Ultimate EOL Ranking discussion

Feature requests and ideas for the new unofficial versions of Elma and general talk related to those.

Moderator: Moporators

User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Ultimate EOL Ranking discussion

Post by Kopaka »

We've had some ranking discussions in other topics here and there, so creating this topic collect it in one place and make it clearer that such discussion is going on.

Since introducing the current ranking system (the main one called Ranking on that page), I've heard many complains. People not understanding it or being low placed even though they are pretty pro. Complains I pretty much agree with. With the current system you cain points according to how many people you beat in a battle and how well placed they are(in ranking). The points you get from each battle is saved and the best 200 from each six months counts towards your ranking. Which means if you play more than 200 in six months, your "worst" results won't count. Also you might get more from a 3rd place in a battle with a lot of half decent players, than being 1st in battle with a few really good ones. It's quite possible to make it more understandable, by showing the exact points you get from each battle and which of these counts, but that doesn't remove the other problems.

After a discussion in another topic stini started working on a system, dunno of his progress, anyway from my understanding his would be more like the current relative ranking we have. This was discarged as the main ranking as you can lose points by playing, and that tends to discourage people from playing, especially if you have already maintained a good ranking.
I also have some ideas myself for a better system.

But before we talk about the specifics of the ranking system, there is another question to be answered.
Question 1: Should we make a new ranking system (and hence delete the current)
This is an important question to answer first because making a new would of course remove the old and the stats that we currently have. Meaning battler of the year 2010 and 2011 would probably change, as would people's current position, battler of the X achievements etc.

I'm making this one question at a time, so to have a clear debate that we can draw some conclusions from. So try to keep to the current question (:
User avatar
abruzzi
Kuski
Posts: 1425
Joined: 17 Sep 2007, 21:07

Re: Ultimate EOL Ranking discussion

Post by abruzzi »

just make elo rating like in chess, everybody will be satisfied, 3/10 loses with 1 and 2 and wins with rest
<Pawq> at a gym you have only 3 options: 1. have your eyes closed, 2. stare at yourself, 3. stare at others, all of which are either super boring or disgusting
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Ultimate EOL Ranking discussion

Post by Lukazz »

yes, change the ranking pls. i understand your explanation of the current system, but it still feels extremely random. in the beginning i was like 20th and stayed on that position for months, no matter how much i played. then within not much time i became better and then was like 8th for a very long time, although i hardly played.

but pls pls pls no system where you can lose points!!
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

Yes I think ELO is the way to go and you should lose more points for sucking in general and for losing to players ranked lower than you. It's tough but without that what's the point of rankings anyway? It's just grinding exp then, which is pointless. ELO discouraging players battling in fear of losing points is a false myth. Current system is a joke, let's try something that's proven to work in countless analogue situations and see what happens. I like the external ranking system of Quake Live. www.qlranks.com
39:37,91
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Ultimate EOL Ranking discussion

Post by Lukazz »

Xarthok wrote:ELO discouraging players battling in fear of losing points is a false myth.
How do you come to that conclusion? Do you really want to compare chess with EOL? Of course, when playing in a chess championship you try to win every game, but in EOL some people - including me - just come online and want to cruise a few battles very often. And if I'd lose points there, I'd definitely skip some battles, especially when I join when there are only a few minutes left or TL starts one of his 60 mins battles and I know I won't play the whole time.
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

If you only come to cruise why the fuck do you give a fuck about rankings? They're simply supposed to be the closest thing to a realistic representation of one's skill level, which the current system is most certainly fucking not. It's not the point of the game.
39:37,91
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Ultimate EOL Ranking discussion

Post by Lukazz »

Xarthok wrote:If you only come to cruise why the fuck do you give a fuck about rankings?
Good point.
Do whatever you want. I'm addicted so I'll play dozens of hours every week anyway.
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
BlaZtek
Kuski
Posts: 391
Joined: 15 Dec 2011, 01:07
Team: CMR
Location: Norway

Re: Ultimate EOL Ranking discussion

Post by BlaZtek »

I only care for Kopasite Kinglist <3 Fuck balles!
Image
[13:22:24] (ImagebEAT) blaz super pro
[17:06:15] (ImageRamone) ok I suckj. blaz rules
User avatar
ribot
Not banned
Posts: 2416
Joined: 19 May 2002, 16:20
Location: Miranda: the true state
Contact:

Re: Ultimate EOL Ranking discussion

Post by ribot »

I think it's pretty common that people don't battle all the levels they try. Some levels are of very bad quality and why should you want to play them to care about ranking?

A ranking system should benefit the collective attitude, imho. In Xarthok's opinion then, why should only stats freaks have high ranking?

A problem arises when zweq tries one bulle and doesn't play seriously. Then WALDO beats zweq this round. Should he really get extra points in that case?

I think that we can agree with a few things however, and we could start from there. If there are very few players in one battle, it is not a great achievement to win such a battle. Then, it is quite normal that not all who try the battle will play it fully. Therefore a high ranking should be achieved when winning a battle against many players. I don't think mediocre results should need any points at all. Only achievements need to be rewarded with ranking points.

Since there is already a lot of statistic data, there could be some tweaking with different point systems to see how it would position certain pro players. A question arises here, of course, about how complex Kopakar is willing to code the point system.

I'm thinking about exponential formula1 points (according to sega megadrive super monaco gp):
1st place - 9 points
2nd place - 6 points
3rd place - 4 points
4th place - 3 points
5th place - 2 points
6th place - 1 point
All the rest get 0 points.

The weakness of this system is of course if there are too few players in an bulle. I think this can be fixed with something like if there are only 10 players, winner gets 4 points, 2nd gets 3, 3rd 2, 4th 1. If there are 15 players the winner gets 6 points, 2nd gets 4, 3rd gets 3, etc (always using the same scale, but not always counting from the top). Only if there are maybe 20 players in a bulle can the winner get 9 points. If there are only 2 players in an bulle both get 0 points (it's so easily cheatable otherwise), and if there are 3 players in an bulle: only winner gets 1 point.
Last edited by ribot on 6 Aug 2012, 20:05, edited 1 time in total.
-
"leader status in the Elma against-the-system underground" - Abula
-
IncrElastoMania - Elma Simulation - Browser Game 2020
Elma Imager - Command Line Tool 2020
User avatar
Madness
35mins club
Posts: 2168
Joined: 1 Jan 2009, 10:51
Location: UK

Re: Ultimate EOL Ranking discussion

Post by Madness »

I agree with Xarthok, it sounds reasonable to lose points for losing. But first of all, it's important to exclude all those random battles from ranking. For example battles with a balletime of under 10 minutes and those stupid idiot modes: See others, No turn, No throttle, No brake, One wheel, Apple bugs, Multi, One turn, Always throttle, No volt, Drunk. Also you shouldn't lose points to someone with the same result just because they did it first. Just my 2 cents!
Image
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

Madness wrote:stupid idiot modes: (..) Multi
ಠ_ಠ
39:37,91
User avatar
Zweq
34mins club
Posts: 4055
Joined: 28 Nov 2002, 15:54
Location: suo mesta

Re: Ultimate EOL Ranking discussion

Post by Zweq »

yea need time to test every style with saveload :roll:
Image
molama
Beginner kuski
Posts: 2
Joined: 13 Nov 2010, 08:50

Re: Ultimate EOL Ranking discussion

Post by molama »

I guess it should be like this if 5 players played:
1. 2P
2. 1P
3. 0P
4. -1P
5. -2P

a basic ranking of 10, and after getting higher up its more and more difficult to earn points.

Eather way how the ranking should be. Wold i loose points just staing online on eol. Without finishing?

Hey Madness, One wheel is cool !
User avatar
Pab
38mins club
Posts: 887
Joined: 4 Apr 2008, 23:46

Re: Ultimate EOL Ranking discussion

Post by Pab »

if u lose points then lukazz ranking would be -100000000 points, he always enter lev and dont play (and ends up last in results)
but right now he is 7th in ranking
My Youtube Channel Level Stats Maker Battle Notifier!

Image
Image
Signatür ruined by SveinR - smaller plz :*
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Ultimate EOL Ranking discussion

Post by Lukazz »

Pab wrote:if u lose points then lukazz ranking would be -100000000 points, he always enter lev and dont play (and ends up last in results)
but right now he is 7th in ranking
true story!
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
Stini
39mins club
Posts: 202
Joined: 5 Dec 2002, 22:15
Team: ICE
Location: Helsinki, Finland

Re: Ultimate EOL Ranking discussion

Post by Stini »

I have pretty much implemented a very basic ELO based rating system some months ago already. Maybe I should finish it and put some results for everybody to see. What is left to do is better handling of new kuskis (Glicko would be better for this but orka) and estimate some parameters from the data etc.

I considered a battle like a chess tournament where everyone plays against everyone and you win your "game" if you get a better time than your opponent. The problem with this approach is that in chess tournaments the results are quite independent, so you might have one very good game but that doesn't mean you are going to win the next game as well. In battles if you make a nice run, you win all your "games". With a bad run you lose many games, which might make a severe blow to your ranking, but I guess choosing a wider probability distribution fixes this to some degree. So theoretically this is not a very justified approach, but in practice it might work decently well and my first test seemed pretty ok, I think Markku was on top and I was better than Spef o/. Something like TrueSkill would be better theoretically, but it's also a lot more complicated to implement. If you can't assume the games to be independent, the math gets a bit harder if you want to do it right.

With this model, your ranking suffers a lot if you tend to give up battles, but I guess those who care about their rankings will try harder. This model "forgets" old results eventually so your ranking 100 battles ago doesn't really affect anything. So you can just stop giving up battles and your ranking returns to "normal".

I also put some random limit if you play less than 3 mins or so, your rating doesn't change, so you can try a lev a bit to check if you like or not. There are probably some other elma specific details to consider as well.
User avatar
Igge
38mins club
Posts: 6393
Joined: 7 Apr 2007, 12:15
Location: Stockholm, Sweden

Re: Ultimate EOL Ranking discussion

Post by Igge »

There shouldn't be a fixed time for checking level though, it should be based on balle length. If the limit was 3 mins, then you could play almost 1/3 of a 10min balle, but if it's a 30 min balle you'd only be able to play 1/10th which might not be enough to fully check it out.

Basically, like so many others have pointed out, battles in elma have many more factors to them than, say, chess. In chess you play a game. In elma you can join whenever in a battle, and the battle can be whatever format, and you might not even like the level so you'll quit. There's so much more to battles in elma when it comes to the play-style, making a set of rules is a lot harder and should be done a lot more carefully as to not discourage people from playing the game.
John: lol hittade ett popcorn i naveln
(19:52:06) (@Madnezz) The Golden Apple Award goes to.....
(19:52:36) (@Madnezz) ib9814.lev by igge!!!
Zweq wrote:99.9999% of nabs haven't even opened the book yet and most of those that have are still on the first pages
User avatar
elmas
Kuski
Posts: 69
Joined: 21 Apr 2012, 17:16
Location: Viitasaari, Finland

Re: Ultimate EOL Ranking discussion

Post by elmas »

ribot wrote:I'm thinking about exponential formula1 points (according to sega megadrive super monaco gp):
1st place - 9 points
2nd place - 6 points
3rd place - 4 points
4th place - 3 points
5th place - 2 points
6th place - 1 point
All the rest get 0 points.

The weakness of this system is of course if there are too few players in an bulle. I think this can be fixed with something like if there are only 10 players, winner gets 4 points, 2nd gets 3, 3rd 2, 4th 1. If there are 15 players the winner gets 6 points, 2nd gets 4, 3rd gets 3, etc (always using the same scale, but not always counting from the top). Only if there are maybe 20 players in a bulle can the winner get 9 points. If there are only 2 players in an bulle both get 0 points (it's so easily cheatable otherwise), and if there are 3 players in an bulle: only winner gets 1 point.
This best so far
User avatar
zebra
Kuski
Posts: 1009
Joined: 23 Sep 2003, 15:35
Team: TAP
Location: Finland
Contact:

Re: Ultimate EOL Ranking discussion

Post by zebra »

- Any system where only fixed points are given (like in the previous post) is bad, because the number of participants vary very much between battles.
- I still think that you shouldn't lose points by playing. The reason is that not every battle you play is serious. There are lots of reasons why you can quit battle. The most common reason is of course that you try the level 2 or 3 times and then see it's very boring and you quit elma and go do something else.
- This "unseriousness" is the main problem here. In chess you play the same game for hours and nobody can really say that you are not seriously playing. But in elma you might come late in battle and have a phone call in the middle of the battle etc. There should be a way to tell if a kuski is really participating in the battle or not.
- If there is a limit (like 3 minutes or 3 tries) which do not yet count, what about zweq making the best time with first try?
- I think there are not any problems with first finish battles, only normal battles. In ff battles you "have" to play seriously and be present from the start of the battle if you want to win.
- This current system (as Kopaka described it) sounds good. What is really the problem there?
- The problem might be that if a good player gets a bad result (not playing seriously), the noobs who beat him, will get easy points.
- If you want to change the ranking system, you can still do so, but I don't really see any reason to do that. If I look at the ranking page, I see only the pros at the top of the list (the kuskis I expect to see there) :)
A winner of 4 GAA's (mc2 included), winner of mkup206, and a proud member of team TAP.
Play uni levels: http://koti.mbnet.fi/zebra/uni.html
Homepage: http://koti.mbnet.fi/zebra/elma.html
User avatar
Madness
35mins club
Posts: 2168
Joined: 1 Jan 2009, 10:51
Location: UK

Re: Ultimate EOL Ranking discussion

Post by Madness »

zebra wrote:- I still think that you shouldn't lose points by playing. The reason is that not every battle you play is serious. There are lots of reasons why you can quit battle. The most common reason is of course that you try the level 2 or 3 times and then see it's very boring and you quit elma and go do something else.
- This "unseriousness" is the main problem here. In chess you play the same game for hours and nobody can really say that you are not seriously playing. But in elma you might come late in battle and have a phone call in the middle of the battle etc. There should be a way to tell if a kuski is really participating in the battle or not.
There's no way to say this. One can start playing when there are 3 mins left and another one can train the level in editor and start "playing" when there are 3 mins left. Why would someone, who doesn't play seriously, care about rankings anyway.
Image
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

1. Leave the system as it is or slightly alter it. Nobody loses points and "everybody" is happy. In return, the "ranking" doesn't represent skill but how much one plays, instead. Like XP in RPG games. People who are good but don't battle as often are placed lower than people who aren't winning as much but simply play more often. Which defeats the whole purpose of it if you ask me. Nobody advocating a "never lose points" system has presented a reasonable defence yet.

2. Or implement some form of ELO or TrueSkill or whatever fits better. People who play "casually" or "not seriously" or just enter the battle and leave a few mins later or join late will be at a disadvantage but if they don't play seriously they shouldn't expect to be treated like everyone who does play seriously. There can remain a parallel "experience" ranking like it is now. If anything, this should encourage (serious) players to battle harder and not half-ass their effort, which isn't a bad thing. Even if you aren't as good yet and end up with an ELO of under 1000 (starting point), at least you know where you stand and you will be able to feel real progress as you get better in the future.

Pick one.
Last edited by 8-ball on 7 Aug 2012, 11:24, edited 1 time in total.
39:37,91
User avatar
Hosp
38mins club
Posts: 2105
Joined: 30 Aug 2009, 20:55
Team: MiE
Location: up yours
Contact:

Re: Ultimate EOL Ranking discussion

Post by Hosp »

I pick 2. )
Image
ToMaT
User avatar
Stini
39mins club
Posts: 202
Joined: 5 Dec 2002, 22:15
Team: ICE
Location: Helsinki, Finland

Re: Ultimate EOL Ranking discussion

Post by Stini »

Igge wrote:There shouldn't be a fixed time for checking level though, it should be based on balle length. If the limit was 3 mins, then you could play almost 1/3 of a 10min balle, but if it's a 30 min balle you'd only be able to play 1/10th which might not be enough to fully check it out.
Yes, I'm aware of this, but I don't think percents are any better. In a 30 min battle, 10 mins way more than enough to decide whether you like the lev or not. In 10 mins you might get a lot of information, like your style is wrong and you don't know the right one and you'll probably get a bad result so you can give up just to avoid losing points. I think it's better to give 3 mins or something that is just enough to decide if you like the lev, but you don't have much time to predict how well you'll do.

Instead of using a hard limit like 3 mins, it's also possible to use some function (perhaps a linear function or some logistic curve) so that how much your ranking changes depends on how long you played the battle.

Sure it's also possible to use ELOs just as skill estimates and calculate rankings with some other method that takes into account how much you play etc. Personally I find it more interesting to know the skill of the kuski quantitatively, I don't really care about ranking. You can see your own skills improve and you can calculate stuff like how likely it is for you to beat Zweq if he plays seriously. Just being able to compare the skills of the kuskis quantitatively would be quite max, just like in chess I can see from ratings how much better player A is compared to B or how much someone has improved in a year.
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

ELO progress graphs would be seriously awesome. Seriously.
39:37,91
User avatar
Pab
38mins club
Posts: 887
Joined: 4 Apr 2008, 23:46

Re: Ultimate EOL Ranking discussion

Post by Pab »

Can win a battle in 10 minutes on 30minutes battle, or First try ez
My Youtube Channel Level Stats Maker Battle Notifier!

Image
Image
Signatür ruined by SveinR - smaller plz :*
User avatar
Igge
38mins club
Posts: 6393
Joined: 7 Apr 2007, 12:15
Location: Stockholm, Sweden

Re: Ultimate EOL Ranking discussion

Post by Igge »

EOL
ELO
LOEL
John: lol hittade ett popcorn i naveln
(19:52:06) (@Madnezz) The Golden Apple Award goes to.....
(19:52:36) (@Madnezz) ib9814.lev by igge!!!
Zweq wrote:99.9999% of nabs haven't even opened the book yet and most of those that have are still on the first pages
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Ultimate EOL Ranking discussion

Post by Kopaka »

Everyone (pretty much) is talking about new systems, so I take it you don't mind potentially erasing the current stats etc.

Must say I'm beginning to warm up to the idea of a relative/ELO sort of ranking as the main one. Although I still have some ideas I wanna try to tweak the current main.The current main uses battles up to 2 years old, the thought was to see how you improve over time, but it might just mean that you are stuck with your bad results for 2 years before you can really reach the top. A problem with ELO/relative ranking is inflation. Some new player joins and get's his 1000 start points, he sucks so he loses a lot and those points are given to the best players, and maybe he quits before getting over 1000 again, which means top players have more points to divide among themselves. So a 2500 ranking in 2010 might be better than a 2500 ranking in 2011. Players with over 1000 will gain points little by little even if they don't improve skills, and historical graphs are useless. Dunno if this is really a problem though, but they do talk about it on ELO wikipedia page aswell. There will also be new pro players who takes points away from top players.

As far as the whole "not playing seriously", "playing only 2 mins" problem goes, we were talking about it in #kopasite earlier and some nice ideas sparkled from that. I find it an important issue, like I said:
(@Kopaka) I find myself not wanting to join a battle midway even without thinking about ranking, having to wait for new battle is one of the "problems" with elma battles imo, so shouldn't be more reasons to not join a battle midway or stop playing midway
The idea:
(@Markku-) should be something like: good place with little playing time, counts normally or even get extra points. bad place with little playing time, counts less.
(@Markku-) or maybe not extra points, then nabs play 90% of the time in editor and come to do one good time
More technically it could be something like this in a relative/ELO ranking:
- percentages here means that the player gets/loses that percentage of what he would normally have gotten/lost, and other players gets/loses that percentage of what they would normally have gotten for winning/losing to him
- If you gain positive points you count 100%
- If you gain negative points:
- - If playingtime > 50% = counts 100%
- - Otherwise counts ((playingtime * 100) / battletime)% (maybe times 1.25 or 1.5 or such to not have such big jump from 49% to 50% playtime)
- For example if you play 30% of time and gain +30 you get +30
- For example if you play 30% of time and lose -30 you lose -9
- 20 mins battle, play 3 mins, means 15% for example

A potential problem:
(Madness) what if i realize after 3 mins that i can't win the battle and stop playing?
But dunno if people really cares enough to speculate in it like that.

Xarthok wrote:I like the external ranking system of Quake Live. http://www.qlranks.com
There's no explanation of the ranking there as far as I can see? It would be nice to see, if nothing else for some ideas, as quake is more comparable to elma than chess (you play many against many).
Madness wrote:But first of all, it's important to exclude all those random battles from ranking. For example battles with a balletime of under 10 minutes and those stupid idiot modes: See others, No turn, No throttle, No brake, One wheel, Apple bugs, Multi, One turn, Always throttle, No volt, Drunk. Also you shouldn't lose points to someone with the same result just because they did it first. Just my 2 cents!
Except see others those are already excluded (multi was/is meant to have it's own battletype like FF etc. but it was insanely complicated to do). And I agree with Jappe2 that for times, first = better, as it has always been in elma. For apple results I agree with you, they should count same. Atm everyone with 0 apples get's 0 points, but for other apple results it's first = better, simply because that was simpler to do.
molama wrote:Eather way how the ranking should be. Wold i loose points just staing online on eol. Without finishing?
If you mean a 0 apple result then yes you do in current relative ranking lose points, and get 0 in main ranking. Some levs may have none or few apples and some people with 0 apples result may have played a lot, so winners should still get points for winning them. If you are not in results nothing happens of course :P
User avatar
Madness
35mins club
Posts: 2168
Joined: 1 Jan 2009, 10:51
Location: UK

Re: Ultimate EOL Ranking discussion

Post by Madness »

Kopaka wrote:More technically it could be something like this in a relative/ELO ranking:
- percentages here means that the player gets/loses that percentage of what he would normally have gotten/lost, and other players gets/loses that percentage of what they would normally have gotten for winning/losing to him
- If you gain positive points you count 100%
- If you gain negative points:
- - If playingtime > 50% = counts 100%
- - Otherwise counts ((playingtime * 100) / battletime)% (maybe times 1.25 or 1.5 or such to not have such big jump from 49% to 50% playtime)
- For example if you play 30% of time and gain +30 you get +30
- For example if you play 30% of time and lose -30 you lose -9
- 20 mins battle, play 3 mins, means 15% for example
Please don't adjust the points to the playing time in any way. It will only lead to everyone playing in the editor for most of the battle.
Image
User avatar
elmas
Kuski
Posts: 69
Joined: 21 Apr 2012, 17:16
Location: Viitasaari, Finland

Re: Ultimate EOL Ranking discussion

Post by elmas »

Im starting to feel that we should keep this old system
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

Yes, old system (current main) should stay but the relative (ELO) ranking should be made "main" and then most central/important one with graphs etc. Inflation is much less of an issue in ELO than the other systems we have.
39:37,91
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Ultimate EOL Ranking discussion

Post by Kopaka »

So the question may be how much time and effort is it worth using to fine tune the relative ranking. There are things like these to consider:
How big amount of points you can win/lose in 1 battle
How much should amount of people in the battle mattered relative to how much their position matters when calculating points
Should we have some sort of you play less than X mins you don't count/count less
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

I think it would backfire to implement any sort of "play only a few mins and lose less or gain more points" because people would practice in editor/saveload.

If you enter the battle you should be expected to battle seriously like everyone else if you care about relative ranking, up to you if you only want to play a few minutes or start late.
39:37,91
User avatar
abruzzi
Kuski
Posts: 1425
Joined: 17 Sep 2007, 21:07

Re: Ultimate EOL Ranking discussion

Post by abruzzi »

PROTIP: listen to xarthok
<Pawq> at a gym you have only 3 options: 1. have your eyes closed, 2. stare at yourself, 3. stare at others, all of which are either super boring or disgusting
Luther
37mins club
Posts: 873
Joined: 20 May 2002, 21:20
Team: ICE
Location: Sweden - Gothenburg

Re: Ultimate EOL Ranking discussion

Post by Luther »

It is impossible to climb in ranking no matter what I do... Doesn't matter if players haven't played for 24 months, there is no way to reach any of them. 700 wins ago my placement was higher than it is now

Would be fun if something actually happened to the ranking
Nekit for president
User avatar
Madness
35mins club
Posts: 2168
Joined: 1 Jan 2009, 10:51
Location: UK

Re: Ultimate EOL Ranking discussion

Post by Madness »

That's because you win at the time when elmers are working. :D
Image
Luther
37mins club
Posts: 873
Joined: 20 May 2002, 21:20
Team: ICE
Location: Sweden - Gothenburg

Re: Ultimate EOL Ranking discussion

Post by Luther »

Yes, but not always :) I think you understand what I mean anyway :)
Nekit for president
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Ultimate EOL Ranking discussion

Post by Kopaka »

As mentioned here I should hopefully soon get around to do this. So if you have more input now is the time. I would especially like some input on these things:

In the current relative ranking there is a big different between winning a battle with 10 players and one with 20 players, should there be?
Should it matter a lot or a little how much the different in ranking is between you and the ones you beat?
How big percentage of your current ranking should you lose/win in a battle?
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

I feel the relative ranking changes a tad bit too quick - a single battle shouldn't have THAT big of an influence.
39:37,91
User avatar
Bludek
38mins club
Posts: 1725
Joined: 23 Sep 2009, 10:56
Team: CART
Location: Some pub in Prague

Re: Ultimate EOL Ranking discussion

Post by Bludek »

Kopaka wrote:In the current relative ranking there is a big different between winning a battle with 10 players and one with 20 players, should there be?
Should it matter a lot or a little how much the different in ranking is between you and the ones you beat?
How big percentage of your current ranking should you lose/win in a battle?
1. NO, coz it disadvantages players, who cannot (or just don't) play in the primetime (~22 EET). It should be imprortant A LITTLE, but deffinitely not so much it is now.
2. It should be considered too, but I think better place in balle should just give you more points. But this point is deffinitely more important than 1st point.
3. If I understand the q correctly, my answer is: winning or losing a balle should not care about you rating at all. It should give you some points and that's it. More points = better rating.
edit 3rd q: I just realized it would be impsy for newer players to catch the veterans... I don't know then :?
Image
EOL top 10 kuski of 2014 and 2015.
21:03:48 <umiz> i like 99% of bludek levels
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

Bludek it doesn't seem like you understand what relative ranking is.
39:37,91
User avatar
Bludek
38mins club
Posts: 1725
Joined: 23 Sep 2009, 10:56
Team: CART
Location: Some pub in Prague

Re: Ultimate EOL Ranking discussion

Post by Bludek »

Oh wow, I was talking about normal ranking the whole time. Sry,my mistake.
Note to myself: read question first and THEN answer to it.
Image
EOL top 10 kuski of 2014 and 2015.
21:03:48 <umiz> i like 99% of bludek levels
Thundr-
37mins club
Posts: 398
Joined: 10 Jun 2010, 17:02

Re: Ultimate EOL Ranking discussion

Post by Thundr- »

One point system i know of thats used im many competative games are such: If you enter in a match and the opposing team/player has a high score/rating, winning gives you more points, and you would loose less points for losing and vice versa.

To incorporate this into elma would perhaps be something like if there are a lot of pathetically low ranked newbs playing battle, and then all of a sudden a high ranked player such as Zero would come in and steal the win, he wouldnt gain as much points as if it were a battle with many high ranked players in it.

For example:
1: Zero - 11.44
2: Noob1 - 12.35
3: Noob2 - 14.43

The points awarded to Zero isnt as high in this battle, but the lowly ranked noobs still get a fair amount of points as they are currently at the bottom of rankings, this is a good motivation for noobs to join battles even if there are many pros in it for example and they feel they have no chance. If you drop really low, its easier to get back up (talking about the very bottom now)

For example #2:
1: Zero - 11.44
2: Zweq - 12.35
3: Markku - 12.43

In this scenario the points awarded for each players is relatively high, or higher compared to example 1.

Using this method eliminates such arguments with a set number of points for each position in battle (if there is only 1 player participating he gets 10/10 points even if time sucks etc). Points dont increase that much depending on how many players are in the battle, but would increase if there are many good players in it. It also doesnt matter what position in battle you have. Winning is of course nice and all, but for those playing at late night when there are no players online it would be many ez wins and high rating for that person, which in reality might not actually reflect his actual skill or prowess. This system eliminates that which I think is good. Winning vs many pros or placing good vs many pros is more valuable than winning versus 3 random super low ranked players.

Not saying you should definately use this method or that it doesnt have any flaws cuz just like any other system it has several. But this is what I know is used, with some variation, in many pvp based games such as starcraft 2/world of warcraft pvp/leauge of legends etc. I dont have any idea for ppl that just come to check out battle and leave though, their ranking would be the same, but the overall ranking for others may change due to it.
User avatar
elmas
Kuski
Posts: 69
Joined: 21 Apr 2012, 17:16
Location: Viitasaari, Finland

Re: Ultimate EOL Ranking discussion

Post by elmas »

I always thought ranking is like this. But if isn`t, best idea so far
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Ultimate EOL Ranking discussion

Post by Kopaka »

Yeah that's pretty much how the relative ranking works, just needs some tweaking.
Thundr-
37mins club
Posts: 398
Joined: 10 Jun 2010, 17:02

Re: Ultimate EOL Ranking discussion

Post by Thundr- »

Its just that there are so much stuff displayed under the Ranking tab, you have experience points, relative ranking, and ranking. Narrow it down to just one set of ranking points displayed, and clicking on a player shows the other stuff for example win/loss ratio and total games played and relative ranking. Right now its like Zweq has highest ranking, Jeppe highest relative ranking, Eddi most experience points, and none of this is explained on the site so people get confused. Perhaps change the name of relative ranking to matchmaking score or something to make it more distinct.

Simple explanation example:
Ranking: is calculated by a number of values determined by a players relative ranking and etc etc
- Relative ranking: takes into consideration the amount of players in a battle and respective players calculated skill/experience. The more skilled players in a battle, the higher your relative ranking gets for being victorious or placing good etc
- Experience etc etc..

Have something like this ABOVE the entire list of players on the site under the ranking tab maybe.
User avatar
Igge
38mins club
Posts: 6393
Joined: 7 Apr 2007, 12:15
Location: Stockholm, Sweden

Re: Ultimate EOL Ranking discussion

Post by Igge »

Yep. There should be only one main ranking, and it should not be solely based on either amount of battles played, nor win-to-loss ratio, but rather a combination of everything like you said. I don't know how hard this would be to make, but surely there must have been a similar system for some sport or w/e at some point in time?
John: lol hittade ett popcorn i naveln
(19:52:06) (@Madnezz) The Golden Apple Award goes to.....
(19:52:36) (@Madnezz) ib9814.lev by igge!!!
Zweq wrote:99.9999% of nabs haven't even opened the book yet and most of those that have are still on the first pages
Thundr-
37mins club
Posts: 398
Joined: 10 Jun 2010, 17:02

Re: Ultimate EOL Ranking discussion

Post by Thundr- »

Clicking on Zweq now you see this:
1st ranked with
26094.7 Points
14th battle experience ranked with
35084 Points
1030 Battle Wins
Best Winning Row
6 Wins

Points points experience wins row wins ranked ranked numbers numbers. Organize this, and highlight thats its acutally the Ranking points that is the mother and the rest are just variables that determines the rest. Also the way its constructed now its basically saying - If you did not win, you lost, which is incorrect as it has already been declated that placing 2nd is different from placing 12th. Hence there is no point in putting so much focus on wins/losses, win% etc on this slate.

Basically what I am suggesting:

Clicking on ranking tab you have the following information displayed:
Player [maybe team also, or nationality flag etc] and then Ranking. just those two or very limited for a nice clean table.

Clicking on a player and entering his profile displays the following information in a more organized fashion:
All the other information that players might be interested in + his relative ranking which should NOT have any ranked position like it has now. The number may be displayed but not like "ranked 14th player in relative ranking, (even though he is ranked 1st in real ranking)" Relative ranking isnt ranking, so it shouldnt be ranked in any way at all - its a variable that more or less determines how valuable it is to place above this person in a battle, and if I will receive many or few points for doing that. Having this ranked like it currently is has just contributed to adding another type of ranking that people obsess or whine about.
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Ultimate EOL Ranking discussion

Post by 8-ball »

I still think pure ELO ranking would be best as the main ranking and would best represent who really are the best in this game. Most competitive games use this. If people don't like losing points when they perform poorly in a battle then, well, more motivation for them to improve and play seriously (unless they don't care about the ranking, in which case it doesn't matter either way)

Secondarily there could be a ranking where you simply get 1 point for every person you have beaten in all battles you have played (any kind, not just normal battles). That would be an easier to understand version of "experience" ranking.

Thundr- you have no fucking clue what relative ranking is

The way "main" ranking is now - playing more often is (slightly) more important than playing as good as you can, which it shouldn't be. Any ranking system that inflates points over time will have this serious fault.
39:37,91
Thundr-
37mins club
Posts: 398
Joined: 10 Jun 2010, 17:02

Re: Ultimate EOL Ranking discussion

Post by Thundr- »

My earliest post I explained my suggestion:
Kopaka wrote:Yeah that's pretty much how the relative ranking works.
Then I guess it isnt then? Whatever. I've like never won a battle and I never play them so im not the one to talk really. Im simply giving an explanation how it could be done as I know it works in many other competative games.
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Ultimate EOL Ranking discussion

Post by Kopaka »

For amount of wins, should I make it so a win only counts if there is certain amount of players in a battle? I imagine around 5 or something.
Post Reply