differences between how Maestro Ugo and mixs did their research, but I cannot judge whose results are "better".

don't know if the information on guidetofootballmanager is realiable but it suggests that consistency raises with good form (playing inconsistent performers against poor opposition to help develop) and if this is true i believe important matches comes from consistent good ratings on big games.

That would make sense, wouldn't it? I also remember reading somewhere that consistency increases with age also, so there could be two mechanisms for growth - "natural" growth due to age, which is perhaps NOT optimal, because players will be at their most consistent when they are PAST their CA peak (which is what happened in my test), and "trained" growth due to actual consistent performance.

That is exactly how I believed it would work. However, so far, Maestro Ugo's results militate AGAINST Work Rate being important. However, mixs's work shows that it can be. If you look at the linked thread, there were important differences between how Maestro Ugo and mixs did their research, but I cannot judge whose results are "better".

Work rate on the field I believe directly correlates with distance covered. A higher work rate player will run much more then someone with a lesser work rate in the same position. If we can prove that distance covered correlates with raising PA then our hypothesis of work rate raising PA would be proven true.

Recently I saw many up arrows next to the attributes of my striker (whose development I monitor the most) just after he had a great match (well, he has many great matches), at least one attribute reached a higher number, soon I looked at that screen again and the arrows were gone (it happens to me every now and then that the arrows seem quite ephemeral; previously I noticed free kicks getting better directly after good free kick shooting, but I don't know if that arrow was ephemeral). So it seems that event-based development raised many of his attributes - and Work rate (and Determination and Professionalism?) might make the difference between scoring 2 goals and scoring 4 goals (as they could make him work for goals which aren't needed for the win).

I've read on one page that a bad match consistency-wise means CA lowered by 10. One of my players started his career with consistency of 9, I believe, I haven't seen it improve, and he seems pretty consistent in delivering top-class performances.