Forged Alliance Forever

Re: Standard deviation of the true skill rankings

2013-03-08T19:24:42+02:00

I seriously suspect, that playes are get matched in very narrow diapason of ratings, or tend to.
This means, the normal distribution bell is much-much steeper. Also, team game only players teams influence it, making it more like a stairway.

Statistics: Posted by Sunny — 08 Mar 2013, 19:24

Re: Standard deviation of the true skill rankings

2013-03-06T14:13:34+02:00

I really don't get your questions.

The average mean is 1500. That's how it set. The real average is now 1421.

The lowest mean is 162, the highest 2560.

The rating in trueskill is not a scale.
The values are entirely arbitrary. You are average if you are around 1500, but "average" now can mean more skilled than "average" one year back. It really doesn't mean anything.

It's designed to compare two players by checking how they deviate from that average, nothing more, nothing else.
Keeping your rating at a constant value doesn't mean you don't evolve, it means that you evolve (or regress) at the same rate as others.

You can ask me for data anytime, it will always be average is around 1500, and the distribution curve will always be like this :

distribution.jpg

Statistics: Posted by Ze_PilOt — 06 Mar 2013, 14:13

Re: Standard deviation of the true skill rankings

2013-03-06T12:54:17+02:00

Standerd Deviation ? lol i remember HATING that subject in math class long ago.

But yeah an SD could be useful, but then again, won't cleaning out inactive players do more to help people find an accurate representation of overall FAF skill rating?

Statistics: Posted by asdfmovie — 06 Mar 2013, 12:54

Re: Standard deviation of the true skill rankings

2013-03-05T21:49:00+02:00

Ze_PilOt wrote:
Take the lowest rating, and the highest one, you have it.

He want to understand, what rating means skill-wise, or wants to compare it to ELO.

I'm not sure why, probably wants to undertand, what global rating - player number density curve is.

For 1v1 it's pretty obviously like ELO.

Statistics: Posted by Sunny — 05 Mar 2013, 21:49

Re: Standard deviation of the true skill rankings

2013-03-03T19:31:39+02:00

Take the lowest rating, and the highest one, you have it.

Statistics: Posted by Ze_PilOt — 03 Mar 2013, 19:31

Re: Standard deviation of the true skill rankings

2013-03-03T17:46:40+02:00

I don't mean the average standard deviation of the players I mean the standard deviation of the normal distribution which comprises the entire player population, that does allow you to work out percentiles as it describes how compressed or spread out ratings are around the defined central value of 1500.

Statistics: Posted by higy3 — 03 Mar 2013, 17:46

Re: Standard deviation of the true skill rankings

2013-03-02T18:06:02+02:00

Can't see how the average Sd will make you able to see that. But well, its was around 70 last time I've checked.

Statistics: Posted by Ze_PilOt — 02 Mar 2013, 18:06

Re: Standard deviation of the true skill rankings

2013-03-02T16:18:50+02:00

Stalewee has it right I am looking for the standard deviation of the entire player population i.e how flattened the normal distribution curve of the rating values is, I am interested to know primarily out of curiosity about what percentile of the player population different rating indicates.

Statistics: Posted by higy3 — 02 Mar 2013, 16:18

Re: Standard deviation of the true skill rankings

2013-03-01T01:33:06+02:00

The inactive players were cleared off the list only a few weeks back, so the current list of 5000 players is probably a pretty accurate representation of the number of active players.

As for the rest of the question, I'm afraid I accidentally the maths.

Statistics: Posted by Eukanuba — 01 Mar 2013, 01:33

Re: Standard deviation of the true skill rankings

2013-03-01T01:31:16+02:00

I don't get what you will do with that exactly.

Statistics: Posted by Ze_PilOt — 01 Mar 2013, 01:31

Re: Standard deviation of the true skill rankings

2013-03-01T01:25:36+02:00

But in that case the SD is 1?

Statistics: Posted by Anaryl — 01 Mar 2013, 01:25

Re: Standard deviation of the true skill rankings

2013-03-01T01:24:24+02:00

Sounds like he's after the mean and standard deviation for the nominal rating over the population of players.

As opposed to the internal mean and standard deviation for a particular player.

Or I could be misunderstanding the question.

Statistics: Posted by stalewee — 01 Mar 2013, 01:24

Re: Standard deviation of the true skill rankings

2013-03-01T01:13:47+02:00

i dont reallyunderstand your question

That's immediately apparent - why answer if you had so little clue?

Your consists of mean and deviation - you can check your deviation by right clicking on your nickname whilst in chat and view "player statistics" - that will bring up your mean and deviation. But are you asking what the SD of a normal distribution is??

Statistics: Posted by Anaryl — 01 Mar 2013, 01:13

Re: Standard deviation of the true skill rankings

2013-02-28T19:59:18+02:00

Pilot said there have beein over 3000 (3246 or something) players logged in in the last week and played at least one game

i dont reallyunderstand your question

Statistics: Posted by ColonelSheppard — 28 Feb 2013, 19:59

Standard deviation of the true skill rankings

2013-02-28T19:49:09+02:00

Hey guys, was wondering if anyone had access to information giving the standard deviation of the true skill rating system normal distribution and the number of active players in total, out of interest of working out the percentile of players I meet and myself, any information appreciated.

Statistics: Posted by higy3 — 28 Feb 2013, 19:49