Subject: little junior talent tool

between 3° and 4° line in my excel sheet (see up), you could see the application of "TRUNC" function...

... it seems to be what The Manager asked...

(edited)

yeah, that's it ;)))

thank you very much, it works ;)

;)

2) Using this function by myself, I noticed that this technique isn't so reliable. It needs a lot of evaluations (according to me at least 15), ad it seems to significantly change talent estimation during weeks even with just one/two evaluation more... what do you think about?

I also have such an Excel sheet and I added a coefficient of determination to have a calculated overview about the spread of the datas. The closer this coef. is to 1, the more accurate is the talent evatuation.

I am using your excel sheet

Nice to know some of you are using it.
I somehow stopped its development but I still plan to improve it.

If one of you is expert in importing XML data into Excel, I'd appreciate any advice.

I added a coefficient of determination to have a calculated overview about the spread of the datas.

What kind of coefficient? I don't understand :/

its called a correlation coefficient in english ;)

e.g use this

http://www.easycalculation.com/statistics/correlation.php

all it does is shows how much the current graphs fail XD
(edited)

It only gets out the big fluctuations... but it does not correct this:

5 5 5 5 5 5 5 5 5

when real progress was : 4 4 4 5 5 5 6 6 6

There is no way to sort this kind of problems out. I see enough not big fluctuation who look great/rubbish after 10 weeks and turns out after 17 weeks the opposite.

I use the Pearson coefficient of correlation applied to the sample of datas.

In Excel (English) you can use the RSQ function.

you sure you should use pearson's instead of spearman's?

do i get it?

these are some values do i get for my players:
0,86 - probably good talentevaluation?
0,12 - not good evaluation
0,93 - probably good evaluation
0,53 - mediocre evaluation

i say "probably" good because of my statement before about the problem of few data. But the 0.12 is definatly not a good evaluation because it has a low correlation. Am i right?

/edit:

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

(edited)

In fact, I use R² which is calculated from Pearson.
Imho, it's enough and correctly related with the poor number of datas we can deal with :)

You are perfectly right.
0,86 is the limit.

You should also make sure that you have at least a 11 weeks history to take this statements in consideration.

(Edit for complements)
(edited)

I think my formula is not right i use 0-34 for 'y', but it give low correlation for low talented players.... do you use something else for 'y'?

/edit: i use 10 weeks for my formula now, but the 0.12 player is 11 weeks in:
6 4 6 6 7 6 6 6 7 6 6 6
(edited)

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...