What are the tiers now?

[KaasMan]

I've reached the point where I basically can't progress. I'm either dominating a T3 game or getting absolutely shredded when I get bumped to T4. It's a vicious cycle every time I log on these days. The game actually encourages me to want to do worse if I want to have fun, even though I'd prefer to get better.

I don't understand how you manage to do this. I'm fulltime tier4 now, but I used to have a time where I also was 'between' tier 3 and 4.
I didn't asswhip tier3, and didn't get asswhipped in tier4 either. Probably you are just an annoying kid who thinks that losing a 1on1 lg mean 'being absolutely shredded' and winning one means 'dominating'.
Anyway, if you dislike tier4, use a handicap in tier3 and quit whining.

And megaman, he said 'top 1/3', you said 'top 3'. There is a difference. For example, someone who is 3rd on teamsize 3, is in the top 3, but not in the top 1/3.

[FlashSoul]

Any skill rating system has to be based on something. The score just happen to be the best thing out there for that. I'd blame the score system for inaccuracies, not the rating one.
CTF : I think the system tries to not rank too high those who are only playing CA in CTF. It seems kinda hard to distinguish between a "useless" frag and a legit frag if it's not close to your flag so, yeah, mid players are a bit screwed on this.
CA : 1 point per 100 damage dealt and one point per frag. It is fair, in my opinion. If you consistently fail to secure frags then you are doing something wrong. An opponent that you let escape with 1 hp can still deal loads of damage; that's why frags are important.
FFA : Eheh but the same problems of damage dealt VS frags arise in FFA!
Duel : It's not exactly based on score but the most important part is that it's based on your opponent's skillrating, which is good.

It's the exact same I said but with more words.

Read both of our posts again : it clearly isn't the "exact same thing". I have no idea why you would say that especially when I'm contradicting you in a part of my post...

[FlashSoul]

Not sure whether you guys have seen this, but there's an ictf stats website that implements TrueSkill ranking system to gauge players in team modes.
I checked the US iCTF ranking (as I am familiar with the top players there) and it looks legit.
http://ictf-stats.com/players

Wouldn't pretend I know a lot about ranking systems, but anyone have any ideas, whether this is a better system? (looks good at first glance)

QLRanks's CA algorithm was based on Trueskill before the recent change and it truly sucked. I'd say it's mostly because teams aren't constituted randomly. If they were, then yes it would produce decent results. But as it is, being better doesn't mean you will win more matches unless you are on the very top (or the very bottom). Why? Because people, and shuffle, always try to give each team roughly 50% chance of winning chances. If you are very good on a 3v3 CA for instance you may always end up with the two worst players of the server in your team to balance your strength. Your win rate will stay around 50% and Trueskill will be useless for you as it's basically the only important thing for it. This sytem is better to detect win joiners/team stackers than good players.

[Ph4nt0m]

QLRanks's CA algorithm was based on Trueskill before the recent change and it truly sucked. I'd say it's mostly because teams aren't constituted randomly. If they were, then yes it would produce decent results. But as it is, being better doesn't mean you will win more matches unless you are on the very top (or the very bottom). Why? Because people, and shuffle, always try to give each team roughly 50% chance of winning chances. If you are very good on a 3v3 CA for instance you may always end up with the two worst players of the server in your team to balance your strength. Your win rate will stay around 50% and Trueskill will be useless for you as it's basically the only important thing for it. This sytem is better to detect win joiners/team stackers than good players.

Flash,

I am guessing that your reply assumes the TruSkill rating shown in the website I shared is purely on win%, where I think it's not. I agree totally with your points if it was so, but if you look here
http://ictf-stats.com/players/rank
the ranks are not in the descending order of win%. I think it somehow incorporates who you are playing against too, and doesn't punish you too hard for playing against tough players (pug vs pub). Maybe its a mix of some sort of Trueskill and elo rating? I donno

[skeletonface]

The other problem with awarding skill rating points for finishing in the top third is that you ARE going to finish in the top third from time to time. Your rating will obviously rise, but you might not be actually getting any better. Which poses a question....
What happens when you are in the top few places, but everyone beneath you ragequits? And what happens to a player in the bottom third who decides to quit? Has he avoided losing his points?
All in all I find most games to be ok, and there's rarely a player who really dominates unless it's a blatant new account set up by some good player who fancies farming him some noobs for a while. The only time it's a real problem is on special, large unrestricted matches. Even then, it wouldn't be a problem if the good players were spread fairly, but people dodge shuffles and winjoin and stack one team with the good players.

[FlashSoul]

Flash,

I am guessing that your reply assumes the TruSkill rating shown in the website I shared is purely on win%, where I think it's not. I agree totally with your points if it was so, but if you look here
http://ictf-stats.com/players/rank
the ranks are not in the descending order of win%. I think it somehow incorporates who you are playing against too, and doesn't punish you too hard for playing against tough players (pug vs pub). Maybe its a mix of some sort of Trueskill and elo rating? I donno

It does use the rating of your opponent and teammates. But it's not enough... the premise is still flawed.
This system is based on the belief that your individual skill has a perceptible influence on the probability that your team wins. It may sound logical, but the individual skill of the majority of players is not high/low enough to have a perceptible effect. Because of the team balancing, the probability of winning will adjust around 50% regardless of a player's skill. I repeat : team win/loss is basically a random event independent of your skill. So what if you win? What if you lose? You gain or lose skillrating, but none of that is justified. It all becomes noise. Yes, losing against a team considered "weak" will lead to a big loss, losing against a "strong" team will be a little loss and so on... but are these teams really weak or strong? Or were they not simply the product of a random string of win/loss with no meaning whatsoever?

So basically most players just win and lose skillrating randomly after every match. Randomness being what it is, some will win more points and some will lose more.
Only those who are strong enough to unbalance the teams (read : give their team an higher winning probability despite balancing) will have a chance to get correct rating after enough matches. Their "correctness" may propagate to those who play with them to some extent but I doubt it can go all the way downward and give everyone a significant rating; it'll most likely be far too chaotic to fight.

Disclaimer ~ what I write should be correct when those conditions are respected :
- The TrueSkill system gives everyone the same default value; there is no "intelligent" initialization using another system;
- The TrueSkill system is used, from the start, on a playerbase which is already correctly rated and balanced by another system;
- The TrueSkill ratings have NO influence on the team balancing.

This is the case with QL and its external rating systems with the exception of some pickups which use QLRanks to balance the teams (I think so... or was it QLstats? Erm... not sure...).

[FlashSoul]

The other problem with awarding skill rating points for finishing in the top third is that you ARE going to finish in the top third from time to time. Your rating will obviously rise, but you might not be actually getting any better. Which poses a question....
What happens when you are in the top few places, but everyone beneath you ragequits? And what happens to a player in the bottom third who decides to quit? Has he avoided losing his points?
All in all I find most games to be ok, and there's rarely a player who really dominates unless it's a blatant new account set up by some good player who fancies farming him some noobs for a while. The only time it's a real problem is on special, large unrestricted matches. Even then, it wouldn't be a problem if the good players were spread fairly, but people dodge shuffles and winjoin and stack one team with the good players.

I'm not sure of the exact details but everyone who played in the match should be counted, even if they quit. However, whose who haven't played enough (2 minutes? or something like this) won't have any skillrating change. I'm not sure if they are still in the list and used in the player sum but I don't think so.
You should also be aware that what's used isn't exactly the score; it's the score weighted by the time played in the match. At the end of the match, the player list is sorted with that field.
About your concern with finishing in the top third from time to time without improving : don't forget that you will finish in the bottom third from time to time too so that should balance it. (top third : gain skillrating; bottom third : lose skillrating)

[68k]

I don't understand how you manage to do this. I'm fulltime tier4 now, but I used to have a time where I also was 'between' tier 3 and 4.
I didn't asswhip tier3, and didn't get asswhipped in tier4 either. Probably you are just an annoying kid who thinks that losing a 1on1 lg mean 'being absolutely shredded' and winning one means 'dominating'.
Anyway, if you dislike tier4, use a handicap in tier3 and quit whining.

It's not like that at all. I want to be able to stay in T4. And, sometimes when I get bumped up, I'll do pretty well in T4, but I can't stay in for more than about 4-5 games before inevitably getting knocked back down. I don't mind T3, either, I just like to feel like I'm getting better, but now it just feels like I'm in limbo. It probably doesn't help that I only really play CA. Perhaps I'll break out of it at some point.

I guess it just feels like T4 games are all over the place - I'll play one, and it feels like I can hang; I'll play another, and it feels like I'm literally a baby just smashing my arms on the keyboard.

[FlashSoul]

I guess it just feels like T4 games are all over the place - I'll play one, and it feels like I can hang; I'll play another, and it feels like I'm literally a baby just smashing my arms on the keyboard.

Your analysis is right. Tier 4 is the tier in which the internal skill difference from bottom to top is the highest. It's the end of a bell curve (or somewhat like one) don't be surprised.

[kodisha]

woah, thanks for the links, i only found out about this thread from the google analytics

I Will leave trueskill part to tyhjyydesta, since he implemented that part.

Second, together with him we added premium matches only ranks:
http://ictf-stats.com/players/rank-premium

And we did all this to help community to track their progress and tho help them realize where they "stand" compared to other players.


We are also working on (actually its done, we are just looking at the quality of the results it gives) on integrating this into ictf pickup channel.

@FlashSoul
And yes, the number alone can't tell you everything about the player, but that's why we keep all kinds of stats to determine best position for a player and create good shuffle.

And also, regardless of rank, if you are that unlucky to always get into loosing teams you can still check your accuracy/captures/defends charts/stats and see how you are doing

And as you get better so will your win ratio. Im looking at this data for few months now, and there are no good players with good stats and low win ratio - you play good, you win. You might have a bad day or noob team, but if the shuffle is good, or you play pickups your rank should be reflection of your skill.