I find a bug from "Genie Scout 26 (GS26)", it can't correctly calculate the ratings. For example, I set the rating as
then I compare two players, Riad vs Leoni, their attributes are
A simple hand-calculation shows, (only need to compute the difference for each key-attribute, the first parenthesis is the difference for physical attributes, the second parenthesis is the difference for mental attributes, Riad is +, Leoni is -)
(1*60 -2*40 +60 -15 -35) + (3*44 -52 +44 +52 -2*56 +44) is GREATER than 0,
so Riad is better as a Center-Back, but GS26 says Leoni is better
The only way to explain this bug is GS26 using some hidden "Extra Curving", and it is fixed in their code, NOT editable. But we don't know how is it written in their code.
For example,
player-A has acceleration 15, pace 15, player-B has acceleration 14, pace 16.
All other attributes are the same. And if we set both acceleration and pace with 100 in the rating file of GS26. Overall, the two players should have the same ratings, namely, (15-14)*100+(15-16)*100=0.
But if GS26 uses some "Extra Curving" in its code, for example, it sets
acceleration with "Extra Curving" weight of 1.1, pace with "Extra Curving" weight of 0.9,
then we get
(15-14)*100*1.1+(15-16)*100*0.9 = 20
Then overall player-A is better. But we don't know how is this "Extra Curving" written in their code!!
OK, I found the source code from Github, although this source code is for GS2011, it confirms my idea in OP, which is they applied extra curving in their source code, here are screenshots, you can also go to Github to see it:
@Rain There is a big issue with GS, due to those hidden curving coefficients. If you have a programing background then you know what I am saying about their source-code.
This actually can explain why sometimes we feel "ridiculous" for the setting values in the rating file. For example,
If I claim in the rating file, I set dribbling = 50 for both center-back and winger, you must think I am stupid and ridiculous. It makes sense dribbling=50 for winger, but why set dribbling=50 for center-back?
This issue is due to those "Extra Curving Coefficients (ECC)", for example, the ECC on "dribbling" for winger is 0.9, for center-back is 0.1, then the actual buffed values are
50*0.9 = 45 for winger 50*0.1 = 5 for center-back
This makes more sense, doesn't it? It explains why we need to set such a crazy value (dribbling=50) for center-back, because the ECC for center-back is too low in their source-code!!!
Finally, go back to my title, there is a BIG BIG problem for the current GS app. The problem is: what if I want to buff some attribute (say, attribute-X) significantly? For example, I want to buff X by 30, here 30 is a true 30. But what if the ECC was fixed as 0.1 for X in their original code?
Because the maximum input for rating file for attribute X is 100, then actual maximum buff is
100*0.1=10
THERE IS NO WAY WE CAN BUFF attribute-X BY A VALUE OF 30! NEVER!!! This is the biggest problem in GS.
Falbravv said: Thanks for this work. This doesn't really matter in my games because i only use CA/PA / Prof / and some others tools.
The game only have 6-7 key attributes, it's easy to compare them directly in game.
But we all know this tool is edit every year by the same guy, who claim the same money, for something which doesn't evolve a centimeter in 10 years. Expand
Nozdormu said: @Rain Because the maximum input for rating file for attribute X is 100, then actual maximum buff is
100*0.1=10
THERE IS NO WAY WE CAN BUFF attribute-X BY A VALUE OF 30! NEVER!!! This is the biggest problem in GS. Expand Sure you can by reducing other weights by a factor of 0,3
I find a bug from "Genie Scout 26 (GS26)", it can't correctly calculate the ratings. For example, I set the rating as
then I compare two players, Riad vs Leoni, their attributes are
A simple hand-calculation shows, (only need to compute the difference for each key-attribute, the first parenthesis is the difference for physical attributes, the second parenthesis is the difference for mental attributes, Riad is +, Leoni is -)
(1*60 -2*40 +60 -15 -35) + (3*44 -52 +44 +52 -2*56 +44) is GREATER than 0,
so Riad is better as a Center-Back, but GS26 says Leoni is better
How is this possible?
========================================================================================
The only way to explain this bug is GS26 using some hidden "Extra Curving", and it is fixed in their code, NOT editable. But we don't know how is it written in their code.
For example,
player-A has acceleration 15, pace 15,
player-B has acceleration 14, pace 16.
All other attributes are the same. And if we set both acceleration and pace with 100 in the rating file of GS26. Overall, the two players should have the same ratings, namely, (15-14)*100+(15-16)*100=0.
But if GS26 uses some "Extra Curving" in its code, for example, it sets
acceleration with "Extra Curving" weight of 1.1,
pace with "Extra Curving" weight of 0.9,
then we get
(15-14)*100*1.1+(15-16)*100*0.9 = 20
Then overall player-A is better. But we don't know how is this "Extra Curving" written in their code!!
OK, I found the source code from Github, although this source code is for GS2011, it confirms my idea in OP, which is they applied extra curving in their source code, here are screenshots, you can also go to Github to see it:
https://github.com/kvarcg/ingamefmscout/blob/master/FMScout11/FMScout/Logic/ScoutContext.cs
Take Center-back as an example:
The coefficients for the extra curving is acceleration is 0.4, jump is 0.8
It also includes the strong foot and weak foot into their extra curving.
To my surprise, "Height" is a KEY attribute!!!, see the green-circled part,
if the height is greater than 190 cm, it adds 100 attributes
if the height is below 190 cm and above 180cm, it adds 50 attributes
It is completely chaotic! We need to re-write a completely new code for FM26 and the future...
@keithb I thought there was no issues with genie scout?
@Rain There is a big issue with GS, due to those hidden curving coefficients. If you have a programing background then you know what I am saying about their source-code.
This actually can explain why sometimes we feel "ridiculous" for the setting values in the rating file. For example,
If I claim in the rating file, I set dribbling = 50 for both center-back and winger, you must think I am stupid and ridiculous. It makes sense dribbling=50 for winger, but why set dribbling=50 for center-back?
This issue is due to those "Extra Curving Coefficients (ECC)", for example, the ECC on "dribbling" for winger is 0.9, for center-back is 0.1, then the actual buffed values are
50*0.9 = 45 for winger
50*0.1 = 5 for center-back
This makes more sense, doesn't it? It explains why we need to set such a crazy value (dribbling=50) for center-back, because the ECC for center-back is too low in their source-code!!!
Finally, go back to my title, there is a BIG BIG problem for the current GS app. The problem is:
what if I want to buff some attribute (say, attribute-X) significantly? For example, I want to buff X by 30, here 30 is a true 30. But what if the ECC was fixed as 0.1 for X in their original code?
Because the maximum input for rating file for attribute X is 100, then actual maximum buff is
100*0.1=10
THERE IS NO WAY WE CAN BUFF attribute-X BY A VALUE OF 30! NEVER!!!
This is the biggest problem in GS.
Thanks for this work.
This doesn't really matter in my games because i only use CA/PA / Prof / and some others tools.
The game only have 6-7 key attributes, it's easy to compare them directly in game.
But we all know this tool is edit every year by the same guy, who claim the same money, for something which doesn't evolve a centimeter in 10 years.
Falbravv said: Thanks for this work.
This doesn't really matter in my games because i only use CA/PA / Prof / and some others tools.
The game only have 6-7 key attributes, it's easy to compare them directly in game.
But we all know this tool is edit every year by the same guy, who claim the same money, for something which doesn't evolve a centimeter in 10 years.
Yes, exactly!
Nozdormu said: @Rain
Because the maximum input for rating file for attribute X is 100, then actual maximum buff is
100*0.1=10
THERE IS NO WAY WE CAN BUFF attribute-X BY A VALUE OF 30! NEVER!!!
This is the biggest problem in GS.
Sure you can by reducing other weights by a factor of 0,3