automated daily skill restricted tournaments on site? (low-mid-high)?

"team ranking"

i had a random idea

do you have an idea?

to fuck off I guess :/

I think this is fully doable even right now. The question is - whether QLRanks devs are interested in it.

tehown has

I think CA ranking is a tad fucked due to the quitters and how quitters are punished on qlranks

imho just use the algo you have right now, but rank every game that ends with 3v3 instead of 4v4 minimum
quitters should just get a normal minus from a loss instead of the constant -40
for games ending with 3v4 or any other uneven number - dunno :/

if the tools are available to you, take all the games you have, and redo the calculation with those slight changes and do some finetuning if necessary

CA ranking is totally and completly fucked anyway. It should be reset or deleted all together. From certain point on it is only possible to lose ranking since no matter what you do it's always get "-" points for each game. Win 10 games in a row -2 per game ... -20 ... which is just fucking stupid. And it doesn't matter against who you play or how well you perform.

it highly depends on who you are up against, if you continue to play shitty pubs and win vs a bunch of 400 elo kids, you deserve it

that's the definition of CA

recently (since the subscription change by id) I got a lot of pretty high calibre prem games going, Swedish, German, UK servers

in terms of elo, that's self-cannibalization of course, but who cares it's good games

this ... eventho my elo plummeted for like half, it was worth the ggs.

... what happened to the team algorithm anyway?

ahem... you meant to say some players got put in their place? :p

i agree but those games are few in between, i'd love to have some sort of fixed "higher skill" servers.

Youd' think that wouldn't you, everybody does, but it doesn't at all. I can play against players who have +1000 ELO points on me and all I can look forward to is either +1 or -1 ELO if i win.

To MailmanMel:
It does work as T1E stated. But to be more specific, it is about the ELO AVERAGES OF THE TEAMS.

Example 1:
Your team's average ELO: 1000
Other team's average ELO: 1500
If you win this game, everyone on your team will GAIN ELO. Even if YOU individually have 2000 ELO.

Example 2:
Your team's average ELO: 1500
Other team's average ELO: 1000
Even if you win this game, you will LOSE ELO. Even if your individual ELO is only 200.

If, let's say, both teams have about same ELO, i think you also gain ELO if you win (or at least +/- 0).

So basically it is all about the team's averages. Make sure to have NOOB ELO's on your team while you are playing against HIGH ELO's on the other team. This scenario is best to GAIN ELO (of course you have to win).

To T1E:
Games of 3v3 do count. It is games that end with less than 3v3 (for example 2v3) that dont count anymore.

The biggest problem I see so far is that quitters get punished with -40 every time. Of course you have to punish quitters somehow, but that's a ridiculous way. Because even switching teams during the game counts as a quit and results in -40, no matter how good you play after the switch.

That, in my experience, is the sole reason why many good players have only 1000 etc... ELO.

Basically right now ELO shows:
it is really hard to get high ELO. So high ELO does pretty much = skillful player. But low ELO does NOT = bad player. Theres many good players with low ELO, and that is because of the quit punishment.

have you seen the CA ladder? Most of the top20 are beyond awful. It's simply broken.

Also, 1v1 and 2v2 CA are the only ones that take a bit of skill (3v3 and onwards is laughable), and you say those don't count?

I'm not talking about the 'bug' where a new player can play 10 games and get 2000 ELO because of his first games giving him +500 or more per game. That is obviously a bug and easily detectable if you look at their QLranks profile.

But once you've played hundreds of games... try getting a really high ELO. It isnt easy. If you say it is, then let me know when you are at the top. Until then... pssst.

Personally I don't care about elo and defeats (there was a time I did in LoL), but anyway.

http://qlranks.com/ca/player/ymcmb
This is my account, and as you can see I'm a pub stomper. You can check it on quakelive.com as well.

First of all, I find funny that it isn't tracking my last games. Or if days are fucked up somehow, it would be really sad that I lost 40 elo on that somewhat damaged game, because we won (as it shows on qlranks) and if it is the same I'm thinking of I had a score of 160 and second place like 50 (even took a screenshot rofl).
It simply is badly implemented that you lose elo when winning, even more so when you are by far the best player of that match.

Secondly, it's just sad that right above me there's this guy that has only 1 game played, which was a defeat. http://qlranks.com/ca/player/FUBUKI

Finally, there are 4 players from my location that are on the worldwide top100 and I'd say I can defeat all of them playing with my feet, monitor turned off and everything else. And I mean it.

There's no interwebz glory on an aleatory ladder. Duel works a like charm, plus it truly is an individual gametype so it is more accurate, but CA... pssst.

And besides, if I wanted to win I'd play only with my team and win 100games in a row rofl. That's pub CA for you.

QLranks doesnt update every CA game like it does on duel. And it seems to lag behind...

But you can MANUALLY UPDATE instantly (for duel or CA). Go to IRC #qlranks and type:
!update username (to update duel)
!updateca username (to update CA)

I updated your CA, so now it has all the recent games listed on QLRanks.

I didnt check the game you are talking about (on Somewhat Damaged), but im guessing if you got a -40 that you either quit/went spec or switched teams during the game. That always results in a -40 (and is what i was talking about earlier - something that needs to get changed).

About the FUBUKI guy: there seems to be a bug? He has 1785 ELO with only 1 game, and that game giving him a +/- 0. That means he either started with 1785 (which would be a bug, since normal is 1250 or 1200, not sure) - or he played a long time ago which resulted in this ELO, but hasnt been active in a long ass time and the QLranks website doesnt show these games anymore? Either way... weird. :)

But anyways, here are my main tips to gaining ELO in CA.
1) Don't ever quit, go spec or switch teams during a game.
2) Use QLPrism and/or the ingame ELO-Announcer to make sure you are playing with LOW-ELOs on your team while playing against HIGH-ELOs on the other team. (Add up all the ELO's and you can tell before you even start the game if you will gain or lose ELO playing it.)
3) and... of course... win :) (oh and btw... your individual performance has no influence on CA ELO. It is just team win/loss based - if my 'research' is correct.)

I literally never quit (98% completed with almost 1000 games must be a worldwide record), so it probably is that I change teams a lot to balance things I bit. Guess I'll shameless roflstomp to the end.

even with manual update via IRC ... it still doesnt collect all of the games. Could it be a donation only feature?

I think it does collect all games if you manually update.

I had a phase where I would manually update right after every game (during map changes) and it ALWAYS updated the game.

(I did this to find out how the algorythm works. If you analye the team's ELOs and then see what effect the win/loss had on your ELO right after the match ends, then you can "learn" which scenarios to avoid... :)

Of course, games that end in less than 3v3 DO NOT COUNT and thus will NOT be updated. So if you play a 3v3 and a player quits two seconds before the buzzer sounds, then the game will not count.

I don't think it is fucked up because of that. I think it should be deleted, since there's no point to spend resources on very badly done ladders, or reworked completely (plus reset obv.).

Does it work with tdm?

What information exactly is available on matches? How much data could be stored about each player? How heavy can the calculations be to rank a player after a match?

one problem that i didn't even expect as one:

http://www.quakelive.com/#!profile/matches/T1...1fd90/ca/1

this game ends with 3v4 cause of a quitter on the winning team
as a conclusion it thinks all rounds were played like that and does -57 on me (losing team) which shouldn't happen

and following that thought, there will be many games where the winning team gets too few + points cause the system thinks they won vs a 3-person team

CA rankings have been garbage from day one, and still is.
You can dominate the server of equal players and get -200.

Fixed it for you:

90% of score related to dmg given. (rail damage should be ignored)
10% of score related to team win

NP.

8DD

I'd rather suggest getting rid of CA ranking and replace it with CTF rankings cause most people (especially those who play a lot of CTF in America, Pickup Games wise :P) could see where there at with CTF in QLRanks.

why would you need to remove CA ranking to get CTF rankings?

If people are complaining about the elo in CA and how it doesnt work and such

Better yet, remove CA and CTF. Keep Duel.

this or even add ffa (which will make sense, unlike the team ones)

edit: 1 extra reason for removing CA is that ppl started lately using the CA ranks as prove of their excellence... go figure

IF you are using an MVC environment you should do:
Expand your database with a foreign key table which contains the range of players that pass a certain filter like having 150 of ClanArena games.
Then you have your controller read the range and use the foreignkey ids to select all the filtered players from the main player table to send to your view that creates the player table on the web page in your template.
Since all players are still entered in the main database table but you've got a front for picking the right players to display, any new player under 150 games can still be linked to his own ELO profile page and can have his games compared to others etc.

It's not simple but it would solve the 2000 ELo palyers with under 100 games for instance.

As for tdm ranking, or any team game with items, it's still not prudent to do anything more than what is being done.
TDM skill of a player is determined by more than what is currently used. For instance Statistics information we get on scoreboards every match would help here a lot. If a player is doing something like +2 net only, but has significantly less resources than rest of his team, then his tdm effort is the same if not greater showing more guise in playing. Also it depends on the other teams players and their perspective scores and statistics if a pikcup match is counted. Comparing them to each and everyone on the other team.
As for team ranking I really don't know what to say. You have teams that play with different lineups, presenting different skills and abilities with each one of those. Also you have teams that have individually really good players but can loose to individually inferior players.

Would it be possible to fix the elo gaining loophole I just came across.
If you host your own server, just wait for someone to join your server and when the game has started just kick them, from your own server. You'll get an instant win and the +elo accompanying it.
In theory it should work until the very top. Even if you actually play the game you can still kick your opponent when you lose RA or MH timing and victory is guaranteed.

and how do you suggest to fix it?

uhm.. that's against QL TOS

send in proof and he gets banned

I don't know but what is the database time of it? Because I find for myself quite old games, some looks like they were from my previous account in late beta times. From my point of view this looks ... (I have 75+ CA games now but they are not listed)