Azərbaycan dili Bahasa Indonesia Bosanski Català Čeština Dansk Deutsch Eesti English Español Français Galego Hrvatski Italiano Latviešu Lietuvių Magyar Malti Mакедонски Nederlands Norsk Polski Português Português BR Românã Slovenčina Srpski Suomi Svenska Tiếng Việt Türkçe Ελληνικά Български Русский Українська Հայերեն ქართული ენა 中文
Subpage under development, new version coming soon!

Subject: Sokker helper - for player price estimation and prediction.

2014-03-10 08:23:58
I really think you're doing a great job, but finding the right estimation for every player with Some calculations is very very hard, if not impossible... Now my normal players their value estimations look pretty good, but my better players their value is 3 Mill too low in my opinion :)

You are really doing great ciury! But i'm getting the feeling the previous player estimations were better?
2014-03-10 14:22:23
D@ni to ciury
What i can see,judging by my team is that young players are really cheap in your predictor and i dont think they are "special" in any matter,what else i can say is that when i bought them,your predictor was pretty accurate about the prices, but now after few pops, some of them are predicted a half of previous value or even less.Really drastic change ... so now it works worse for my opinion altough older players still looks fine or even better at this moment of the season ;)

Keep up good work,this is a really great tool I really enjoy using it ;)
2014-03-10 18:59:42
exactly that.
young players are waaay off on price calc.

i think its best have the old formula for players below 20-21 years of age maybe
once players get through U21 people lose interest in them
2014-03-11 09:56:22

(edited)
2014-03-11 11:46:04
Thank everybody for the feedback. I am checking, and trying to improve. If I can't improve, I will switch back to the neural network estimation. Meanwhile, I increased the transactions which are part of the calculation to last 40 000, which might improve the situation...
2014-03-19 22:54:15
Now, depending on how scattered are the points which are used for learning, and how you train the black box, the output can differ. You can have a more smooth function, but in these case the "edge" cases will not be very well aproximated, or you can have a "pointy" function, and in this case based on bad values (far away points) the aproximation could differ.

Could you show me an example? Post chart or sth..? Iam curious about it.
2014-03-24 09:44:19
Nice Work ciury! :)
2014-07-03 08:49:43
up
2014-07-03 09:30:08
i suggest you to add the real ages (17,5 year, 18,1 ...) for entries in player transfert database
and using it as a criteria to calculate distance. Price are not the same (not only for young's) if a player is sell at the beginning of the season or the end.

Eg. with a 17,5y players sell on week 7 (7weeks /15 total weeks a year=0.5) then we using the diff age for comparison and distance like any other skills.
Maybe this diff criteria is too large (17,5y compare to 17,0y the -5 is maybe too much), i suggest you to adjust to div by 2 round to max (in that case : -5/2 round max=-3)
2014-07-03 09:31:55
Din0 to juytt
good idea, prices are bad now ;)
2014-07-03 09:40:39
juytt to Din0
eg. this one
Sahari Ilyas [29278252]
Age 16
Stamina 1
Pace 5
Technique 0
Passing 4
Keeper 0
Defender 8
Playmaker 4
Striker 4

Evaluate to 533 782 €

At the beginning of the season certainly ^^ but certainly not in week 14 ! :)
2014-07-03 10:49:58
7weeks /15 total weeks a year=0.5

You mean 8 weeks / 16 total weeks a year=0,5?? :)

2014-07-03 17:21:22
Message deleted

2014-07-04 00:05:06
forgot the week 0... 8/16 ;-)
2014-11-06 10:13:42
You may try to adapt some parameters because for this player, after adding one year is estimated as more expensive than his younger version, which is obviously wrong.
2015-01-18 23:39:55
any updates??