Active Clan ELO ranking 2011-[update 31 dec 2011]

[Eyestone]

The quality of a clan isn't necessarily reflected in the players score, you should know this. There are several generals on the site that I wouldn't let into AOC just like that because they don't know how to play team games (freestyle farmers or plain standard escalating experts for instance). And there are several examples of great players who has a low rank because they play different type of games rather than focus on having the highest rank they possibly can. You will no doubt find some correlations between newcomers having a low ranks compared to old timers, but I doubt that you will find that the "clan scores" correlate perfectly with the power ranks. It just doesn't work that way.

[ga7]

Just a suggestion... It would be a lot of work but using seq team games maprank with maybe a factor on their provided opponent maprank (equalitarian, brawler, etc) would provide a fairly accurate picture of the actual clan strength.

[Rodion]

The quality of a clan isn't necessarily reflected in the players score, you should know this. There are several generals on the site that I wouldn't let into AOC just like that because they don't know how to play team games (freestyle farmers or plain standard escalating experts for instance). And there are several examples of great players who has a low rank because they play different type of games rather than focus on having the highest rank they possibly can. You will no doubt find some correlations between newcomers having a low ranks compared to old timers, but I doubt that you will find that the "clan scores" correlate perfectly with the power ranks. It just doesn't work that way.

Eyestone is right.

Another thing to consider is that, even if points were a perfect representation of a player's skill, some players end up playing more games than others and a simple arithmetic mean wouldn't do the trick. To ilustrate that, imagine a clan whose members are only field marshals (all at 4500) and cooks (all at 500) in equal proportion (50/50). Considering this clan has power "2500" would not be accurate if the field marshals played the lion's share of the games and left only the bare minimum to the cooks. In fact, even a weighted mean wouldn't provide an accurate value, because who's to say the cooks actually think with their own heads as opposed to following every field marshal's advice?

Not that anything I wrote in the previous paragraph matters, because, as I said, Eyestone is right. I'm just trying to show how it would be impossible to measure the "depth" of a clan.

[Bruceswar]

so by going off your rankings... clans would want to farm for win % seems lame?

Really? And here I thought clans continually tried to win to increase their wins. Surely your clan tries to increase its wins? Well, by default, if you win more, your win% goes up.

Seriously, Bruce, I get what you're saying but... I doubt decent clans fight "lesser clans" just for easy pickings, and if the clan isn't decent, it's not likely to continue a winning streak to make this a concern, because better clans will overwhelm them. Eventually it will be noticeable if a clan is refusing battles with "reputable clans" in favor of "easy pickings" to increase win %.

Contrarily, just because a known clan agrees to a battle with an unknown clan, doesn't mean they're doing it because they think it will be an easy win. For example, as I recall, BSS, a fairly decent clan, will often give unknowns a chance at them... I doubt they're doing it simply because they think they'll swarm the opponents to increase their win%.

You obviously were not around when clans farmed for ladder spots on JP's ladder. I forgot which version it was, but DLPP, and THOTA(Possible - Their war with DBC 35-5?) Now DLPP stated that is what they were doing. THOTA and LoW (vs PE) got raw shakes, and now both those results are basically voided out. DLPP was not a slack clan either. It was a pretty solid clan when they were at full force.


To expand on Eyestone a bit, sure while rank does not mean much, it can help in a situation like this. In general clans with higher overall ranks are better than those who are composed of sgt's and the like. Not to say all are like that but ....

[Qwert]

by Eyestone » Sat Mar 05, 2011 4:03 am

The quality of a clan isn't necessarily reflected in the players score, you should know this. There are several generals on the site that I wouldn't let into AOC just like that because they don't know how to play team games (freestyle farmers or plain standard escalating experts for instance). And there are several examples of great players who has a low rank because they play different type of games rather than focus on having the highest rank they possibly can. You will no doubt find some correlations between newcomers having a low ranks compared to old timers, but I doubt that you will find that the "clan scores" correlate perfectly with the power ranks. It just doesn't work that way.

I know that,but these mean alot.
Look example of Tofu vs T4c-using rank score you can almost predict who will going to win. Try these on CC2,and you will see effect.
Also effect in Thota and Kort its also very close. Slightly THota are in advantage, but bouth clan are close to be equal quality.
I think that these need more time,and more diferent games to show much more efect on scoreboard, and i thing that give best picture of clan ranks.

aka vs osa 17-25 [41]
legion vs nemesis 30-40 [69]
empire vs ID 23-18 [34]
foed vs dbc 28-12 [25]
time vs t4c 27-13 [40]
time vs bofm 28-32 [46]
bofm vs bss 39-21 [59]
tofy vs T4C 40-10 [23]
dbc vs RA 26-34 [45]
----------------------------
look these results,what i get so far from 2011- ,and all clans who are average score rating higher then hes opponents win in 100% cases. If i realised these earlier, with these formula,i will take prize for "predict Clan League winner"
So far these ranking are quite succesful, because favorites have much high % of success,then low ranked clans.

by ga7 » Sat Mar 05, 2011 5:18 am

Just a suggestion... It would be a lot of work but using seq team games maprank with maybe a factor on their provided opponent maprank (equalitarian, brawler, etc) would provide a fairly accurate picture of the actual clan strength.
Just a suggestion... It would be a lot of work but using seq team games maprank with maybe a factor on their provided opponent maprank (equalitarian, brawler, etc) would provide a fairly accurate picture of the actual clan strength.

Almost imposibile for me, because its have over 30 clans,and who know how many plauyers are involve in clan wars.

Another thing to consider is that, even if points were a perfect representation of a player's skill, some players end up playing more games than others and a simple arithmetic mean wouldn't do the trick. To ilustrate that, imagine a clan whose members are only field marshals (all at 4500) and cooks (all at 500) in equal proportion (50/50). Considering this clan has power "2500" would not be accurate if the field marshals played the lion's share of the games and left only the bare minimum to the cooks. In fact, even a weighted mean wouldn't provide an accurate value, because who's to say the cooks actually think with their own heads as opposed to following every field marshal's advice?

well theory its one ,and reality its another thing. I think,and i belive that points are perfect representation of a player skill, but in Team games you need to take average points. Just try to use these parameter in CC2, and you will be supprised how high will be yours prediction %..
Clan who have 2700 or high average point score,have very big chance to defeate clan who have 1600 average point score.

[Chariot of Fire]

bookmark

[Qwert]

update with all results for cl3 who are finish in 2011.

[Commander62890]

Are these rankings going to be used to sort teams for any future Leagues or Cups?


If so, I might have to tell my buddies to start farming some noobs so we can get the #1 seed.

[jrh_cardinal]

Are these rankings going to be used to sort teams for any future Leagues or Cups?

not unless qwert runs a future league or cup

[Frederik Hendrik]

I personally like the idea of a ranking based on results. But jpcloet has tried it many times, and didn't really succeed in producing a working formula, so I think it will be very hard.

I would implement in your method the score of the clanwar. So if a war ends 30-10 the winner gets 30 x (loser points/winner points) - 10 (winner points/loser points). That would avoid farming, because you could lose points while winning the war when there is a large difference in points.

FH

[Qwert]

by Frederik Hendrik » Sun Mar 06, 2011 12:59 am

I personally like the idea of a ranking based on results. But jpcloet has tried it many times, and didn't really succeed in producing a working formula, so I think it will be very hard.

I would implement in your method the score of the clanwar. So if a war ends 30-10 the winner gets 30 x (loser points/winner points) - 10 (winner points/loser points). That would avoid farming, because you could lose points while winning the war when there is a large difference in points.

FH

Well im all ready give examples on page 2,and these formula based on average score- prevent farming,because if you are stronger clan, you get less points,no mather with what score you win. Farming are imposibile here.
(look page 2 example TOFU-T4C)

[jpcloet]

This is essentially ELO with a modifier. It promotes winning frequently and higher % and disregard quality of opponent. I think I've seen this before. Take it as an alternative measure of clans. The eventual winner will be a clan that wins a lot in the mid-range clan zone. You should call this Qert's Active Clan ELO ratings.

[Chariot of Fire]

cl3:LOW vs TOFU 4-4 [0]

This can't be right. The game total must add up to 9.

[Qwert]

by jpcloet » Sun Mar 06, 2011 3:23 am

This is essentially ELO with a modifier. It promotes winning frequently and higher % and disregard quality of opponent. I think I've seen this before. Take it as an alternative measure of clans. The eventual winner will be a clan that wins a lot in the mid-range clan zone. You should call this Qert's Active Clan ELO ratings.

Qwert Active Clan ELO rating 2011

ok if you think its a proper name for topic,i can change name .

by Chariot of Fire » Sun Mar 06, 2011 4:43 am

cl3:LOW vs TOFU 4-4 [0]

This can't be right. The game total must add up to 9.

Not mine fault, they not update table.
I check all games-LOW-TOFU 5-4
I will add these to score table.

[Qwert]

man,amazing i just add ongoing battles. there have 35 ongoing events.
do i miss something

[jpcloet]

man,amazing i just add ongoing battles. there have 35 ongoing events.
do i miss something

Not sure, active challenges has not made it to page 2, however, we might get there soon.

[Qwert]

30 clans so far have finished games in 2011

[Qwert]

2 chalenges added to ranking.

[Arama86n]

bofm vs bss 39-21

They were ranked #15, we were ranked #19, according to the power-ranking that was available when that war started
(writing this for the record as qwert threw this result out there in the middle of a discussion on farming.)

Oh and in general about this thread, I'm hardly going to object (too sternly) to any list that has BoFM at the top
Jokes aside though, having a ranking based on *results* from the current year sounds very interesting, and I applaud your intent. Just going by score does pose problems though. As it was, BoFM beat a higher ranked clan, thus making it a worthy victory, but what if we'd beat a new clan 39-21, if we're still at the same place in the rankings there is obviously a problem, the system would lack needed depth. As it is I didn't grasp exactly how you reached your results.
Whatever the case, it's only March. It will be very interesting to see this list, in comparison to the power-rankings, come December. then we can talk

Good luck

[Qwert]

thanks,its still early, and its look that these year will bring amazing number of chalenges.
Well OSA are new clan,and they up,because have quite quality clan. Its all related on results, if your opponent are less quality(based on average point score),then you will get less number of points, and opposite,if you win against high ranked clan,you will get big number of points.

[FarangDemon]

I applaud your efforts to use a formula to rank clan performance.

I like the concept of Elo and that you've started each clan with 1000 points. But I think one way to improve what you've done would be to ditch the average clan scores formula (CC player score is a can of worms best avoided I think) and instead have each clan's score calculated as a function of their enemies' current performance ratings at time of war (the value that starts at 1000 points for each clan).

http://en.wikipedia.org/wiki/Elo_rating_system

Performance Rating is a hypothetical rating that would result from the games of a single event only. Some chess organizations use the "algorithm of 400" to calculate performance rating. According to this algorithm, performance rating for an event is calculated by taking (1) the rating of each player beaten and adding 400, (2) the rating of each player lost to and subtracting 400, (3) the rating of each player drawn, and (4) summing these figures and dividing by the number of games played. This can be expressed by the following formula:

Performance rating = [(Total of opponents' ratings + 400 * (Wins - Losses)) / Games].

So if Clan A (1000) beats Clan B (1000), Clan A's rating becomes (1000 + 400) / 1 game = 1400, Clan B becomes (1000 - 400) / 1 game = 600
Say, Clan C and D play, C wins, Clan C is now 1400, and Clan D is 600.

Then if the winners play, and Clan A (1400) beats Clan C (1400), Clan A's rating becomes (1000 + 400) + (1400 + 400) / 2 games = 1600, Clan C would have (1000 + 400) + (1400 - 400) / 2 games = 1200

Then, if the losers from first two games play, and Clan B (600) beats Clan D (600), Clan B's rating becomes (1000 - 400) + (600 + 400) / 2 = 800
Clan D becomes (1000 - 400) + (600 - 400) / 2 = 400

Under this system, the figures used in the calculations are the opponent's current rating, it doesn't get recalibrated as the opponent's rating changes over time.

One cannot farm for points. In fact, if one clan has more than 400 points than another, the higher-ranked clan will not benefit from playing the lower-ranked, even if they win.

I think using this system, over time, a clan can only get to the top by beating the winningest clans. It's a simple formula, I'm sure it's been discussed before, so please forgive my ignorance if it has been rejected for good reasons. I'd be interested to know the grounds for rejection, though.

[FarangDemon]

Just had another idea. The formula proposed above could be modified to weight each win/loss according to how many games were played in the set.

So given A vs B (A wins 40 game set), C vs D (C wins 40 game set), A vs C (A wins 20 game set).

A's new score is [ (1000+400)*40 + (1400+400) ] *20 / (40 + 20) = [56000 + 36000] / 60 = 92000 / 60 = 1533

So given weighting by number of games in the challenge, challenges comprised of relatively more games have a greater effect on the performance rating.

Btw, given this scheme, it only matters whether one clan wins or loses a challenge - margin of win does not affect score. This could be considered a drawback, but it does simplify things.

[Chariot of Fire]

Thanks for putting some thought into a solution FD. I do think though that the result of a challenge (i.e. the final score) should have some bearing on points won/lost. If such a system was in place it would then make challenges between elite clans and lesser clans still be competitive, e.g. the elite clan could earn points if they were to win by a margin of say 20 games. Without such a system in place I just don't see the incentive a leading clan would have to play one of the minnows as there wouldn't be any reward whatsoever.

I believe a handicapping system is the best solution - one that's also very easy to calculate. I proposed something a long time ago, since buried in the threads somewhere, whereby each clan is given a handicap which factors into each result and becomes elastic. The ladder is then based on best handicap (at #1) on downwards.

I'll try and find the time to recap what I'd suggested once before. In principle it takes your ELO system in addition to allowing an easy calculation to see what score a clan must achieve to better its handicap and move up the ladder.

[Qwert]

well like im all ready give example for TOFU-T4c chalenge- Tofu are big favorite,and in these ,you can see that in example of 40 games(they play 50 games)
-----------------------------------
tofy win points against t4c:
1622/2707*40=23 ranking points (50 game chalenge TOFU get 30 ranking points)
T4C win against TOFY:
2707/1622*40=67 ranking points (50 game chalenge T4c get 83 ranking points)
---------------------------------------------------------------------------------------------
For TOFU its Better to play with much better clans like THOTA,IA,KORT, or some clans who have 2000 pt average,because will get more points. More average pt clan,more ranking points.
From other way, for T4C are best if play also with high ranked clans,because they will get quite large number of points, unfortunatly in all these games,low ranked clan, so far(and these is fact) can not win against stronger clans.

[White Moose]

This ladder doesn't really work at all... No disrespect qwert, you clearly put some effort into this one, unfortunatly that doesn't make it work :/

('Make it work'... that reminded me of that dude in Project Runway, I forget his name and I'm too lazy to google it)