[breadfan]

Originally Posted by RuineD

i dont think u can really limit bias or opinion with something like this.

i mean this is a stretch but the problem with that is the lans might be quite small and meaningless whereas a big tournament win is worth a lot more. for example, revival has won like 50+ playhem dailies. but compare that to say one MLG win or IEM win, which would u place ahead? i know thats not SEA but similar problems could arise.

Not at first. Over time, you can, but you need to start somewhere. The power ranking becomes more accurate as time goes on and rankings can be adjusted based on results alone

[Delraich]

Originally Posted by Cyanide

suggestion:

make a list that TRIES to LIMIT bias as much as possible

e.g. Assign EQUAL values to things such as

Tourny winnings
tournies joined
tourny placements
popular vote
ladder ranking in say (NA or KR)
Giant Slayer points (points given for defeating/taking games off international t1/2 players in tournaments)
etc.

the more data we can pull from various things, we're limiting problems like whether
Person A winning 20 $20 tournaments in SEA
should be placed ahead of
Person B who won one LAN that paid out $500
or the other way round

I like this idea. I wonder if something like this would work (random numbers but I am just trying to get the idea across):
Each small/ non specific tournament that resulted in $50 or less = 1 point, $51-100 = 2 points, etc.
Specific tournament(s) (e.g. a high profile one) = 5 points.
Player vote average score: 1st place = 5 points, 2nd place = 4 points, 3rd place = 3 points, etc.
Slaying a high end player (e.g. Violet/ Huk) in a tournament = 3 points

Each player then submits their annual performance with a total score. Highest number wins.

[NvRossi]

i've been around a lot longer then you Pet ^^

[ROOTPetraeus]

i wasnt complaining....

Quick Comments

NvRossi:

twas just a casual response!

[breadfan]

Originally Posted by Delraich

I like this idea. I wonder if something like this would work (random numbers but I am just trying to get the idea across):
Each small/ non specific tournament that resulted in $50 or less = 1 point, $51-100 = 2 points, etc.
Specific tournament(s) (e.g. a high profile one) = 5 points.
Player vote average score: 1st place = 5 points, 2nd place = 4 points, 3rd place = 3 points, etc.
Slaying a high end player (e.g. Violet/ Huk) in a tournament = 3 points

Each player then submits their annual performance with a total score. Highest number wins.

You just assign a point value to the 10 power rankings. Tournaments are worth points equal to the number of power ranked players attending.

So Glade is worth 10 points, Pig worth 9, Mafia worth 8, tgun worth 7, and so on - just as an example, pick your points system as you please. If everyone attends, that tournament becomes worth more points (so an ACL national event is worth more than an ACL regional, etc). Just default any international tournaments to be worth maximum power ranking points, no need to mess around with bonus points for killing a "high end player", because not all "high end players" are an equal challenge. Just base it purely on placing.

[UHF]

I didn't really agree with the list provided, as I feel like it didn't capture the spirit of the whole year. I spent a while last night & this morning going over the years results, and i tried to find as many events as I could that had results listed. Here's what i did:

1. entered 40+ events listed in my spreadsheet.
2. SEA events only - if you think you can work out the international stuff, be my guest.
3. I gave each event (or event category) a points system for up to the top 10 placings.
4. I adjusted some events depending on certain circumstances. I also combined some quals/finals to make things both fairer and easier to work out.

ATTEMPT 1

The system i used was a fairly archaic way to rank things (basic descending numbers), and mightn't be the proper way, however the ranking seemed to turn out fairly accurate.

RANK	PLAYER	POINTS	# RESULTS
#1	mOOnGLaDe	231	18
#2	Mafia	189	20
#3	PiG	170	17
#4	tgun	142.5	15
#5	Light	75.5	6
#6	Rossi	70.5	15
#7	TargA	70	6
#8	JazBas	68.5	7
#9	iaguz	65	10
#10	Ninja	55	11
#11	YoonYJ	35	4
#12	SeNSei	30.5	5
#13	TechTron	30	5
#14	deth	25.5	5
#15	yang	23.5	4
#16	Pokerface	23	5
#17	Pinder	22.5	4
#18	Myuu	19.5	6
#19	KnighT	18	2
#19	Petraeus	18	3
#20	EdgE	17	2

ATTEMPT 2

I did a bit more digging and tried using a system similar to tennis, much bigger numbers and much larger gaps as move down the positions. I also changed the weight of some of the tournaments.

After the totals were tallied I divided by 10 to make the numbers more manageable (yes glade was on 19260).

RANK	PLAYER	POINTS	# RESULTS
#1	mOOnGLaDe	1926	18
#2	Mafia	1049	20
#3	PiG	1034	17
#4	tgun	623	15
#5	TargA	505	6
#6	JazBas	287	7
#7	Rossi	277	15
#8	Light	269	6
#9	iaguz	160	10
#10	Ninja	160	11
#11	deth	142	5
#12	EdgE	118	2
#13	TechTron	118	5
#14	SeNSei	108	5
#15	YoonYJ	87	4
#16	Petraeus	75	3
#17	Myuu	71	6
#18	KnighT	66	2
#18	Fourby	60	2
#20	Pokerface	56	5

OTHER STUFF

Either way the top 4 stayed the same, using to completely different scoring systems - and I'm happy to lock that in. I prefer the second attempt as it's based directly off an already used tournament system. Beyond that there is some variance, mainly due to how I grouped certain event types together. It still needs work, and the points system needs to have more variables that change each individual events weight (as no two events are the same). You would need to spend a LONG time really working out how to rank everything, because there are way too many variables to quickly put together lists.

A couple of names I took off the list as they had a single result all year round, which I felt defeats the spirit of making a list like this. The rest of the numbers are untampered. I included both TargA and Kingkong, as they were/are living & playing in our competitive scene.

I think this is as close to a proper 2012 list as you'll get, other than substituting your own event weights and adding in any events I missed. All i can say is if you want something for 2013, you should probably start working something out now.

Quick Comments

	breadfan:	I think you need to rank all tournaments equally to begin with, but otherwise nice
	NvRossi:	impressive!
	nirvAnA:	best post in this thread so far

[Cyanide]

i think delraich kinda got what i was getting at.

note that i said LIMIT instead of ELIMINATE bias. The good thing about assigning equal value to all statistics is that you can get someone who wins one big tourny vs someone who wins 10 small ones on "equal" footing. obviously, this isn't ideal in that sense.

Ultimately, for every new statistic we pull from, we give everything else less weightage. Also, as time goes on, you'll see that the ranking will eventually even out for everyone who still plays competitively.

[Delraich]

Originally Posted by breadfan

You just assign a point value to the 10 power rankings. Tournaments are worth points equal to the number of power ranked players attending.

So Glade is worth 10 points, Pig worth 9, Mafia worth 8, tgun worth 7, and so on. If everyone attends, that tournament becomes worth more points (so an ACL national event is worth more than an ACL regional, etc). Just default any international tournaments to be worth maximum power ranking points, no need to mess around with bonus points for killing a "high end player", because not all "high end players" are an equal challenge. Just base it purely on placing.

Just thinking aloud, but how do you know Glade is rank 1 and worth 10 points, Pig is rank 2 and worth 9 points in the first place? It seems like you need a system to rank and then once you know that ranking you use this data to re rank? Maybe a variant on this idea could be to give the top 10 players a flat value of 10 points each? If each ranked player is worth points, how much is number 11 or 12 worth? I do like this concept/ approach, but I am struggling to see how it would assign the ranking scores correctly if the ranking of the top 10 is not clear in the beginning.

[breadfan]

Originally Posted by Delraich

Just thinking aloud, but how do you know Glade is rank 1 and worth 10 points, Pig is rank 2 and worth 9 points in the first place? It seems like you need a system to rank and then once you know that ranking you use this data to re rank? Maybe a variant on this idea could be to give the top 10 players a flat value of 10 points each? If each ranked player is worth points, how much is number 11 or 12 worth? I do like this concept/ approach, but I am struggling to see how it would assign the ranking scores correctly if the ranking of the top 10 is not clear in the beginning.

Yes, you base it on the current power rankings (or something else). I'm not talking about creating a power ranking here, I'm saying that to begin with, you can't be totally fair and accurate from the outset. The goal is to end up with a way to rank different tournaments against each other.

You have to accept that to begin with, there's a little bit of arbitrary assignment. Whether you do it by votes from players, equally rating all tournaments, or whatever, it doesn't matter. What matters is that you get a list. Once you have a list, it gains credibility in 2013 as tournaments can actually be compare, and players can move up and down with their results.

[ToRSpartaz]

Cute man that coding is fantastic I love it!

[iM MaFia]

finally a power rank that actually had some effort and thought process to it

Quick Comments

	ROOTPetraeus:	cry more plz
	SLCN.NXZ:	^
	FaDeBadger:	Someone has his panties in a twist

[TAEdgE]

hahaha 2 results reppin'

[nGenLight]

UHF is like Hollinger to basketball for SCII.

Based on his results... The biggest rivalries in SEA SCII by results are:

Mafia/Pig
Light/Rossi
Iaguz/Ninja

[nGenLight]

Also... where the **** is the Singaporians? AARON COME BACK!!!!

[Cyanide]

another thing to note. this is an ANZ power ranking list, understandably due to things like scene size etc.
perhaps there is a way to be more inclusive to other countries in the region. unfortunately they do not have something like ACL

Quick Comments

SLCN.Kez:

Show me all the online tournaments sg players won and ill glady put them above any of the others.

[x5.RunaMoK]

power rankings suck.
end of discussion

[Eddie]

Strong list. Order is always going to depend upon individual opinion when there's no set system in place to judge. Good job i thought.

Originally Posted by Cyanide

another thing to note. this is an ANZ power ranking list, understandably due to things like scene size etc.
perhaps there is a way to be more inclusive to other countries in the region. unfortunately they do not have something like ACL

Well for online events, 2013 will have SEA power rankings. Major SEA online tournament ranking system. Masters Cup + others coming together.

[UHF]

I was going to try and include every country in the region, but the amount of time I'd need to sink into doing that would be ridiculous (it could be done though!). The ANZ group was the simplest to put together.

This table is about as unbiased as I could be. I simply got the results, weighed the events, and output the numbers. The only way it's skewed, is if my event weights are off (which I don't think they are, or not much by much) - or if I've missed events (which is possible, but there is only so much time I can put into trawling for events).

The one thing I have seen though is a big shift in the mid-tier, simply from the movement of players (retiring, going inactive, new players coming in). Also the lower tier, of where breakouts look like they might occur, but then fizzle out. It's fairly interesting to look at, and I might look at expanding it for 2013.

[breadfan]

Originally Posted by UHF

I was going to try and include every country in the region, but the amount of time I'd need to sink into doing that would be ridiculous (it could be done though!). The ANZ group was the simplest to put together.

This table is about as unbiased as I could be. I simply got the results, weighed the events, and output the numbers. The only way it's skewed, is if my event weights are off (which I don't think they are, or not much by much) - or if I've missed events (which is possible, but there is only so much time I can put into trawling for events).

The one thing I have seen though is a big shift in the mid-tier, simply from the movement of players (retiring, going inactive, new players coming in). Also the lower tier, of where breakouts look like they might occur, but then fizzle out. It's fairly interesting to look at, and I might look at expanding it for 2013.

How did you weight the events? I don't think it's possible to do initially, there has to be some arbitrary assignment surely?

[UHF]

Originally Posted by breadfan

How did you weight the events? I don't think it's possible to do initially, there has to be some arbitrary assignment surely?

Yes the initial weight for the 1st event was, but there really is no other way to do it. I picked 3 events (ACL regional, ACL regional final, ACL online) and got standardized weights for each of them. That gave me something to work on for other events bigger or smaller. After this, I had a look at prize pools, participation numbers, the actual player pool in some cases, invite vs open bracket, etc to find an appropriate weight class. Weighing events against each other is fairly tricky, as each has equal opportunity - which I also think is important.

FYI - the events i used were basically between WCS Oceania and Masters Cup level.

Quick Comments

breadfan:

Ah I see, makes sense