How does the ranking system work?

[Elduke]

Hey, have been playing on UT3 on the PS3 and am getting a Little confused with what is going on with my rank. As my kill/death ratio grows bigger along with my killing spree etc my rank is dropping instead of going up the leaderbords.
When I finished up last night I was at about 150 rank.When I started playing today my rank had dropped to like 300. Far enough but then I started checking out some of my mates who are ranked above me and there states arnt as good as mine.
My kill to death ratio is something like 2100/1100 while i have a friend with something like 2100/1700.Plus all my killing sprees, rampages etc are all higher than his but he is ranking in the 150's.
So what is the deal?

[kaluro]

It doesn't.

[Trepænima]

Rank is determined by Elo. The Elo system works "okay", as long as everybody is dueling, but they're not, so it's not exactly representative of a true rank. Your rating/ranking goes up as you kill higher quality opponents.

Suppose the following scenario: In DM, let's say there is one player who went 22 and 5, and another that won the match at 30 and 31. Let's say out of those 30 kills, he only killed the 22 and 5 guy once, and 22 and 5 guy killed the leader once. By Elo they are "tied", but since the 30/31 guy finished the match with more kills, he gets another "win" on the 22/5 guy, despite the K/D ratio. So 30/31 guy gets more Elo rating, 22 and 5 may get some Elo, but not as much as the winner of the round, as not only did he get 30 kills, but also an extra "win" against every player in the server who did not outscore him. Thus people who play Warfare are likely to have higher rankings than DM players, just because Warfare server sizes are just that much more.

The higher rated a player is, the more Elo pts you get for killing him, the less he gets for killing people with lower ratings. Dominant players wont' gain much from massacring a noob server, they may even lose a little if they manage to get killed.

This is how it worked in UT2004 stats and it still appears to work the same way now - except the Elo rating now gets affected by activity. Inactivity will cause you to lose Elo rating (thus rank) - this fixes one of the most annoying aspects of UT2k4 rankings - some players would rack up #1 rankings, quit their "player", start a new one, and continually spam the ranks with their new "players".

[bclagge]

It's actually kinda simple, if you know how it works. For a more detailed explanation, check the wikipedia for ELO.

It's based on an ELO rating system. You can check yours at enemydown.co.uk. Higher is better. Here's how it works: If you kill someone with a much better rating than you, you will improve a decent amount. If you kill someone with a much worse rating than you, you see almost no improvement. The same is true for being killed by someone, it hurts your rating depending on what theirs is related to yours. Also, killing someone of a much higher rating is better than killing someone only a little better.

Therefore, kill/death ration means nothing. It's not about how many you kill, it's who you kill.

Edit: hehe, beat me to it

[bclagge]

Trepænima, I'm a little confused by your paragraph about wins. Are you saying winning is related to ELO rating? I thought it was kills only.

Also I think you're wrong about players atrophying with inactivity. Mysterial and Strippedoflife nailed the #1 and #2 rank with a rating of 1930 weeks ago. They seem to have stopped playing and have maintained the #1 and #2 rank until recently when everyone else caught up and then passed them. I'm 16 points behind them so I wish what you said was true, but I don't think it is.

[Elduke]

Thanks for the info. Seeing as they have the rank system like this then there should be a match making option so that you are playing more rank friendly opponents!

[Trepænima]

Trepænima, I'm a little confused by your paragraph about wins. Are you saying winning is related to ELO rating? I thought it was kills only.

Kills = Wins. Efficiency means nothing, you gain more points by killing the top guys, than dying from them.

Also I think you're wrong about players atrophying with inactivity.

You're probably right, but I know rank drops with inactivity, haven't paid much attention to rating.

Elo is best for comparing one player to another (as in a Duel situation). When you place first in the map, you gain a "win" over every single person you outscored, so #1 guys gain TONS on large player-count gametypes, like Warfare.

The problem with the current system is that all players are grouped together. It'd be nice if the rankings discriminated between gametypes. Example - most of the top 10 guys right now at enemydown are regular warfarers.

[Mr Evil]

I wonder if you could manipulate the system like this:

Create an account, call it "Fodder". Fodder goes out and plays as normal, rising through the ranks as far as possible. Now create a second account, call it "Leech". Every so often Leech and Fodder join a private game together and Leech kills Fodder as much as possible, increasing Leech's rating. Because Leech never plays anyone else, there is no risk of losing rating. You could even go one level further and have a "MetaLeech" who does nothing but kill Leech, to boost the final rating even higher.

It might take quite a bit of work to get to No. 1, but it could be done by anyone, regardless of actual skill. I would try it, but I think it would need two CD keys.

[bclagge]

Kills do not = Wins in Warfare which is what I play. Sure if you have a lot of kills you're more likely to be on the winning team, but they are not directly connected. Even for DM they're not equal. A highly rated player playing a bunch of zeros will go down in rating if he dies even once.

I know ELO isn't the ideal system for this kind of thing but it's what we're using. And it's my understand of ELO that you change your rating by killing or being killed. I don't think you gain anything in the rank department from winning (WAR).

I've noticed one thing in the last month, and it's that the ELO rating of the player in first is getting higher and higher. Slythe the current leader has a rating of appr. 2050, whereas two weeks ago Mysterial held it with 1930. The bar is constantly being set higher, and I don't think it's going to peak for some time. You lose ranking through inactiviy because people pass you by if you don't play.

You're right about the top ranks being WAR players. The top eight players were all tearing it up today in a UMG League Warfare server. We meet there almost every night if anyone wants to come join the fun!

@Mr_Evil: I was thinking something similar today I wondered if myself and another player could duel and kill 1:1. In theory both our ratings will go up if we're about equal right?

[Trepænima]

Kills do not = Wins in Warfare which is what I play.

When I say "wins", I mean it in the chess fashion. In Warfare, you kill a guy, BOOM, that's a "win" for you. Over the course of the match, you "beat" him 5:3, you "won" more than he did. Kills are treated as single encounters vs. an opponent, not as totals, so you don't get one win from that 5:3 match, you get 5 wins vs. 3 losses in Elo terms. You also get Elo bonuses from score, number of nodes captured, nodes/cores destroyed, etc. There are many more bonuses to be had in Warfare than there are in DM, hence the skewed ranks.

The bar is constantly being set higher, and I don't think it's going to peak for some time.

It'd be interesting to see the ceiling, Elo chess ceiling is just about 2800, and although a couple people have surpassed it, it's rare.

[btcomm]

Kills = Wins. Efficiency means nothing, you gain more points by killing the top guys, than dying from them.



You're probably right, but I know rank drops with inactivity, haven't paid much attention to rating.

Elo is best for comparing one player to another (as in a Duel situation). When you place first in the map, you gain a "win" over every single person you outscored, so #1 guys gain TONS on large player-count gametypes, like Warfare.

The problem with the current system is that all players are grouped together. It'd be nice if the rankings discriminated between gametypes. Example - most of the top 10 guys right now at enemydown are regular warfarers.

There is only 1 person that plays warfare that is up there in ranks I think.

Maybe a couple but most are DM players.

[theeDEATHMASTER]

It doesn't.

omg I said out loud this JUST before clicking on the thread lol.

[Ikaros]

Well, I still like to track EFF so see if I'm improving or not. I don't care about the ELO and ranking with others. We're not playing Chess one on one!

In athletic sports you rate on a level playing field with everyone's personal stats and don't handicap your own personal stats on how good an individual opponent is. You rate yourself on how good you do.

In general you personally do better when you kill more than be killed. So what I do and take my kills and death totals and the EFF formula would be EFF=Kills/(Kills+Deaths).

[bclagge]

When I say "wins", I mean it in the chess fashion. In Warfare, you kill a guy, BOOM, that's a "win" for you. Over the course of the match, you "beat" him 5:3, you "won" more than he did. Kills are treated as single encounters vs. an opponent, not as totals, so you don't get one win from that 5:3 match, you get 5 wins vs. 3 losses in Elo terms. You also get Elo bonuses from score, number of nodes captured, nodes/cores destroyed, etc. There are many more bonuses to be had in Warfare than there are in DM, hence the skewed ranks.

It'd be interesting to see the ceiling, Elo chess ceiling is just about 2800, and although a couple people have surpassed it, it's rare.

Well if Trepænima is right, then the top ranking players are not just the best at killing, they are also benefiting from node capturing etc. That would make it pretty accurate at defining who's better and worse all around. Not just straight DMing. Which does beg the question why aren't the gametypes separate? If Warfare players really are getting extra bonuses for objectives, then the DM players are handicapped (assuming you care about your rank).

btcomm: Just tonight alone I've played with 13 players in the top 20 all in a Warfare room. Maybe they play DM too, but I see them in Warfare. The top 7 ranks are regulars in the Warfare room I frequent (counting myself )

Trepænima: I would love to know how the ELO scoring breaks down. Do you have anything you can link?

[FACEST0MPER]

Whatever it is, its retarded. My rank went from 500 to 100 to 26 to 100 to over 1000 in 4 play sessions. Right now it seems to be hanging out somewhere between 100 and 500.

It changes way too much IMO. They need to give something like number of kills or number of matches won some weight so it doesn't fluctuate so much.