Power Rankings

[Foxglove]

Everyone sucks.

Well really, this philosophy can be encompassed by reframing the exercise. Let's call it the "everyone sucks" Impotence Rankings, to determine how weak and ineffectual clans are perceived to be in relation to each other. Clan A losing to Clan B, for example, would increase the inadequacy factor of Clan A.

....
....
....

Thanks, Chuuuuck, for organizing and leading this effort. Very nice!

[MTIceman41]

Nice work...this looks pretty solid

[niMic]

KoRT's low deviation is quite interesting. Am I remembering it wrong, or didn't we have a rather big deviation when this was done for the Conquerors Cup?

Great work, btw.

[Chuuuuck]

It was a little bit bigger for the cup, 2.5.

Part of that difference is from the decision to throw out the highest and lowest vote for every clan which I didn't do in the cup. It makes the standard deviation a little smaller and removes any extreme lonely outliers.

[Chuuuuck]

Also, for those of you that don't know. Standard Deviation is a number used in statistics to show the variance of the samples. A good rule of thumb is that 85% of the samples fall within the standard deviation. That means all votes except for 2-3 probably fell within the standard deviation for each clan.

For example, Empire's average is 6.38 with a standard deviation of 1.71. This means that approx. 85% of the votes for Empire fell between 4.67-8.09.

When a clan, such as THOTA, has a standard deviation of less than 1 then that means that most votes put them exactly where they are.

This is just an FYI to give everyone a better idea of how the votes came in.

[waseemalim]

Also, for those of you that don't know. Standard Deviation is a number used in statistics to show the variance of the samples. A good rule of thumb is that 85% of the samples fall within the standard deviation. That means all votes except for 2-3 probably fell within the standard deviation for each clan.

For example, Empire's average is 6.38 with a standard deviation of 1.71. This means that approx. 85% of the votes for Empire fell between 4.67-8.09.

When a clan, such as THOTA, has a standard deviation of less than 1 then that means that most votes put them exactly where they are.

This is just an FYI to give everyone a better idea of how the votes came in.

only if you assume normality of rankings.

[freakns]

And what of Nemesis and their performance vs. THOTA? Does it seem right that the 8th or 9th ranked clan should be giving #1 a run for their money?

---C

in all fairness, i dont see us better then no8-9 atm

I'd probably have Nemesis in the top 5 if you didn't have freakns in there!

^
what he said! its obvious we have players because of who we will drop in rankings! now, who the hell is that freakns? we should kick HER from the clan!

[Chuuuuck]

only if you assume normality of rankings.

Yes, that is true, but I didn't want to go into too much detail. We don't have enough votes to even get anywhere near a normality for the sample. BUT the standard deviation still represents all the votes somewhere in that proximity (85%). That is why I just left it at approximately 85%, for ease of understanding.

And I just checked and verified a few. For Empire, every vote but 2 falls within the standard deviation. KoRT also has every vote but 2 within theirs.

[Snowgun]

. A good rule of thumb is that 85% of the samples fall within the standard deviation.

FAIL

Plus or minus one standard deviation from the mean encompasses ~%68 percent of the set.

2 STD's plus minus is ~95, 3 is ~99.

Here is a correction:

For example, Empire's average is 6.38 with a standard deviation of 1.71. This means that approx. 68% of the votes for Empire fell between 4.67-8.09.

And we can safely assume a normal distribution here, since there are no real live factors that would cause this distribution to be skewed, like some sort of heavy weight (like income per population of a country)or a cut off value (like 0).

However, data on a set like THOTA might be skewed due to the cut-off. For instance, many people might vote them at 1, with the rest trailing down to 2, or 3. A histogram will show an exponential like curve instead of a bell curve. However we can just assume it's a bell curve that has been shoved at the top of the range and cut off, so no worries.

[Chuuuuck]

You are correct. I just looked it up. I guess my statistics is a little rusty since I haven't used it in a while.

[Snowgun]

You are correct. I just looked it up. I guess my statistics is a little rusty since I haven't used it in a while.

No worries, as long as you calc'ed the data right we are good to go.

[Drew483]

Interesting stuff...and a good read throughout the thread...thanks for doing this.

[Lubawski]

This looks great Chuck. Thanks for doing it.

KoRT's low deviation is quite interesting. Am I remembering it wrong, or didn't we have a rather big deviation when this was done for the Conquerors Cup?

You all opened up a can of whoop ass on G1 and us...and everyone saw it.

[Dako]

I guess we will need another vote after the Cup finals.

[Chuuuuck]

I guess we will need another vote after the Cup finals.

Ya. I will host votes every 4-6 months as needed. So I am guessing after the finals end is probably a good time for the next vote.

[Caymanmew]

can you add my clan? or do we have to finish a clan war?

[ljex]

can you add my clan? or do we have to finish a clan war?

you will be added for the next vote as your clan was formed after this vote. That would be somewhere in 4-6 months from the sound of it.

[Chuuuuck]

can you add my clan? or do we have to finish a clan war?

you will be added for the next vote as your clan was formed after this vote. That would be somewhere in 4-6 months from the sound of it.

That is correct. Each vote represents current active clans from the time of the last vote until the time of the current vote. Since this vote was the first vote it represents all active clans for approximately 3-4 months before the vote.

Since you all haven't gotten started yet, you will be added to the next vote which will be done 2-3 times a year.

[Chuuuuck]

The next vote I run will be sometime around December. I hope to have more participation each time around because more stats = better results.

[Jimmy V]

Interesting stuff.

As a newcomer to clans, I saw a lot improvement in my group (LOTZ). Much like the college football polls, it isn't about where you are when the season starts, it is whether you are moving up or down. I will be curious to see the changes on the next poll.

[Chariot of Fire]

I'm not sure you'll see much change Jimmy unless the ranking system adopts two distinct guidelines (and thus a need for two separate charts).

i) is a ranking based on historical data (where you see perennial favourites THOTA topping the standings - and justifiably so - followed in 2nd by TSM).

ii) is more of a 'seedings' ladder where current strength and form determine the standings (and on this you would more likely see LoW & TOFU higher up the placings, whereas TSM have had a quieter year).

So it will always be a conundrum if only one ranking ladder is used, due to the disparity between historical data and current 'form' data. If, for example, half of THOTA & TSM broke away to form a new clan would you put them at the bottom of a ladder as they are 'untested' (when on paper you know they would beat allcomers 9 times out of 10)? On a Top 10 clans list they wouldn't appear, but as seeds for a tourney (e.g. Conqueror's Cup) they would really have to be ranked #1 or #2.

Thus the current systems are far from perfect and until such time as every encounter, every challenge, every achievement (as a clan) receive points based on an approved system then the standings in any ladder will always be contentious.

[Incandenza]

If, for example, half of THOTA & TSM broke away to form a new clan would you put them at the bottom of a ladder as they are 'untested' (when on paper you know they would beat allcomers 9 times out of 10)? On a Top 10 clans list they wouldn't appear, but as seeds for a tourney (e.g. Conqueror's Cup) they would really have to be ranked #1 or #2.

That would indeed be a pretty badass clan...

You make some good points, especially because you've essentially lived the above scenario, tho in somewhat milder form. Part of the problem with too much historical weight in rankings is simple turnover. Every clan loses members, whether through the slow drip of retirement or the sharp shock of mitosis (for lack of a better way of putting it) or disbanding. Given the length of challenges and the work involved, a clan may be a substantially different animal than it was 4 challenges ago. But save something catastrophic happening to the clan itself (which really muddies the waters), a historical component to the rankings can still be helpful because it gives people a frame of reference. That frame might lead to prejudicial thinking/ranking, but that's just the cost of doing business.

[trestain]

wow things have really changed round here.

once i have found my cc legs again i hope to see some of you on the battlefield.

lots of updates to look at and learn.

any BW still around?????

[Jimmy V]

I'm not sure you'll see much change Jimmy unless the ranking system adopts two distinct guidelines (and thus a need for two separate charts).

i) is a ranking based on historical data (where you see perennial favourites THOTA topping the standings - and justifiably so - followed in 2nd by TSM).

ii) is more of a 'seedings' ladder where current strength and form determine the standings (and on this you would more likely see LoW & TOFU higher up the placings, whereas TSM have had a quieter year).

So it will always be a conundrum if only one ranking ladder is used, due to the disparity between historical data and current 'form' data. If, for example, half of THOTA & TSM broke away to form a new clan would you put them at the bottom of a ladder as they are 'untested' (when on paper you know they would beat allcomers 9 times out of 10)? On a Top 10 clans list they wouldn't appear, but as seeds for a tourney (e.g. Conqueror's Cup) they would really have to be ranked #1 or #2.

Thus the current systems are far from perfect and until such time as every encounter, every challenge, every achievement (as a clan) receive points based on an approved system then the standings in any ladder will always be contentious.

I guess I was looking at it incorrectly- I thought it was about the players' rank in the Clan, not the historical Clan rank. If a couple of people that have ranks as high as yours form a Clan, it should be top ranked, I think, despite no history.

Odd, but good information...

[Chariot of Fire]

Don't be fooled by everyone's rank Jimmy. A lot of the guys high up the chart are predominantly freestylers who rely on speed and quick wits to earn their points - a format that isn't played in challenges where sequential team games are the norm. And then there are the farmers.

A lot of the best team players are at major/brig level and tend to remain in that bracket as that's where the points start to stabilize, esp when points are shared between a team of three or four.