Subject: little junior talent tool

this fictual player gets a correlation of 0.14
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 6

You're right!
That's sounds weird. I'll try to sort it out...

do you use something else for 'y'?

No, I also use 0-34

I am a litle bit testing with that correlation and i get some very strange results...

this gives a correlation of 0.6:
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 5

and this a correlation of 0.8:
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 6 6 6 7 7 8 8 8 9 9 9 10 10 10 10 10


I know the second one will never happen in SK, but i made it to look how this function works. For now I doubt it gives a good impression about how good the evaluation is. I think i stuck with the median(5)lineair regression as most reliable function on this moment.

its correlation. clearly that last point in the first one is an outlier.

the value of 0.6 tells you this fact. this is what its for.

it does not tell talent, anything like that, just how well all the values fit and if one point is possibly causing a talent graph to be incorrect.

i mean if you add a 5 onto the end of the second one also, it will massively throw off the correlation result

Ofcourse i know it says nothing about the talent itself but about the talent evaluation.

But what i have seen after some testing is that the correlation does not say realy much about the consistence of the evaluation. Just try it out your self :)

"the value of 0.6 tells you this fact. this is what its for."

yes about 1 outlier, the other one has much more of them and a way high number...that is just my point.

And it is one below, when i put a 9 on the other on the end i get 0.76....

/edit:

with the first line ending on 6 -> 0.76, with 5 on the and 0.6 it is a big gap that shows that 1 datapoint on 35 datapoints gives a big different result. But it is just a difference of 1, what they always will have in youth. how many matches they play, it will always be a range of 1 at least.
(edited)
(edited)

this is the sheet i use: link to xls

(It is in dutch)
It doesn't load xmls, on 'input' i copy paste sokkerorganizer youth page every thursday/saturday, but further it ia ll manual.

/edit:
i made a quick translation to english. link to xls

comments are very welcome
(edited)

I read your consideration about correlation factor.

In my Excel graph (some post before) Excel calculates R²=0.87... so it's a good evalutation for you?

Well... I believe we have to be careful with what we call evaluation.
Evaluation is first given from your jr. coach. I assume a high overall skill coach will give better evaluations than a poor one. Coach evaluations may also vary if the player spent time on the pitch during jr. games or not.

Back to the R², 0,87 means the data are quite close to the line drawn on your graph and that the correlation between your data and a linear function is high.

Though, there are some cases where R² is maximum (1) and data are weird, and also the opposite cases when data are very regular and R² is low.

Just consider R² as a check point or a weighting argument when you calculate the talent from a linear regression.

I personally consider that R²>0,86 gives me enough liability to my talent estimations if I have more than a 10 weeks history.

agreed with that. Just use the correlationnumber as what it is, something that can help you with your analyzing. But the analyzething is still a brainfunction ;)

But the analyzething is still a brainfunction ;)

I agree... ;)

ok 5-6 mln

Somehow I can't get this sheet to work :\ Mostly if I have more filled data for any player I have got N/A at the data/talent part.

I got some problems with installation framework 3.5...

I wrote Excel macro for detecting YP talent . may be It' will useful for somebody

screen 1
user instruction

downloads:
1)rapidshare
2)megaupload

I get an error it says,

cannot find file ATPVBAEN.xla cannot be found when i click on check talent after allowing the macro to run.

Any help will be appreciated

Best junior-tool-ever. Really miss it :/

R.I.P.