F|A server overrated??

[german]

hey there..i just wonder
-->
25-01 13:52
mp_theriver_2nd

F|A #2 XPSAVE FOREVER
Pro*D!Zz* Rate: 29.1176
11.6 / 0 min K/D 35 / 5 Rate:18.36 (202 xp) +0.01 or
Minas Tirith FP 3

F|A #2 XPSAVE FOREVER

25.8 / 0 min K/D 70 / 10 Rate: 21.8 (545 xp) +0.16 Average Rating 10.59 and Average PPM: 6.41
and he get fuuucked up higher rating..
i did today this
Capuzzo Airport Beta 1 PolskiSerwer XpSave

24.7 / 0 min

K/D 75 / 11
Rate:21.79 (523 xp) -0.12 Average Rating 10.69 and Average PPM: 10.17
and i went facking 0.12 down..i have a rate of 26..pro dizz a rate of 29
he went up +0.16 and i went down -0,12?? is this normal??? i have 3 rating points less..only 26 and i go down..so why does he increase rate soo much with a much higher rate of 29...
become this f|a server better rated?? or can someone explain pls??this is a little bit strange..cuz on this f|a server is kinda easy to do 20 ppm.
ps. nothing agains you pro*dizz if you read this..i just wonder...everyone goes up with 19 or 20 ppm on fa 2..on other server, players with less rating points loose rating by doing more ppm ...and average ppm and average rating on the map is almost same..

thx
sry for bad english

[JoNny]

1st: its all normal and always like everything...
2nd: it depends what other players make ppm so if dizz makes always 25ppm e.g. And otger players which just do 5ppm then his +rate will be higher as nornal ppm is 10 of some players..
;)

[Spl@$h~]

Here is the difference :


Rating 10.59 and Average PPM: 6.41

Average Rating 10.69 and Average PPM: 10.17

It's huge one ..

[testforecho]

imho all elo-like systems fail hard when you don't know the rating of newcomers and/or you don't adjust it really quick, because the whole system of comparing rating values is based on the assumption that those ratings are correct, and reflect really a player's skill.

In chess elo it's not a big problem, in et with all the renaming and stuff it gets huge... I don't know how tb deals with the problem of new nickname's initial rating, maybe making a 'standard' fixed average rating would probably lead to less trouble, since after all the average skill of pub players is generally the same ('bit higher on etpro servers actually, but nobody rates on etpro so...)

A newcomer (new nick) should start with a rating similar to his real skill, otherwise = system fail

[Paul]

imho all elo-like systems fail hard when you don't know the rating of newcomers and/or you don't adjust it really quick, because the whole system of comparing rating values is based on the assumption that those ratings are correct, and reflect really a player's skill.

In chess elo it's not a big problem, in et with all the renaming and stuff it gets huge... I don't know how tb deals with the problem of new nickname's initial rating, maybe making a 'standard' fixed average rating would probably lead to less trouble, since after all the average skill of pub players is generally the same ('bit higher on etpro servers actually, but nobody rates on etpro so...)

A newcomer (new nick) should start with a rating similar to his real skill, otherwise = system fail

Unfortunatly, we can never know on what value we should start initially as players mostly play some few rounds with their new name, then reserve it. The only possible (partly) fix I would see in here is to set the rate to 5 on all new names :)

[german]

well i just wondered..because there are many over examples..ppl playing on fa 2 having a rating of 27-28 making 19!!!ppm and increase +0.1 and other ppl with rating of 22 are going down -0,15 with 20 ppm..average rating same and ppm are 2 ppm less on fa2..it looks a bit strange..for sure trackbase tries to calculate this stuff fair..but in my eyes it doesn´t look fair sometimes..dunno exactly how many values become calculated!! but to be honest..it is noooo big deal to get 20 ppm on fa2..really..just bit adrenaline and so on..but it is much more harder to make 20 ppm on other server.
thx for infos..

[Scarhand]

Testforecho once mentioned doing a comparison with FA and another server and this is his results:

I am sure the real formula is more complex, and I don't know it :), but to make a simple one let's say that a server has:
average rating 10, average ppm 6, then if you wanna go toward 30 rating, you need to make 18ppm (30/10*6) so to make a good rating, you need 3 things:
1. high average rating and low avg. ppm, so you need low ppm
2. the actual ability to make those ppm in that server.

examples taken from 2 random real sessions I found, and calculated with my simplified formula:


server avg.rate avg.ppm ppm needed for 30 rating
FA2 11.02 6.3 17.15
BiO 9.05 6.24 20.69

In this case, it's obvious that you need to play in fa2, since not only you need to make only 17 ppm and not 20, but also because making 20ppm in bio is probably much harder than doing 20 in fa

that's why. Even if my formula is simplier than the real tb formula you can get the idea

As you can see, the average rate is very different proportionally to the average ppm there, which doesn't make sense because it should balance out. My belief is that it is due to team-stacking, which I often hear complained about. Hopefully, TB which switch to team-based comparisons which will greatly improve the accuracy of rating.

[german]

hope so..because i will never play on fa2 for rating..first of all sucky ping..sucky admins..most of them are just some kids who are kicking/baning good mature ppl (i don´t have been baned till now :) but some friends in the past) well an exact rating isn´t and won´t be able i think..there are soo many values..like team stacking or skilled players switch to another name and than this renamed players starts with low rating but has good ppm..much those stuff...hope this new trackingsystem with skillpoints will work better :)

[testforecho]

Unfortunatly, we can never know on what value we should start initially as players mostly play some few rounds with their new name, then reserve it. The only possible (partly) fix I would see in here is to set the rate to 5 on all new names :)

well if we take for granted that most new nicks are not real new players, but an old player who changed nick, the highest chance to get a realistic one is giving an average rating, so why not let them start from the middle, who goes up who goes down, most will stay.

Because the 'average rate' thing is strongly based on the fact that rate is correct, and if that is the starting point for all calculations...

And maybe if a nick is new its rating should move faster for x days, I don't know if you have that already implemented, I am just guessing

Etpro players change name every other day, and that keeps their rating low