Pimp my SL again... Options

[kresala]

¿Any explanation about the flags, please?



greetz

[sl.Proddi]

¿Any explanation about the flags, please?



greetz

http://wiki.splatterladder.com/Sessionhistory/en
not complete, but we work on it.

[sunone]

ok . you made me write my frist post on this forum
first of all - changes - just one word to describe it : WUNDERBAR

I was also wondering about the flags and thnx to Prodi for that link. just one thought when we are at this point. I think you are based in Germany, right? because some of the translations into English aren't precise. please don't get me wrong, I think you are doing great job, but just had mind mind what can be done better
for example description of one flag -"to less players" it is not gramatically correct in English, but if you translate it directly to German then I am sure it is. In this example I would rather write - not enough players . just a small suggestions. hope you don't mind.

and a question from me. i see rating is updated after each game, but what about the rank?

thanks in advance.

sun

[sl.Proddi]

thank you for your comment. i know, my english is not good, but i'm a programmer - not a translater
the helpfiles was designed in german .. u're right ... and the english files are snapshots from the germans last date.

but its wiki, u can edit the pages and help us with the gramatic-snake if u want

...and ..btw. congratulation to your first post

[sunone]

you welcome. i see wiki has been updated already
but still I do not see answer to my question ...

btw i see rank was updated ^^ i just asked cos last time it was updated abou 3/4am so in the morning i always saw it fresh ^^

This post has been edited by sunone: Oct 27 2006, 08:17 AM

[sl.Proddi]

yes .. u're right again ... the ranking. now the ranking was calculated at 9 MET .. in 30 minutes from now (lookat my posttime)

[sunone]

hmm what a good start of a day. got info which i wanted. thanks again.

next thing : signature. i see it is a beta version. before the pimpin' it worked 'fine' but now some of the signatures aren't avaiable


and also a question how often those are updated, or how do they work, because when i generated mine now, it says old stats.

sorry for being so picky and impatient ^^ it comes with the age

[sl.Proddi]

this signature doesnt work because our sig-gen have problems with the new 64bit-technology. we work on it.
and this signature was updated every 3 minutes or if u played, 10 seconds after the serverscan. if u have problems with old signatur-images, check your pc-clock or press shift+reload to kick out this image from the cache.

ps: the ranking is done.

[sunone]

i see the rank has changed. nice.
as for signatures - i deleted temp int files and works now.
too bad that its the only way, cos then i have to log in to every forum account etc and I am not good with remembering passwords

This post has been edited by sunone: Oct 27 2006, 11:02 AM

[WTF!-B|ack]

I still think Sniper Maps should be ranked. Sniping is also a skill that can be beneficial to the game. The only XP you gain on a sniper map is through killing....why should it be skipped?

[Warmonger]

I mentioned this in another (evidently the wrong) thread.

Everytime a new player joins a game, or any other player even changes his/her name to something new, or changes colors...they come in the game with a 1 (or is it 0?) rating. This throws the whole system out of whack unless that person is so unproductive as to approximate his non-existant rating.

If two players A and B, both rated at 20.00, are dominating a server and both get similar points/minute...they should get +/-0.00.
If player A is using an alias, his rating will be nil, and player B will be punished for using his REAL name with a rating loss.

If player A is rated a lot higher and gets a higher score/minute, the point changes should be minimal...
If he's using an alias with that newborn rating, B's REAL name gets REAL screwed.

I absolutely LOVE the new setup and extended stats and more frequent scans. I even think using an Elo based system is a great idea, but it can't work properly in a open system such as this. Elo is for closed, controlled player sets. Innacurate ratings simply CANNOT be allowed to affect accurate ones.

A possible solution would be to consider players with too little rating time as 'unrated'. While they are accumulating enough rating time to become 'rated', their rating does NOT enter the calculations for the rated players in that session...but their rating change is calculated normally, or even weighted. Also, new names could begin with an 'average' initial rating, as opposed to 0.

Comments?

This post has been edited by Warmonger: Nov 7 2006, 10:17 PM

[sl.ErdnussJack]

Thats right Warmonger. Firstly I think you have already asked this in a other news thread?

In any case, we have already modified the "average rating" (from all players in a session-map). As you say the lower "outlier" are filtered now.

[Warmonger]

ErdnussJack' date='Nov 7 2006, 05:35 PM' post='29590']
Thats right Warmonger. Firstly I think you have already asked this in a other news thread?

In any case, we have already modified the "average rating" (from all players in a session-map). As you say the lower "outlier" are filtered now.

Yes I did mention it elsewhere, as I said above. Wrong thread, my bad.

Um...I dont think it's workin. I can see a CLEAR difference bewteen how the rating changes look with and without no-name no-time players in the game.

5th high-scorer...all realistic ratings, all the higher-scorers were rated lower than me, but not by much...I went 0.00

4th high-scorer...4 single-digit rated players (no rating time), one of them scoring higher than me...NEGATIVE 0.23

What's being filtered, exactly?

Is it just me? Am I crazy? Thanks though, for all the hard work...and I hope I don't sound like a pain in the ass. :)

This post has been edited by Warmonger: Nov 8 2006, 05:28 AM

[sl.ErdnussJack]

...no you are attentive

anyway the calculation problem of these "outliers" should work now. give us your meaning sessions and we take a look on them.
perhaps you haven't understood the system either. it's also different to explain how the new sl ranking exactly works but i will try to give some based details:

firstly if one wants to understand the sl-ranking then it's based on the ELO_ranking (in essential features) with a strongly modified calculation for the ET_statistics. furthermore we have introduced two "session-boni" within the session-calculation.

broadly speaking
The rating calculates itself from the distribution of the SPM within a Session. A better SPM is a better rating for you and a bad one for the other ones. The rating number of every player comes into a kind of pot. These points in the pot are distributed with the SPM-ratio on the players. If you have a twice as good rating as the rest, then you must play also twice as well (to get at least +-0). Otherwise you can by all means lose against better players and you get a green +. the rating of other players is a decisive point.
To our boni:
The, let me call, adjustment-boni is the correction of pro vs. pro and noob vs. noob sessions. the second, the player-boni, is independent from the average rating vs. player-rating of a session.

ok so far, so good.
Our modification (the problem with 0-rating outliers) takes effect in the adjustment-boni with the average rating. all lower third rated players get the average rating of the session. now the effect is thats the unrated player only increase faster in the ranking until to his level.

a simple example: (all in points)
player1 has 0 rating (first time in et)
player2 has 2 rating (its a noob too)
player3 has 0 rating (its a little bastard nick changer)
player4 has 15 rating (its a pro)
--------------------------------------
average rating = 4,25

player1-player3 gets the average rating for the further session-calculation.

P.S: sorry for some bad english in the morning :)

[Warmonger]

Thank you very much for taking the time to discuss this with me/us. (...And your English is fine. Way better than my German at any time of day...)

So, in the above example...player 4 would be expected to get more than TRIPLE the spm as each of the 3 other players (15 / 4.25, the avg. used for 1-3)

Easy to triple the noob's spm, but if the bastard nick changer is actually a 15 rating also, and gets 15 spm...wouldn't our hero, player 4, need to get over 45 spm just to break even?

What if player 3 is really a 20? Or higher! and SCORES like a 20? No matter how much averaging you do, 4 is still being affected by innacurate ratings.

The "Lower Third" is not what should determine an inaccurate rating. TIME is. What about players who just ARE rated in the lower third?

Players 1-3 should not affect 4's rating AT ALL as they have NO BASIS in fact, but THEIR ratings should be weighted according to how they perform compared to #4...the only one with a REAL rating in that example.

Your average is STILL based on INACCURATE RATINGS.

This post has been edited by Warmonger: Nov 10 2006, 10:13 AM