Subject: Sokker Asistente

Strange. I've sent you an sk-mail

I have player with really top talent...

terrion look it... mín talent can to be so low?

http://prntscr.com/vmduda
http://prntscr.com/vmdv2i
http://prntscr.com/vmdvcj
http://prntscr.com/vmdvml
http://prntscr.com/vmdvyq

Your players are great, I didn't have such good players to test the accuracy of the minimum talent. It should be close to 3 but, again, it was just an estimation

Thank you...

3.8 / 3.1 talent is not a surprise. Im using stats and for many players even with lot of data (weeks) i see this min/max ratio.

I'm thinking on giving the chance to set the talent by hand. This way, when you have too much data, you will be able to correct that ugly numbers

I'm thinking why you don't remove the min and max value of the talent and set One average number, that will be calculate like -> min+max = average.

I think your talent estimation is accurate enough and there is no way of avoiding such differences. No way, because of following 2 factors:
1- Change of assistants or change of first coach
2- Skill level 16 can be different than skill level 16. I'll explain what I mean by an example: If my head coach has 16 e.g. in pass and in playmaking, there can be differences between the speed of training in pass and playmaking, because of decimals. If pass training happens faster, skill 16 (of coach) may be 16,9 and if playmaking training happens slower, playmaking skill of coach may be 16,0.
According to data of training pass, the talent estimation based on pass will be better than the estimation based on playmaking.

My opinion: Leave it as it is! :)

> I'm thinking why you don't remove the min and max value of the talent and set One average number, that will be calculate like -> min+max = average

That's how it works to calculate the training points estimation, but there's a big difference in a 4 talent if it's due to a [3.9, 4.1] range or due to a [3, 5], so anyone can decide how to interpret it
(edited)

> My opinion: Leave it as it is! :)

It would be optional. Some friend like to make me change the talent of his players to adjust the prediction. I think it can be a good tool for everybody

A while ago I experienced the same like jezo and juytt with my very talented goalkeepers. Since then I dont rely on the talent-estimation anymore, BUT this is NOT a critic to Terrion since it's honestly almost impossible to build a talent-predictor that is really accurate in all situations given how many assumptions have to be made to get to a result.

Based on what I understand the estimation depends on:
- level of coach and assistants
- training-points per skill and week (considering the difference between friendly, league, and NT-match)
- exact base-level when player starts his training (e.g. 16.0 vs 16.9)
- a certain coefficient which Terrion assumes is always 1.094. But it is eventually just an estimation based on his past players' skill developments and if we could analyse all sokker-players that exist, this coefficient could be different.

A few months ago I created my own talent-predictor based on the excel-data provided by tsolias (though he actually doesn't know about that ;-)). The predictions were slightly different from those of Terrion, partly because the underlying coefficient is not the same. The difference between Terrion-predictions and my Tsolias-based predictions were around 0.5 talent-points even with players that I or my manager-colleague trained for over 30 weeks. In my view this reflects how much assumptions and uncertainty is in play.

My opinion: If it doesnt take much effort, consider adding the option to set the talent by hand (so it's the manager's choice if he/ she wants to use the calculated or the manual value). Yet I really wouldnt make too much effort since the entire talent-calculation might change completely by the release of the new Sokker-version in about 2 seasons from now.

@Terrion: I appreciate your efforts and your good work and I hope the new devs will appreciate your (and also my) work one day as well...

:)

I just sent before some minutes a sk-mail to Terrion and I informed him about my method of talent estimation. Both methods are very similar. Imho the biggest difference is on the result of the talent estimation (not in the way of the estimation)



My estimation is stricter - that's why you had different results.
(It is stricter, in order to have the same talent, that was estimated on junior school. Eventhough, also my method is not entirely accurate, it's not possible)

I wouldn't compare the talent of players calculated with one method with the talent of other players calculated with another method, since there is no public theory about how you can calculate it. I use my own assumptions, which of course may be improved

I also assume that pace and striker are a 21% slower than the rest of skills, althought I'm not sure about the right numbers

Pace could be a 20% (I don't think the difference is much more than that 1%)

About striker, I have more doubts. Now I think it is lower but my data is hard to analyze :P

i do not problem with estimation players... i see that my players are great and if see talent 2 know that his talent will be 3 :D

my youth and he has sure the best possible talent: https://prnt.sc/rf4iy4 https://prnt.sc/rf4gyc

SA estimation is right: http://prntscr.com/vnl6tc

and this have youth talent 3,49 but only 13 youth coach... so i mean that his reall youth talent is for 16 coach 3,1-3,3

and SA estimation:http://prntscr.com/vnl7qs

i can say, that estimation is not 100% but really near facts and it is enought for me and other...

(edited)