I agree, one main and easy fix for CA would be to have a minimum number of games played/tracked. 30 sounds good, or maybe even 50.

That would eliminate all of these "fake" top players, who have only played 5 games.

And of course the -40 handicap. If a complicated fix is too much to ask for, at least make it a smaller number. -10 should suffice, since it counts going spec and/or switching team as a quit too.

if you switch teams I don't think you get minus 40
-10 is just a thought, but it should at least be the same as the other losers on the team

elo resets when a player changes ql nickname. might as well add that to the list =)

very good point

the only way to find that out automatically is to check ones profile for old games with the old nick, if he didn't know that

Just purge CA compeletly, just do it. Delete it.

imho the system works reasonably well if you remove 4:3 games or the ones who abuse the system by winning 10 games and quit playing on the account

I just replied on another thread here on ESR, that I got them to remove fazed_ from the Nr. 1 CA spot (bc of only 2 games played).

And they said they will be working on a fix, so that players with a low number of games wont get displayed.

As long as it's possible to lose 500 elo in one match against a vastly better team, it may as well not exist.

link?

http://www.qlranks.com/ca/player/tcm

+434 for winning a game vs a much worse team
-486 for losing to a much better one

You just can't gauge overall team skill with the current system

there is weird stuff happening in the first 100 games, that's how that trueskill MS(c) thing works, and as you see they were immediatly after another..

No, that is NOT how True Skill works.

True Skill depends on simple rules:
- good team has high rating
- bad team has low rating
- your rating depends on the rating of those who you beat/lose to
- yes, initial games have very low confidence, hence very high jumps. But you will float to correct rating very quickly.

Meanwhile, in QLRanks for team games (TDM and CA):
- losing doesn't count if you quit
- winning counts almost as loss if enemy quits
- losing counts almost as a win if mate quits

And additionally for CA:
- quitting means -30 or whatever (I normally get like -2 if I lose in TDM)
- everyone has negative ratings because of this
- your rating depends on your number of quits more than others' ratings
- your rating still depends a bit on rating of others, but their rating depends on their number of quits, so...
- overall, everyone's ratings depend on overall number of quits



Look, this quit business is retarded.



Remove quit punishment.

As game finishers, count people who played more than half of game's total time.

That's it! Winjoiners won't be ranked, late quitters will be ranked. Lose-joining won't be punished. (I miss times when I joined losing team in last minutes to help them, now I don't because of QLRanks)



No, it's not technically hard to do. It's TRIVIAL. I'll do it in a day (including learning a new language) if you give me access. Ask wision for references.

Yes, it's probably hard to recalculate for past games. In TDM's case, you can just leave what's there and let it smooth out. With CA, it's all meaningless anyhow and can safely be wiped out :)



TL;DR: you're all trying to fix symptoms, meanwhile fixing the underlying problem is trivial.

that's what I'm trying to express with this post and my last one on the same topic :~

also those +400 -450 things are expected in trueskill, overall it's just a -50

I was under impression that the parent (Tcm) was mainly surprised that bad team had higher rating than good team.

Yes, he mentioned "lose 500 vs good team" to which you refer. But then he explained that the problem is: the outcome shouldn't be negative score. And he's right.

As long as the rating depends on quits instead of match results, it's going to be like that, though.

I have somebody's script installed that pulls player CA Elo rankings and prints them in-game.

While flawed, I've found that the relative CA numbers are pretty indicative of the strength of the players. That is, when the teams look pretty balanced relative to this, they're usually fairly close. When the QL Shuffle thinks the teams are balanced, but the rankings that are listed show heavy weighting to one side, it's often a blowout.

Definitely room for improvement, but I don't think it should be scrapped.

add ctf.

yes. add ctf or die.

And FFA.

promised on july 4th 2012

hall of shame instead of removing banned

Good suggestions. Purge inactive/banned/fresh CA

For the record: I remove players from my rankings if I have no confidence in the players current rating. It makes a lot of sense to do so because you don't want new players be part of the ranking.

Another problem is that you need to punish players who does not play for a long time.

- hide all CA accounts with less than 30 games played

fixed

Limbus is in the TDM pool. Been there almost since it came out. The games are tracked too.

hm.. seems true, but unfortunately the tdm frontpage qlranks.com/tdm doesn't seem to be correct then..

Yeah. Same with duel, though.

Meanwhile Campgrounds, Retribution, Realm of Steel Rats and Terminatria are still absent. All are in #tdmpickup pool (Terminatria in the main pool, the rest in optional pool currently).

Retribution and Realm of Steel Rats are free this month and we're playing those on publics (along Purgatory).

I'm surprised there's only 4 times as many CA games ranked as there are TDMs, as the map restrictions removed huge amount of TDMs (esp. in the first year when dm6 was free).

For players who haven't been playing for more than 6months - 1year should at least have some icon next to it's name (inactive) or something like that, so you know that he's not playing.

just do something! it has been several month without any actions, no small fixes, no new features, nothing. start slow, make tweaks here and there, or ask for help, if you don't have time and motivation any more.

it's kinda funny that first thing they made is the ql tv, which was the thing i was quite obsessed with, but couldn't make due to the location reasons. :)

as i answered so many like you before.

good stuff but sadly this will never happen