Thema: Sokker Tools by Geston

he started at 9 while it's obvious that starting max is 7.00-7.99 (I would assume it's 7.2 as 40%), so linear regression might base on the following weeks and just create the line so it would cross (0, 7.2)

It seems that adjusted progress is round to higher integer, but it should be round to lower integer (it’s how it’s work in sokker). There is a mistake in adjusted graph since the trend line and initial point is correct (7.5 at w.0) according to linear reg.
(Edith war hier)

I will reply at today's late evening.
To much work today :)

So...

About "talent": +1. But what’s for min ? 5.0 ?

Everything here starts from linear regression, adjusted progress aswell. Im calculating components "a" and "b" of linear function "y = ax + b". "b" is not that important right now. Most important now is "a" component, this is actual talent in linear regression. What I'm doing in this step is checking if "a" is higher than 0,(3), cause 0,(3) is max (best) talent in theory. Why? Long story short: 1 / 0,(3) = 3.0.
What about minimum (worst) talent? I was collecting some data from my youth school for several months and juniors with 15+ weeks in ys had worst talent around 8.0-9.0. But I cant be sure that 9.0 is the worst talent could happend here, Im just checking if "a" component is lower that 0,05, which is 20.0. It is probably overestimated, but for now ok, I must collect more data.

@kryminator: he started at 9 while it's obvious that starting max is 7.00-7.99 (I would assume it's 7.2 as 40%)
This is very valuable tip! "b" component could be set to max 7,2 value for example, or first week level, than second set min to 8. Worth discuss and consider.

@juytt: It seems that adjusted progress is round to higher integer, but it should be round to lower integer (it’s how it’s work in sokker). There is a mistake in adjusted graph since the trend line and initial point is correct (7.5 at w.0) according to linear reg.

Adjusted progress is nothing else but rounding liner reg, not to higher integer, neither lower, but to closer one. this is why green line goes in one part below blue line, and than pops up above.
True, that in sokker skills are rounded down, but liner reg (blue) line is not a line of exact skill level, blue line is already below true skill level. This is why green line is calculated correct in that matter.
But it is simple to test and prove it. Just imagine junior of any talent you want, write down his level at paper and further pops for next 20, 25, 30 weeks without "mistakes". Without mistakes green line always(?) sticks to red line.

But to be sure, to start chaning anything here, I need to collect more data of juniors with 30+ weeks in youth school.

----

Im not professional programmer, doing it after work as a hobby, exercise.
When I was starting this calculator, I didnt even think it can grow up that big. I mean I wasn't planing anything, it was just drawing graph at the beggining. Next came linear regression and adjusted progress. Next - reading json files, uploading images, more and more staff - not everything is visible. Without plan at the beginning my code became a big mess , it already needs to be rewrited to continue doing improvements.
(Edith war hier)

but liner reg (blue) line is not a line of exact skill level, blue line is already below true skill level.... (...) Adjusted progress is nothing else but rounding liner reg, not to higher integer, neither lower, but to closer one

In definitive the true levels of your junior are supposed to be the points on your trend line, if not, supposed calculated talent in regression and the linear function are wrong.
Since the Adjusted graph (green) is build upon true levels of trend line and not build upon on (red) "mistakes" there is no reason to "smooth" things by rounding to the closest integer but it should be round down like sokker do.
(Edith war hier)

Problem with talent label when scaling - fixed

So, I have a question maybe you or someone else can answer.

I see a youth talent calculator, but is there a calculator for advanced training talent? Something that factors in the skill being taught, age, and quality of training?

I am still trying to get a feel for 2-3 of my 10 advanced trainees talent level, and I'm going largely from memories of a decade ago. I wonder if there's a tool out there that can help with this?

Sokker asistente does that, try it our :)
(Edith war hier)

Thanks!

Updated counting talent algorithm. Should predict even better now.

That is a really good job. This page is useful. Now I can selected easier my juniors.
I tested this. I felt I have a first divine junior. But as I used this page.

Now the page say he will be brilliant (14). He is now brilliant. But if I see the pop ups and he i 3.3 talent, I feel he will be level 16 junior.
After 8 week I will now it.

Geston, how many weeks do you count as:
Minimum to use talent prediction;
Good sample to make adequate prediction;
Week number for very reliable prediction.
Thanks!

For my personal youth school control I dont even look at talent below 5-6 weeks in school.
And am not taking so serious talent of juniors in school less than 10-12 weeks.

But I can't tell there is something like threshold, prediction from 6 weeks is always better that from 5 weeks, from 12 weeks is better that from 11, from 18 better than from 17 etc etc.
The longer junior is in ys, the better prediction is.

Added "expected sum-skill" label. Prototype!
New bb-code includes also link to page with current junior data

> Youth Talent Calculator

Hey geston, thanks for tools!
Any idea about the tool to determine best stadium ticket price?
It could be overall or advanced for a certain situation, where you put your rating, opp rating, stadium size, fans mood, etc.
It would be really great, I think this is space where a lot of users losing so much money.

I would love to see that you can change price for every side of stadium with one clic. not 8 x same actions :D