Algorithm of 400 Clan Ranking (Sep 30, 2011)

[Dako]

Any chance we can get a user script that is compiled from this updated list of clans that could show clan names (clan rank too? maybe) next to ppl's names? Right now, just using CC Clan Watch from Userscripts, and it seems to be outdated since there are a lot of new clans since its revision. Maybe there is a script already that I haven't found, so if one already exists, please offer a link for me, please.

Thanks for the work. Seems like the weighting is really the key to figure out something agreeable by all to represent a snapshot overview of where clans are at. I think the annual clan tournament could officially be the "weighted figure" that is used to determine who has bragging rights for a year. But this works as a record keeper. Unless someone set up a "season schedule" that forced everyone to play each other throughout the year, then we could use a simple season record, otherwise this is great.

Cheers

If you click "compile clans" it will refresh clan setting up to today. :)

[Master Kai]

Edit: So I went in to the .js script itself and added the missing names. But couldn't figure out why it wasn't populating with new names.

Thanks for the reply so quick.

I have hit the compile clans button hoping it would do just that. Thus far it does not update the newer clans. ie, it doesn't have PACK or Wild Geese, or MM.
By not updating, I mean, their clan name does not show up next to their names like the rest of them do. Those three are examples from my compiled list.
They also do not show up in the drop down menu when comparing clans.

Thanks in advance.

[Chariot of Fire]

....

[Qwert]

ACC- can someone explane to me how will these chalenges be count? They have some strange point calculation, who are not depend of number of games played in one chalenges.

[FarangDemon]

I've updated the rankings as of Aug 31. Hopefully soon I'll have implemented something that tracks change in ranking for each clan between a previous time period. But I didn't get around to it yet.

ACC- can someone explane to me how will these chalenges be count? They have some strange point calculation, who are not depend of number of games played in one chalenges.

If a war contains some singles, like ACC, I include those games weighted according to the formula used in the tournament.

So if the winner of 5 singles games gets one point, I just award one point to the team that wins the set.
But if winning a singles bout or a doubles game is worth 1 point, and trips/quads is worth 2 points, then I scale everything down so that the greatest weight for any single game is just 1 point.

So the singles bout is worth 1/2, a doubles game is worth 1/2 and trips/quads is worth 1 point.

The reason for the downscaling is that I don't want the war to have inflated weight due to its weighting system. I could do a better job of normalizing - because the way I have currently suggested doing it results in the war being slightly deflated in weight (because some games are worth a fraction of a point). Perhaps the best way would be, I downscale all weights by the same fraction, such that when you add up all the weights, it equals the total number of doubles, triples and quads games played in the war. That way the winner and win margin is determined according to that war's intrinsic weighting, and it's weight in my rankings is determined by how many actual games were played (doubs trips and quads or some function of singles, too).

I don't know what the raw data was, but the score jigger gave me was:

PIG OSA 10 9.5 08/31/11

So let me know if that is consistent with my initial proposition.

[FarangDemon]

Update - I can now track change in ranking between any two time periods. See first post. viewtopic.php?f=438&t=147062#p3211823

In the official rankings, Mythology moved up 3 spots to 16th.
Biggest fall was ID, slipping 5 spots down to 19th.

Among the provisionally ranked, the biggest gain this month was Pig Renters, moving up 9 spots to 16th.

[Dako]

You'd better use not minus and plus signs, but arrow up and arrow down. They are easier to understand. (html codes are ↑ and ↓)

[FarangDemon]

You'd better use not minus and plus signs, but arrow up and arrow down. They are easier to understand. (html codes are ↑ and ↓)

Thanks, I'll try that out. Yeah I was thinking that minus being good was not intuitive, but didn't think about up/down arrows.

[Qwert]

But why we can not have normal results for ACC- if they play 30-40 game chalenge ,why are we can not use normal result?These confuse me most. Well i will try to count each game one point, maybe its best for ranking.

[FarangDemon]

But why we can not have normal results for ACC- if they play 30-40 game chalenge ,why are we can not use normal result?These confuse me most. Well i will try to count each game one point, maybe its best for ranking.

Well I'd like the actual winner of the event to be the one who gets the points according to my ranking system. If I don't use the same relative weights that the organizer of the event has defined, then I won't get the same winner.

In that case, the winner of the actual event might not be awarded a win in my ranking system. I want to avoid this because I respect the judgement of the event organizer to choose an appropriate weighting system to determine winner.

[Teflon Kris]

But why we can not have normal results for ACC- if they play 30-40 game chalenge ,why are we can not use normal result?These confuse me most. Well i will try to count each game one point, maybe its best for ranking.

Well I'd like the actual winner of the event to be the one who gets the points according to my ranking system. If I don't use the same relative weights that the organizer of the event has defined, then I won't get the same winner.

In that case, the winner of the actual event might not be awarded a win in my ranking system. I want to avoid this because I respect the judgement of the event organizer to choose an appropriate weighting system to determine winner.

Thanks

PIG OSA 10 9.5 08/31/11

So let me know if that is consistent with my initial proposition.

Yes, that works for me - essentially halving the ACC points score would give something proportionate for comparison with scores in clan wars and elsewhere (whilst retaining the tournament's own eighting system).

Best way to incoporate scores into the algorithm in my view.

[Lubawski]

I can't wait to see how our score changes after the TOFU was is officially over. I think it's going to put us even with KORT (because their win over FOED is a little sizable as well).

[josko.ri]

I can't wait to see how our score changes after the TOFU was is officially over. I think it's going to put us even with KORT (because their win over FOED is a little sizable as well).

I think that it will put you a above us. your win is bigger (by only one) but more important, TOFU has more than 200 points more than FOED.

just look in graph how much KORT jumped when beat THOTA, and how much THOTA fell. and there was only 33-27, and point difference between those 2 clans were lower than now between Empire and TOFU. I would say, in September ranking Empire may reach score that has never been achieved in this type of ranking system. it is always great pleasure to kick asses of #1 clan

[shocked439]

Thank you to FD and Jigger for maintaining this. It is the most valuable ranking system for clans that we have and is fun to watch.

I'm not saying the others should stop I love power rankings in sports and polls in college football but I like the way this system works so thank you!

[Bruceswar]

I can't wait to see how our score changes after the TOFU was is officially over. I think it's going to put us even with KORT (because their win over FOED is a little sizable as well).

I think that it will put you a above us. your win is bigger (by only one) but more important, TOFU has more than 200 points more than FOED.

just look in graph how much KORT jumped when beat THOTA, and how much THOTA fell. and there was only 33-27, and point difference between those 2 clans were lower than now between Empire and TOFU. I would say, in September ranking Empire may reach score that has never been achieved in this type of ranking system. it is always great pleasure to kick asses of #1 clan

Yes if it does, that just means more points for us when we beat them OK light banter aside.... Hell of a ranking system.

[FarangDemon]

You'd better use not minus and plus signs, but arrow up and arrow down. They are easier to understand. (html codes are ↑ and ↓)

Thanks, I'll try that out. Yeah I was thinking that minus being good was not intuitive, but didn't think about up/down arrows.

Implemented the up/down arrows. Also updated graph of performance over the last 12 months.

[Dako]

Awesome!

[Lubawski]

I can't wait to see how our score changes after the TOFU was is officially over. I think it's going to put us even with KORT (because their win over FOED is a little sizable as well).

I think that it will put you a above us. your win is bigger (by only one) but more important, TOFU has more than 200 points more than FOED.

just look in graph how much KORT jumped when beat THOTA, and how much THOTA fell. and there was only 33-27, and point difference between those 2 clans were lower than now between Empire and TOFU. I would say, in September ranking Empire may reach score that has never been achieved in this type of ranking system. it is always great pleasure to kick asses of #1 clan

Yes if it does, that just means more points for us when we beat them OK light banter aside.... Hell of a ranking system.

I love when Bruce starts to trash talk. It's like a nun walking into a strip club. There's a little bit of shame involved, but you can tell his panties just got a little damp.

I whole-heartedly agree with a couple of the above posts. I think this is the most accurate ranking system we've had so far. It gives good balance to both past and present results, clan strength at the time of defeat, etc. I'm no where near smart enough to understand how it all works, but great job Farang.

[Chariot of Fire]

What kind of scary ass nuns live in your neighborhood? 'his' panties?

[Lubawski]

What kind of scary ass nuns live in your neighborhood? 'his' panties?

There are pictures in the EMPIRE forum. You'll see them soon enough.

[josko.ri]

I am looking at your graphs and one thing isnt clear to me, weight factor. I thought weight factor is function of time passed from war and total number of games played in the war.

What I dont understand, KORT has the same weight (6%) for war vs G1 and for war vs LoW. but both of these factors are in huge favor of war vs G1, because there were 41 games vs G1 (19 vs LoW) and G1 war is from June while LoW war is from March.

so, did i interper factors for counting weight wrong, or there is some mistake in calculating?

[The Voice]

I just noticed that Memento Mori has only 828 points in the first chart, but in the chart segmented into three lists, the clan has 920 points. Are the numbers being calculated differently?

[FarangDemon]

I am looking at your graphs and one thing isnt clear to me, weight factor. I thought weight factor is function of time passed from war and total number of games played in the war.

What I dont understand, KORT has the same weight (6%) for war vs G1 and for war vs LoW. but both of these factors are in huge favor of war vs G1, because there were 41 games vs G1 (19 vs LoW) and G1 war is from June while LoW war is from March.

so, did i interper factors for counting weight wrong, or there is some mistake in calculating?

Good question. Your assumption is correct, weight is a product of games in war times a decay factor.

I looked into the data. There were three recent wars between Kort and G1. April 2010 (41 games), Dec 2010 (18 games, probably CL3) and June 2011 (41 games). I'm guessing that the June 2011 had not been entered into the datafile yet at the time I generated that output (it is dated to July 1, but since the war is dated June 20, it is possible that it didn't make it yet).

Since this data was computed July 1 (I have this written right above the section), then for the wars in April 2010 and Dec 2010 we have weight calculated as:

KoRT-G1
July 2011 - April 2010 = 15 months old --> (24-15)/24 * 41 games ~= 15.4 weight points
July 2011 - Dec 2010 = 8 months old --> (24-7)/24 * 18 games ~= 12.8 weight points

KoRT-LOW
July 2011 - March 2011 = 3 months old --> (24-4)/24 * 19 games ~= 15.8 weight points

So these values seem to correspond with the relative weights (expressed as percentages) of 6%, 5% for the G1 games and 6% for the LOW game, respectively.

I do plan to update these old images to show the ratings basis for current ratings when I next update ratings for Sept 30. Sorry for the confusion caused by my non-updating of this info.

I just noticed that Memento Mori has only 828 points in the first chart, but in the chart segmented into three lists, the clan has 920 points. Are the numbers being calculated differently?

No, but the chart segmented into three lists that contains details about each war is not up-to-date. It is current as of July 1. I will update it next time.

[Georgerx7di]

so, just curious. Who do we have to beat to move up on this thing?

[malevolous]

THOTA would be a good start....