So we looked at the data..

[Trump]

And what we found surprised us
For whatever reason, they won't tell us the basic mechanics of this game. So I decided to investigate some of them myself.
Here's what I've found out in regards to regen quality (immediate effects only):

Youth Importance = NO EFFECT
Youth Coaching = Changing from '20' to '1' reduces PA by ~40%, CA appears unaffected. '10' appears equal to '20', but this requires further investigation.
Youth Facilities = NO EFFECT
Youth Recruitment = Changing from '20' to '1' reduces both CA & PA by ~25%. Changing from '20' to '10' reduces CA & PA by ~10%. Effect is also influenced by youth recruitment of other clubs, exact amount TBD.
Club Reputation = Changing from '10000' to '1' reduces CA by ~10%, PA appears unaffected. Changing from '10000' to '5000' reduces CA by ~4%.
Training Facilities = NO EFFECT
Nation Reputation = NO EFFECT
Nation Youth Rating = Changing from '200' to '1' reduces CA & PA by ~25%
Game Importance = NO EFFECT
City Attraction = NO EFFECT
City Inhabitants Range = NO EFFECT

Based Nation or Unique Division = One of these appears to have a significant effect. Possible hidden nation variable.
City competition (clubs existing in same city with higher youth recruitment & reputation) = Modest reduction of CA (range of perhaps 5-10%?), no effect on PA.

Notes:

The above was tested on FC Bayern, edited into a 'perfect' club in a 'perfect' region, 'perfect' country, etc.

'CA' and 'PA' here refers to the cumulative total of one youth intake (16 regens). This is because the highest individual regen PA can be highly variable - you might get one 190 PA regen, or several 160 PA regens, but the cumulative total PA of the intake only has a random variation of about ~10%. A more accurate method may be to compare the median PA of the intake, but comparing the cumulative total seems to do just as well. For instance in one of my tests for changing 'Nation Youth Rating' from '20' to '1', median PA decreased by 28.6% and cumulative PA decreased by 29%.

I'm sure most people here already know this, but for those who don't, youth and training facilities only affect CA increase over time, not the initial CA/PA.

The caveat to my testing is that it only shows the difference between the absolute best and the absolute worst. The effects could be curved instead of linear when plotted on a graph, but my testing does give a general idea at least as to what does and does not effect regen quality.

To sum up:

Nation Youth Rating = pa/ca factor (25% effect)
Youth Recruitment = pa/ca factor (25% effect)
Youth Coaching = pa factor (40% effect)
Club Reputation = ca factor (10% effect)

 

[crusadertsar]

Is this a joke?

[Trump]

Just now, crusadertsar said:

Is this a joke?

No, why?

[Kawee]

2 hours ago, crusadertsar said:

Is this a joke?

The thread or the mechanics?

[Keyzer Soze]

So, a guy called Trump started a thread about the hidden secrets of the mechanics of the game.... 

Fake news? 

Now, serious, those results were just from one save and one youth intake? Or this are the average results from various saves/youth intakes? 

 

[Bigpole]

I could agree with some of these points but I don't think Youth Recruitment has an effect on CA/PA, because it is the club scouting range. And scouting range itself does not have any effect on ability.

[herne79]

7 hours ago, Trump said:

For whatever reason, they won't tell us the basic mechanics of this game.

There's no big secret about what has an influence on the quality of newgens.

Simplistically: The scale of talent being produced by a nation is determined by the nation's Youth Rating.  In other words you're more likely to produce a greater quantity of decent newgens in Brazil than Togo.  Clubs with a higher Youth Recruitment and Reputation then become more likely to pick up the better newgens.  So expect Bayen to get more quality newgens than Ingolstadt.  The quality of those club's Youth Facilities and Junior Coaching then contribute and help determine the CA and PA of those newgens when they appear on Youth Intake Day.  Your Youth Coaches and Head of Youth Development can also influence the quality of newgens.

Two other things should also be noted: 1) nothing is guaranteed, we can only influence the likelihood of something happening.  2) There remains a randomness factor, thus there is always a chance (albeit small) that the next 200 PA newgen will appear in Togo rather than Brazil, or that you'll get a great intake one year but a rubbish one the next at the same club with no changes.

What SI haven't shared - and nor should they - are the exact equations of all of these calculations.  New player generation is very closely monitored to try to ensure things are realistic as possible and to maintain a decent spread of new players coming into the database at all levels of the game.  Literally thousands of soak tests are carried out internally to ensure this - not just one test at one club which (as good as the intention is) I'm afraid is nowhere near good enough to assess how things are working.

You also mention Youth Importance.  This the priority your Board places on Youth Development.  So it'll be easier to ask the Board to improve your Youth set up if they have a high Youth Importance but don't expect much willingness to invest if they have the opposite.  This Youth Importance can have an indirect influence on newgen quality at your club over time.

So TL;DR, what it all boils down to is: improve things (Facilities, staff, winning matches) that are within your control to give you the best chance of getting better quality youth.  And always remember that if you're a club in Togo, you can have the best staff, facilities and reputation in the world but you'll still be in Togo.

I will however give @Seb Wassell a nudge here as he's the guy who looks after all of this and is always willing to look at data.  Personally I feel the posted "results" are fundamentally flawed for two reasons:

1) Sample size.  A single save using a single club is nowhere near large enough to start to draw conclusions as there are too many variables involved.  Run the test 100 times at the same club and you'll see different results.  Run the test 100 times at 100 different clubs and you'll get a different set of results again.

2) Methodology.  You've used the Editor to remove (or negate) some variables. I understand why you've done that however in doing so you removed parts which everyone else would use.  Thus any results you now show can only be used in relation to your specific edited database, which nobody else uses.

[TheJanitor]

[Seb Wassell]

10 hours ago, Trump said:

For whatever reason, they won't tell us the basic mechanics of this game.

Youth Recruitment
The Club’s recruiting of Juniors, those of home and foreign origin.
Youth Facilities
The facilities available for Juniors.
Junior Coaching
The quality and quantity of Junior Coaches at the Club. 
Nation Youth Rating
The potential maximum quality and quantity of Juniors produced in that Nation.
Game Importance
How important football is considered to be in that Nation.
Producing Newgens
The Clubs with the best Youth Recruitment will generally pick up the best Junior talent from that Nation first, the scale of that talent being determined by the Nation Youth Rating and Game Importance. The lower the Youth Recruitment the further down the pecking order a club will find itself. Two clubs with identical Youth Recruitment will be sorted by Club Reputation. Being lower down this pecking order does not mean quality Newgens cannot be produced, it simply lowers the chances.
Youth Facilities and Junior Coaching then simulate and determine how that Junior progresses in the Club’s Junior system until a Newgen is produced and appears in game. It is at this point that the Current Ability (CA) and Potential Ability (PA) of the youth players are decided.
The above factors all contribute to both CA and PA.
 

[Trump]

16 hours ago, Keyzer Soze said:

So, a guy called Trump started a thread about the hidden secrets of the mechanics of the game.... 

Fake news? 

Now, serious, those results were just from one save and one youth intake? Or this are the average results from various saves/youth intakes? 

 

Averaged, but small sample size. The only one I'm not too confident about is club reputation has a 10% effect on CA, it could be just random variation. I plan on doing more testing.

As to what has been posted by the forum mod and SI researcher above, the first thing I have to say is that what got me started on this is that the information posted in the forums over the years by SI staff and moderators has not only been lacking/vague, but (from memory) they have even contradicted each other.
 

Quote

I could agree with some of these points but I don't think Youth Recruitment has an effect on CA/PA, because it is the club scouting range. And scouting range itself does not have any effect on ability.

Youth recruitment has a strong effect on CA/PA. On average it seems around 25% on both, or ~20-30% with the random variation. I don't know what 'youth recruitment' interacts with, but I do not think it relates to the club scouting range.

Bonus note: It seems that players entering from an affiliate club into the youth intake, are affected by their nationality's youth rating. Or the aspects of the affiliate club perhaps. I discovered this when in my testing with FC Bayern, I noticed that the American affiliate regen that would always come through, it typically had one of the lowest PA and seemed to drag down the total PA of the group as well.
 

Quote

The quality of those club's Youth Facilities and Junior Coaching then contribute and help determine the CA and PA of those newgens when they appear on Youth Intake Day.

Based on my testing, part of this is false. Youth facilities do not affect regen starting CA/PA at all. Perhaps there is a small indirect effect where Youth facilities improve club reputation, thus increasing regen CA, but given that the difference between '1' and '10000' club reputation is ~10% CA only, the effect would be negligible.
 

Quote

There remains a randomness factor, thus there is always a chance (albeit small) that the next 200 PA newgen will appear in Togo rather than Brazil, or that you'll get a great intake one year but a rubbish one the next at the same club with no changes.

This is sort of correct, but also sort of wrong. There is a randomness factor, and in my testing it appears to make about a 10% difference one way or the other, in regards to total PA or median PA. What you imply that is wrong is that any club in any nation can produce a 200 PA regen. Technically this is possible I assume, but not practically possible. For instance just by setting 'Youth Recruitment' from '20' to '1' in an otherwise 'perfect' club in a perfect country, typically the highest PA regen you'll get is ~120PA. Suppose only 1 in 10 intakes then has a 150 PA regen, what are the chances a 190 PA regen will be produced? Perhaps a 1 in 10,000 chance?

What throws some confusion on the matter is that many of us have had the experience of getting that wonderful regen from Saint Pierre and Miquelon - but it's important to note here that you most likely didn't get that player from a Saint Pierre and Miquelon club, but from either non-club generation or generation at your own club using only the nation youth rating as a point of difference. Remember, there is only a 25% CA/PA difference between nation youth rating '1' and '200'.

Quote

Your Youth Coaches and Head of Youth Development can also influence the quality of newgens.

I highly doubt that any staff member affects regen CA/PA. HoYD does have an effect on the personality values of the regens. Youth coaches? I don't think they affect regens in any way at all, but I'm happy to be corrected.
 

Quote

So TL;DR, what it all boils down to is: improve things (Facilities, staff, winning matches) that are within your control to give you the best chance of getting better quality youth.

I think my findings provide some important insights.

For instance, if you are in the Conference North, what you might want to do is focus on getting higher CA players first up. In that case, it would be important for you to know that 'youth recruitment' increase CA by 25% while 'junior coaching' increases CA by 0%. Conversely, suppose you're at an elite club or you can afford to take the low CA now and reap the increased PA later, you'd want to know that 'junior coaching' increases PA by 40%, while youth recruitment only increases PA by 25%. The order of what you improve could make quite a significant difference.

A lot of people are also operating on false assumptions or even false information being told to them. I myself used to believe 'junior coaching' and 'youth facilities' had each other's roles. It's not just about knowing how to develop one's own players either, but knowing where to look for players. Looking in the editor, one might think it's a good idea to keep an eye on Egypt with it's 138 youth rating. But no Egyptian club has 'junior coaching' above 13. One might also discount Czech Republic because their 'game importance' is less than 'very important', or have no clue what impact a club's reputation has (i.e. 'Should I keep an eye on Ivory Coast's ASEC Mimosas, given their reputation is only 4750?'), etc.
 

Quote

1) Sample size.  A single save using a single club is nowhere near large enough to start to draw conclusions as there are too many variables involved.  Run the test 100 times at the same club and you'll see different results.  Run the test 100 times at 100 different clubs and you'll get a different set of results again.

You say if I'll see different results if I run the test at the same club multiple times. That's not what I've found. There is some random variation, around ~10% either way. For instance, testing the entirely 'perfect' club, I got a low of 1810 PA and a high of 2159 PA. When I set the 'Nation Youth Rating' to 1, my first result was 1405 PA. The median PA went from 110-128 PA to 85 PA. Clearly there is a discernable difference. The chances of getting 1405 total PA at a 'perfect' club just through random chance is probably something like 1 in 10,000 I'd guess.
 

Quote

2) Methodology.  You've used the Editor to remove (or negate) some variables. I understand why you've done that however in doing so you removed parts which everyone else would use.  Thus any results you now show can only be used in relation to your specific edited database, which nobody else uses.

I chose to use FC Bayern instead of creating an entirely new club/nation so that no hidden variables would come into play. I also only tested the difference of one variable at a time, and used the maximum difference to make the effect as clear as possible. I suspect that the variables, and perhaps their interactions with each other, do not result in linear differences, but what we can say for sure is that 'junior coaching' does not affect CA at all (except perhaps it's small indirect effect via club reputation), which is a big deal!
 

Quote

Game Importance = no effect on newgens itself, but determines potential number (by small amount). When you have two similar population nations, one like Canada where hockey is main discipline and Poland where football is one of the main disciplines - you know which one has better chance of quality newgens.

From my testing, game importance did not change the number of regens (always 16) nor CA/PA. I don't know if it effects the number of non-club regens produced, or their quality, but a theory I have read is that game importance effects how many staff are generated in that country.
 

Quote

Game Importance
How important football is considered to be in that Nation.
Producing Newgens
The Clubs with the best Youth Recruitment will generally pick up the best Junior talent from that Nation first, the scale of that talent being determined by the Nation Youth Rating and Game Importance. The lower the Youth Recruitment the further down the pecking order a club will find itself. Two clubs with identical Youth Recruitment will be sorted by Club Reputation. Being lower down this pecking order does not mean quality Newgens cannot be produced, it simply lowers the chances.
Youth Facilities and Junior Coaching then simulate and determine how that Junior progresses in the Club’s Junior system until a Newgen is produced and appears in game. It is at this point that the Current Ability (CA) and Potential Ability (PA) of the youth players are decided.
The above factors all contribute to both CA and PA.

My testing disagrees with two claims here.

FC Bayern with 20 youth recruitment and 10000 club reputation produced the same amount of total CA/PA whether Germany's 'game importance' was set to 'very important' or 'completely useless'.

Are you perhaps saying that 'game importance' determines how many clubs can get high quality regens? For instance a lesser game importance may reduce the possible number of great regen producing clubs from 10 to 2?

The second thing is that setting 'youth facilities' from '20' to '1' did not effect starting CA/PA at all.

As you can see from my attachments:

'Perfect':

877 CA
1927 PA


'Completely Useless' game importance:

902 CA
1956 PA


'Youth Facilities' set to '1':

870 CA
2087 PA

[Trump]

Another example of 'completely useless':

935 CA
2107 PA


Highest individual PA of 164 also suggests an unaltered distribution

[Trump]

Youth Recruitment 10

1739 PA
849 CA

1832 PA
842 CA

1711 PA
815 CA

1760 PA average (~11% decrease)
835 CA average (~10% decrease)

The effect is close to, but not exactly linear.

Youth Recruitment 10 + Club Reputation 5000

1800 PA
819 CA

1851 PA
812 CA

1873 PA
819 CA

1841 PA average (~7% decrease)
817 CA average (~11.5% decrease)

Although there is just a 1.5% decrease in CA compared to above, CA usually follows PA in ratio, so to see CA decrease while PA has increased by 4%, it suggests that making the club reputation 5000 probably has about the -5% CA effect we would expect from the linear extrapolation of 10000 rep to 1 rep resulting in -10% CA.

Notably there is no evident interaction between the factors here. They seem to behave together as they would separately.
 

Next I decided to try and knock out three birds with one stone by setting Youth Recruitment '10' with Club Reputation '5000'. The three things this tells us is:

1. The effect of Club Reputation '5000'
2. The effect of interaction, if any, between Youth Recruitment and Club Reputation
3. The effect of higher youth recruitment/reputation clubs in the same city on regen quality

For the third, I edited the City to include several of the top German clubs and have an inhabitants range of just 5,001-10,000 in case that affects some kind of regen 'pool'.

Youth Recruitment 10 + Club Reputation 5000 + Competing teams in same city w/ higher club reputation & youth recruitment

1777 PA
720 CA

1760 PA
784 CA

1769 PA
777 CA

1768 PA (~10.7% decrease)
760 CA average (~17.7% decrease)

It appears increased city competition is having a non-negligible effect on CA at least. The effect is modest but significant. PA is within the bounds of random difference; more testing can be done to confirm.

I will add these findings to the OP.

[herne79]

@Trump If you believe your analysis is showing things working in ways other than described, the best thing you can do is upload your raw data and methodology for @Seb Wassell to take a look.  Seb will then be able to tell if there is an issue or not.

[Seb Wassell]

13 hours ago, Trump said:

As to what has been posted by the forum mod and SI researcher above, the first thing I have to say is that what got me started on this is that the information posted in the forums over the years by SI staff and moderators has not only been lacking/vague, but (from memory) they have even contradicted each other.

I am responsible for this area of the game. This is how it works. The conclusions you have drawn above are speculative. I'd be happy to discuss this with you but we'll need robust, reproducible data from a large and diverse sample size in order to compare with our internal analysis.

[Keyzer Soze]

@Trump

Thanks for the reply. 

I got say, I admire your research work. It's always nice so see someone taking time to try to understand how the game works underneath the hood. 

But, a little bit of advice... Next time don't start a thread with a attacking mentality. Simply because, you will face a parked the bus opponents and they will trash you in deadly counter attacks. 

Start the thread with a more balanced mentality, and don't rush into conclusion so much. 

 

[sporadicsmiles]

You definitely do not have enough data to draw any conclusions from here. Youth intakes are inherently noisy. Sometimes you get a wonderful class with a lot of great players. Sometimes you get a crappy class without a player who will make it at your club. This sort of randomness requires that you do more than 1 youth intake for each of the things you test. Otherwise, there is no statistical significance to your data. Indeed, it does not even seem you account for this at all, based on how I understood your post (you state in the OP it is averaged over a single youth intake). 

If you really want to make this result robust, you first need to quantify the variance in a single data point. I'm not actually saying you are wrong, and I like to see research. To do this properly, I'd suggest at least 5 repeats for each setting you used (pain in the ass, I know, I work in science and have to do this as a routine). If you have 16 youth players per intake, that 94 players. You can then plot a histogram for those players. This allows you to see (a) if you have a normal distribution and if you do, (b) have direct access to the variance. It also gives you a visual representation of your data (I would not be surprised if this is what is done during testing for FM). A figure should always be preferred to a table full of numbers. Tables are nearly useless unless you want to share data so others can use it. The upshot of this is you will be able to talk about your data with a great deal of confidence, because you will really know what it says.

More than happy to help you out with any of the statistical side of things, if you would like. I'm happy to encourage people exploring data of any kind.

[Trump]

10 hours ago, Seb Wassell said:

I am responsible for this area of the game. This is how it works. The conclusions you have drawn above are speculative. I'd be happy to discuss this with you but we'll need robust, reproducible data from a large and diverse sample size in order to compare with our internal analysis. 

I'm happy to share my method so anyone can verify, but to nip this in the bud I've decided to do a new test with just one edit of setting England to 'completely useless'. The reason why I made FC Bayern a 'perfect' club was because I wanted to measure the full effect of the options, which isn't necessary here. This method also addresses the points that maybe my results don't apply for the typical user, and that different nations/leagues/teams produce different results due to hidden variables.

My settings for the game:



To save on time, I've started unemployed and holidayed until 27th Febuary. I've then added a human manager of West Ham with optimal attributes and holidayed until 2nd March (youth intake day is 1st March).

West Ham - Normal

1756 PA
773 CA

1575 PA
623 CA

1800 PA
771 CA

1710 PA average
722 CA average

West Ham - 'Completely Useless'

1609 PA
698 CA

1840 PA
793 CA

1783 PA
746 CA

1744 PA average
746 CA average

So it's clear that the 'completely useless' setting has no effect on regen quality, but there's more to be gleaned from these tests.

The average of those 6 totals are:

1727 PA
734 CA

The average I have for a 'perfect' FC Bayern (many samples, some tainted by a few foreign nation regens, which I'll fix later) is:

1964 PA
906 CA

Given my findings, that are summarized in the OP, I would expect the following for West Ham compared to 'perfect' FC Bayern:

6800 rep (-2.7% CA)
15 youth recruitment (-4% CA & PA)
12 youth coaching (-13.4% PA)
nation youth rating 120 (-8.4% CA & PA)
City competition from 3 teams w/ higher reputation & youth recruitment (-5% CA)

totals: -25.8% PA, -20.1% CA

The real difference is:

-11.3% PA, -19.1% CA

This stumped me for a while. Is the curve for the variables different from what I have assumed? Are there really hidden variables I cannot account for? Well thankfully it turns out neither is the case. Turns out that setting 'youth coaching' to 10 produces similar results to youth coaching 20. I don't know why this is the case, but it does seem to fully explain the difference, for if we remove the -13.4% PA from the youth coaching effect, it becomes -12.4% PA, -20.1% CA which is very close to the actual averaged difference of -11.3% PA, -19.1% CA.

For those confused about % figures in the above calculations, I've assumed that the variables all follow the same curve, where '1' is 25% worse than '20', but '10' is only 10% worse than '20', and '15' is only 4% worse than '20', etc.
 

Quote

To do this properly, I'd suggest at least 5 repeats for each setting you used

I do at least 3 tests for each setting, some I've done more. My conclusions are based on the averages of these tests (and whatever else I notice). I've tested 'perfect' FC Bayern 12 times, but I will need to redo this data as the foreign regens that come through sometimes will have skewed the results downwards. I think 3 tests is good enough to get a general idea of what the effect is, and 10 is ideal.

I've thought about making some graphs, particularly the curves of the variables once I flesh them out fully. I might do it if I have the time.

[Guest]

Training & Youth Facilities isn't responsible for PA of a player, but helping them out to reach it. What we are looking here?

[phnompenhandy]

9 minutes ago, Cadoni said:

Training & Youth Facilities isn't responsible for PA of a player, but helping them out to reach it. What we are looking here?

Not them.

[sporadicsmiles]

19 hours ago, Trump said:

So it's clear that the 'completely useless' setting has no effect on regen quality, but there's more to be gleaned from these tests.

Your conclusion is based at looking at the averages of a very small sample size, and I do not think it is correct. You need to look at the histograms here, because they are much more informative, potentially, than the average numbers. I did this, briefly, and there are small differences that may or may not be significant (again, there is not enough data to say anything with certainty, especially since we seem to be dealing with small changes). I say there is not enough data because you do not have complete histograms, which means you do not have access to the true mean values yet. You have to be quite careful when you are talking about mean values that have a large variance. The larger the variance, the more data points you need to get close to the true mean (and that assumes you have a normal distribution, otherwise the mean does not really mean much - stat pun!)

What it seems to me is that there may be a shift of the distribution towards lower values when you have completely useless. However, like I said, I cannot back this up because the frequency counts are really low. I find this kinda stuff very interesting though. I would help you out, but currently I am running the beta, so I cannot actually use the editor until the updates go live. Which sucks in general.

Also, do we know how the game itself changes things you have edited? For example, will the game become "very important" again, rather than completely useless, before the regen date? I actually have no idea about this, and it would be very hard to regulate I guess. 

[Trump]

11 hours ago, Cadoni said:

Training & Youth Facilities isn't responsible for PA of a player, but helping them out to reach it. What we are looking here?

What has an immediate effect on regen quality
 

4 hours ago, sporadicsmiles said:

What it seems to me is that there may be a shift of the distribution towards lower values when you have completely useless. However, like I said, I cannot back this up because the frequency counts are really low.

Well I look at it this way: If the effect is so negligible that it's not apparent from 3.. 6.. or even 10 intakes, then it's not even worth identifying. I don't know about other people, but I've rarely played beyond a few years when I play.

Now, you posit that there is a shift of the distribution towards lower values, but my testing shows that the average remains the same. The average would only be an inaccurate judge in this case if along with the lower-shifted distribution, there were more high outliers to offset it - but if you look at my results, the highest outliers for 'completely useless' were 135, 138 and 138. For 'very important' the results were 134, 143 and 146. What about if we look at medians, is there a difference there? 105, 116 and 117 for 'completely useless' vs. 104, 109 and 114 for 'very important'. To add to the West Ham results, I can also add the samples from my FC Bayern testing, so we have a sizable sample size now showing no observable effect of 'completely useless'. Also, another reason to doubt that higher outliers would be created to offset the shift in the average, is that it wouldn't contribute to the realism of the game - why would a completely useless nation be more likely to produce top-tier regens in the game?

I've thought about 'game importance' might actually be doing. I noticed during my testing that 'very important' seemed to produce a bunch of non-club low quality regens every second intake test or so. It could just be coincidental, but it would make sense as a mechanic - more relevant nations (i.e. England) produce a few more low/mid-tier players (and staff?) to sort of fill out the game for immersion, and less relevant nations (i.e. Afghanistan) don't get this to save on performance. Another thing it could be doing is that perhaps it doesn't effect the human manager's regens, but maybe it does effect the AI's regens. This kind of makes sense if you consider a human player managing in India - if that human player gets East Bengal to championship-tier quality within a few years, do you think it's realistic or rewarding for the other clubs in the league to get up to a similar standard also, simply because one team keeps winning all the cups? I plan on looking more into this, but from my cursory glances I didn't notice a change in regen quality for the AI.

You say that 'completely useless' might shift back to its default setting within a year. It's possible, but I don't think it's likely. I think if it did happen, it's most likely to happen as a bug, for if the setting is dependent on other factors, why have it as an attribute in the first place? If, for instance, 'game importance' is ultimately determined by a combination of 'reputation' attributes, then why have a separate setting for it - or how would even changing it effect anything? The number of staff and/or non-club regens explanation seems much more likely to me.

 

[Trump]

A separate post for this.

I have been attempting to further confirm my values of the factors by testing comparing a club in Romania Div. 2 (unedited) to my West Ham results (unedited). If my factor values are correct, then I should be able to predict accurately the difference in regen CA/PA by calculating with the factor values.

Unfortunately, it's not lining up at all. I suspect it is because I am not accounting for other factors, in particular league reputation, which I've completely forgotten about. I'm thinking perhaps league reputation, because the difference between West Ham (premier league - '172') and FC Bayern 'perfect' ('200') would be easy to overlook, whereas the difference between West Ham ('172') and Romania Div 2 ('86') is significant. That said, there are probably some other factors at play, or I have to reassess the curve/interactions of my factor values.

[sporadicsmiles]

5 hours ago, Trump said:

If the effect is so negligible that it's not apparent from 3.. 6.. or even 10 intakes, then it's not even worth identifying.

 

5 hours ago, Trump said:

Now, you posit that there is a shift of the distribution towards lower values, but my testing shows that the average remains the same.

It is if you want to talk about the mean. You are currently giving a population mean, which is not the same as the sample mean. Population means are only as good as the amount of data you have. Besides, at the moment everything you are doing is assuming a standard distribution, which may not be the case. What if the effect you look at is only shifting a small part of the distribution to lower values? You then have two distributions, each of which has a mean. The data you show also has an enormous standard deviation. Statistically, you could not say an average of, say, 40 and 60 were significantly different. This is obviously extreme, but if the effect you are looking for is small (and it clearly is), then it is hard to know for sure. Whether it is worth while is another matter, you can judge that I guess. There are likely much more important factors. By the way, I am not trying to discourage you or criticize you here, I think you are doing something interesting and trying to understand how things work.

5 hours ago, Trump said:

You say that 'completely useless' might shift back to its default setting within a year. It's possible, but I don't think it's likely.

Honestly, I have no idea if it does or not. I am a scientist, so I am just good at spotting potential issues in methods (and I like your methods for this). If you have the in game editor, I think another way to do these experiments (with more control) would be to have exactly the same initial game setup, and then using the in game editor to change settings of the club you are studying just prior to the regen date (another thing, which I do not know, the regens are all generated on regen date based only on the state of the club/country at that date?), then holidaying to get the regens. That would give you much tighter control over your parameters so you can deflect annoying questions from bothersome people like myself!

[Snorks]

On 25/02/2019 at 14:40, Trump said:

A separate post for this.

I have been attempting to further confirm my values of the factors by testing comparing a club in Romania Div. 2 (unedited) to my West Ham results (unedited). If my factor values are correct, then I should be able to predict accurately the difference in regen CA/PA by calculating with the factor values.

Unfortunately, it's not lining up at all. I suspect it is because I am not accounting for other factors, in particular league reputation, which I've completely forgotten about. I'm thinking perhaps league reputation, because the difference between West Ham (premier league - '172') and FC Bayern 'perfect' ('200') would be easy to overlook, whereas the difference between West Ham ('172') and Romania Div 2 ('86') is significant. That said, there are probably some other factors at play, or I have to reassess the curve/interactions of my factor values.

Don't want to appear flippant or anything - but am I the only one who pictured Doctor Who saying this?

I really admire your dedication Mr Trump, but seriously, just play the game.

[Trump]

I've been doing more testing, and I've discovered something of significance importance.

Either the based nation or unique division has a significant effect on regen quality.

If you put the top English teams into Romania Div 1, the best regen of the lot will only be around ~150 PA.  In the English Premier Division, this would be ~190 PA. The same thing happens if you put Romanian teams into the English Premier Division -  their top regen goes from 150 PA to 190 PA. It's not just the top regens that are changed, it's the average/median of the youth intake too.

Factors that have been controlled for (that is, made equal):

Nation youth rating, nation reputation, economic factor, nation attendances, nation ranking points, nation region, nation max youth age, nation state of development, nation federation financial power, nation tactical profile and preferred formations, division reputation, club continental cup nation set to the nation club is sent to

Other factors I have tested and have found to make no difference:

City, City attributes, Stadium city, City nation

The city (or stadium city?) of the team determines a regen's nation. I found that whether English or Romanian regen nationality, the results were the same. Therefore it's either the nation the club is based in, or the unique division itself, that is the factor, not nationality itself. I suspect it is the unique division itself rather than the nation, because I've found an unexpected difference between the Premier League and English League 1 too (though this could be due to division level or city competition for recruitment).

There are some things I haven't copied over, such as competition records, but I think this is unlikely to make a difference.

This all means that youth rating is not the only national factor we have to take into account. We cannot merely presume that because Ivory Coast has a high youth rating, its domestic clubs can produce good regens. It seems that each division must be tested individually, as there is a hidden factor at work. Perhaps by testing each division I can create a ratio table by which to multiply, for there are only a few dozen divisions included by default in the game, and it wouldn't change much from year-to-year.

[herne79]

So there is a greater chance of getting better quality newgens in England than Romania?  Shock.

I'm sorry Trump but with your other closed thread and the language you used in it I'm no longer willing to pander to your "results" which are nothing more than subjective hyperbole which you're trying to dress up as fact and thus potentially mislead other readers of the forum.

There is no discussion here, just subjective opinion and incorrect conclusions because you simply don't know all of the variables involved, thus you make many (incorrect) assumptions.  Perhaps if you hadn't accused SI of being racist, discriminatory and regressive (or apologised for doing so) I'd be more willing to leave this thread open.

You've been asked multiple times to upload your data so SI can take a look.  If you are going to do that, please start a new thread in the Training and Medical Centre forum and upload your data.  And if you do, keep it constructive without any further abuse of SI and it's staff.