The North Carolina Melee Power Rankings! Updated 8/14/14!

[~Twitch~]

sorry brow but dotpwn was my first. you were just my second.

 

[Dorsey]

OK, sorry for the insults. Allow me to explain how simple it is:

the data is only bad because you guys are using data that can even potentially be "bad"(weighting wins/losses in regard to the previous wack PR periods over the course of time[6 months], and going by who plays who as if the seeding means nothing whatsoever). Why don't you just do the rankings according to the data? Then you get ranked in how you did in the tournament, no johns, and seed / do the pools right instead of have a gay *** panel in the first place, unless if two happened to be tied or very close in a pt system like cam suggested, then you can use a little discretion so you guys feel important.

 

[Dr Peepee]

OK, sorry for the insults. Allow me to explain how simple it is:

the data is only bad because you guys are using data that can even potentially be "bad"(weighting wins/losses in regard to the previous wack PR periods over the course of time[6 months], and going by who plays who as if the seeding means nothing whatsoever). Why don't you just do the rankings according to the data? Then you get ranked in how you did in the tournament, no johns, and seed / do the pools right instead of have a gay *** panel in the first place, unless if two happened to be tied or very close in a pt system like cam suggested, then you can use a little discretion so you guys feel important.

We didn't weight anything about the data related to past PR periods.....

If you think it would be better in a different way then the results are in the links in the OP at the top so you can look at the data yourself and tell us where we made mistakes. There was not much we could have done with how things went this time but if you aren't convinced then you can look for yourself like I said.

This isn't even about feeling important lol. Panels have always been around for this stuff in every state for a really long time and if you feel the need to disagree with that system then I suggest a better argument against it than "they want to feel important." I'm not saying that is your argument but you should give a better reason for why a panel isn't necessary is all.

 

[Dorsey]

I didn't mean all periods. It was plural because I was referring to you doing it every time with the past ranking perioD. My argument had nothing to do with you guys feeling important. I thought it was funny, but you took offense to it and based half your post around it. Nice. lol

 

[Dr Peepee]

I didn't mean all periods. It was plural because I was referring to you doing it every time with the past ranking perioD. My argument had nothing to do with you guys feeling important. I thought it was funny, but you took offense to it and based half your post around it. Nice. lol

Can you tell me where exactly the panel weighted things unfairly in their decisions?

 

[Bl@ckChris]

well, i'm not sure if his no weight policy makes it so that it doesn't matter who you beat, as long as you place high (ie, beating scrubs to get 7th > beating pr'd people and getting 9th) or if it matters who you beat but it doesn't matter how they were ranked before.

can you clarify which one of those two it would be dorsey?

 

[Dark Hart]

Dorsey's model seems to reward... consistency?

Interesting

The current model does too kinda though...

 

[OmniOstrich]

this thread reminds me of HoN atm. A nice mix of try-hards, rage, bias, misinformed players, and some just downright trolls that are somehow supposed to mesh as a team.

 

[DJRome]

Something missing in this thread.....................









Oh yeah...

loooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooool

****ing rofl rofl rofl











anyways, i already thought long enough about this to make it pretty simple.

there is no mathematical way of determining what prs are. at least not in any "objective" way that math is supposed to. because data is done one time, bayesian concepts are needed. however, the data is so sparse that it is impossible in more ways than one.

you cannot do it on the basis of first, that any distribution you come up with to predict would have a ****ty confidence interval. secondly, the method of prediction is completely subjective and has no way of confirming variables' weightings to any degree of reason.

thus, you are left with 2 options.

either: do what the panel does now or use a very clear, simple point-based method as cam as suggested. the latter is clearly not necessarily "accurate" for anything BUT it has the advantage of being simple to calculate, easily understood, and consistent in how it is calculated.

 

[Sneak8288]

Shoulda just used a point system a bajillion months ago when I said it, and as far as previously unranked but good players we can probably assign them a seperate point value. Players who were unranked due to inactivity could have like a point value of 5 as a starting point compared to the players unranked due to bad performance. We can always try it and if it doesn't work out we can try something different

 

[Lightsyde]

If everyone who is in this thread discussing these PR's came to a tournament, we'd have an awesome tournament.

I ****ing love this state. Carry on, Gents.

 

[Dark Hart]

Lol sneak looking for credit

 

[Sneak8288]

Cam already gave me credit, im just saying its something that should've been done a long time ago

 

[Lightsyde]

My biggest problem with a point system is that it becomes a bigger test of your ability to attend tournaments than a test of your skill.

Similarly, with a point system, it will be much harder for players further away to possibly maintain the kind of numbers that players close to most tournaments have.

Maybe if we started using points, there'd need to be some kind of other variable included to offset your accumulated points by the number of tournaments you attended.

For example, if two players ended up with 20 points at the end of a ranking period, with one player having gone to 2 tournaments and the other having gone to 3, the player who went to less should get the higher spot (more points for less tourneys means overall higher performance).

 

[[FBC] ESAM]

Losses should also take away points, such as a ranked loss making you -1 and an unranked loss causing -2?

 

[Sneak8288]

We could always use the who beat who system for things the point system doesn't determine as well couldn't we? But you do make a good point josh and the one with less tournaments would most likely have better overall performance. We can always use this period as a means to test it out

 

[lord karn]

My biggest problem with a point system is that it becomes a bigger test of your ability to attend tournaments than a test of your skill.

Similarly, with a point system, it will be much harder for players further away to possibly maintain the kind of numbers that players close to most tournaments have.

Maybe if we started using points, there'd need to be some kind of other variable included to offset your accumulated points by the number of tournaments you attended.

For example, if two players ended up with 20 points at the end of a ranking period, with one player having gone to 2 tournaments and the other having gone to 3, the player who went to less should get the higher spot (more points for less tourneys means overall higher performance).

I agree with this. It also makes it difficult to assign value to non-ranked players. As I pointed out earlier, a non-ranked lozr is much different than a non-ranked random. Having a point system creates value purely by the old PRs, which is something we tried to avoid with the current system. Analyzing set counts makes it so the old PRs don't matter, aside from the way they influence seeding as I said in a earlier post (but that only affects the data we get, not the way we analyze it).

 

[Dorsey]

Why is it so difficult? Is it that hard to assign a pt. value to lozr and some nobody? It would give the panel something to do, by discretion, if a point system is used... lol...

and karn, the old PR's do matter... I hear you guys stating all the time v. a ranked player or v. a non-ranked player... you guys weight wins/losses by the previous PR's, and over time[6 months], a changing variable, is where you get skewed data.

With a pt system, although logically the old PRs would be a good starting point for a pt system, it would become self-sufficient and you wouldn't need to use any previous PR list after that D:

As for attendance, it gets factored a good bit if you weight the tournaments with a point value as well. Tournaments that people would be less likely to come to would have a lesser value, and the value is of course determined by the sum of the participants pt. values. Otherwise i'd say, ya gotta go to a tournament to get ranked. Yeah, a much larger tournament would be worth 3x more pts than a regular one........but........anyone who is serious about getting ranked would go to that. and if not then, **** happens.

IDK, I just don't see how you can be content with saying this ranking period has "bad data" and not a pt system because(example) it's difficult to assign a pt value to a non-ranked lozr and a nobody.

Very simple D:

Sorry again for how I came across earlier, the whole situation is just pretty amusing especially when I'm watered.

 

[Dark Hart]

Then apologies sneak

 

[Bl@ckChris]

...so are people gonna go to the gold mine then?

or am i really gonna have to go to georgia the week after for some decent competition?

 

[Dr Peepee]

Shoulda just used a point system a bajillion months ago when I said it, and as far as previously unranked but good players we can probably assign them a seperate point value. Players who were unranked due to inactivity could have like a point value of 5 as a starting point compared to the players unranked due to bad performance. We can always try it and if it doesn't work out we can try something different

The panel should speak as one unit. If you disagree with what we're doing then you should take it up with us not the public. That's why you're on the panel to get influence like that in the decision process. We just gotta be consistent with how we present our arguments I feel.

meh half asleep

If everyone who is in this thread discussing these PR's came to a tournament, we'd have an awesome tournament.

I ****ing love this state. Carry on, Gents.

^^^^jhyuvfhsafiaji^^^

Why is it so difficult? Is it that hard to assign a pt. value to lozr and some nobody? It would give the panel something to do, by discretion, if a point system is used... lol...

and karn, the old PR's do matter... I hear you guys stating all the time v. a ranked player or v. a non-ranked player... you guys weight wins/losses by the previous PR's, and over time[6 months], a changing variable, is where you get skewed data.

With a pt system, although logically the old PRs would be a good starting point for a pt system, it would become self-sufficient and you wouldn't need to use any previous PR list after that D:

As for attendance, it gets factored a good bit if you weight the tournaments with a point value as well. Tournaments that people would be less likely to come to would have a lesser value, and the value is of course determined by the sum of the participants pt. values. Otherwise i'd say, ya gotta go to a tournament to get ranked. Yeah, a much larger tournament would be worth 3x more pts than a regular one........but........anyone who is serious about getting ranked would go to that. and if not then, **** happens.

IDK, I just don't see how you can be content with saying this ranking period has "bad data" and not a pt system because(example) it's difficult to assign a pt value to a non-ranked lozr and a nobody.

Very simple D:

Sorry again for how I came across earlier, the whole situation is just pretty amusing especially when I'm watered.

It's not HARD to assign a point value, but it could very easily be done badly or just plain suck.

We rank according to who we know will be ranked(those who show skill we can obviously see....part of the reason humans are needed) and then as things get lower on the list we just see whose records are looking the best vs those who are ranked above them as well as vs anyone who isn't being considered for being ranked(like how we considered GofG's record). In other words, yeah we actually can use all the data in one period to rank for every period.

Why is a point system better than what we do now though?

There's an attendance requirement too for our method. All tournaments are weighted equally under our system as everything in a tournament environment should be taken seriously. The problem with attendance this period was there were about 4 legitimate tournaments and some people only attended 1. They got punished for only attending 1 because the ranking period was so big and there were so many tournaments in it. This is usually not the case though, as we usually use 3 tournaments in a period so 1 tournament is enough to make the attendance requirement(and this will be the norm once again). Basically, I don't think attendance will be an issue with the regular system anymore. How hard is it to make 1 out of 3 tournaments at the least, especially if you're trying to be ranked?

I am content calling the data "bad" because a lot of better players didn't enter much and data towards the bottom of the list conflicted strongly but also closely and in such a way that kept me from being able to identify who most deserved the spot. Usually such abnormalities don't happen, or if they don't they aren't this numerous, so I can feel comfortable calling this particular data set "bad" by comparison if nothing else. It shouldn't be "bad" again as long as everything stays on schedule though.

...so are people gonna go to the gold mine then?

or am i really gonna have to go to georgia the week after for some decent competition?

EVERYONE GO TO GOLD MINE GOGOGOGOGOGOGOGOOGGOGOGOGO

 

[0Room]

Boone is coming to Gold Mine I'm 99% sure
At least me and one other person

One of the better players here is staying because his girlfriend is visiting

As far as PRs go, the only things I have to say are this:
It's impossible for old PRs NOT to influence you. It's a built in bias of the past influencing the future. That's just how life works. You always use past experience to influence future decisions.

Now whether or not you want to continue with this, or move on to the points [or more specifically Elo system that was talked about a little while ago] is up to you, I'm fine either way.

Active/non active member debate does get a little clearer on a points system though imo.

But that's all I have to say about that. I do have a question about who's going to Apex/how/how many people but I'm going to move that to the Nc social boards

 

[lord karn]

Why is it so difficult? Is it that hard to assign a pt. value to lozr and some nobody? It would give the panel something to do, by discretion, if a point system is used... lol...

and karn, the old PR's do matter... I hear you guys stating all the time v. a ranked player or v. a non-ranked player... you guys weight wins/losses by the previous PR's, and over time[6 months], a changing variable, is where you get skewed data.

With a pt system, although logically the old PRs would be a good starting point for a pt system, it would become self-sufficient and you wouldn't need to use any previous PR list after that D:

As for attendance, it gets factored a good bit if you weight the tournaments with a point value as well. Tournaments that people would be less likely to come to would have a lesser value, and the value is of course determined by the sum of the participants pt. values. Otherwise i'd say, ya gotta go to a tournament to get ranked. Yeah, a much larger tournament would be worth 3x more pts than a regular one........but........anyone who is serious about getting ranked would go to that. and if not then, **** happens.

IDK, I just don't see how you can be content with saying this ranking period has "bad data" and not a pt system because(example) it's difficult to assign a pt value to a non-ranked lozr and a nobody.

Very simple D:

Sorry again for how I came across earlier, the whole situation is just pretty amusing especially when I'm watered.

Actually, we don't weight previous PRs at all in our discussion. It may seem like that, but that is only because the same players tend to do well every period, which makes sense. The players that end up in contention for the powerrankings are the ones who consistently beat the other players early on in bracket. Once those players have been weeded out, we compare records among the people left. Those players that are left are almost always the same, give or take one or two people that get better or fall off.

And the difficulty of assigning point values to non-ranked players is that it requires our discretion, which is something that we try to avoid. We shouldn't have to decide how many points lozr is worth, his performance should decide that for him, which is how our current system works.

If we did a continually progressive system, it would still not be considering ranking outright. It would seem that way, but only because every data point added into the system is just continuous with existing data, which would preserve the players previous records in a sense. It would be something that is continuously shifting, and older data points would probably slowly degrade in importance or something like that.

And DJ is right, the system has a really low confidence interval because we have so little data (statistically speaking). It's pretty much always going to be that way, but it's better than having a point system (unless we somehow turn what I've been talking about into a point system), because a point system is ultimately arbitrary.

Sorry if this post is confusing, I just took a ton of medicine because of a migraine. @__@

 

[Moophobia]

I've said in the thread I'm going to tgm but Nahhhhhhh *****es don't notice. Bout to get slept on and about to regret it.
Real rage

 

[*P*L*U*R*]

Lol sneak looking for credit

Sneak actually suggested points a long time ago.




Also, points should be wiped after every ranking period, ie the previous ranking period should have no effect on the current one. It'll reward consistency so that that constantly hazy 10-7 area isn't so hazy anymore. And isn't the ideal ranking period for NC like every 3-5 tourneys? You should only be eligible for ranking if you attend most of the tourneys that period.

 

[Dr Peepee]

Depends on what you mean by ideal. Ideal for having more data with more tourneys, sure. For making a PR that has to deal with nasty attendance issues and more things we have to subjectively weight? Not so much. That's why I've always been behind 3 tournament ranking periods because we can require 1 tournament to be ranked but we update fairly often so you can't get away with not coming too long regardless. I think there are other benefits to a 3 tourney period as well but that's not related to your post so lol.

How much is "most?" 3 out of 5? 2 out of 4? Why is that much a good idea?

 

[bossa nova ♪]

cool, positivity about smash and NC


all it took was the panel to make the worst top 10 in history from the ****tiest period that i've ever experienced in my 2 years in this community



hope to see everyone at TGM3!!


EDIT:

Spoiler

bayb <33

 

[*P*L*U*R*]

Until someone tells me what TGM is, I am going to continue thinking of random assortments of words in my head to craft the meaning of those three letters.

Transcontinental Grimlin March

 

[*P*L*U*R*]

thread possibly derailed, btw.

 

[Dr Peepee]

The Gold Mine 3!

brb thread link grabbing

edit: grabbed: http://www.smashboards.com/showthread.php?t=309162&page=34


Double: If 20+ Melee players come then we get M2K. COME ON GUYS LET'S SEE M2K YEAAAAAHHHH

 

[Dark Hart]

**** m2k

Hater

 

[Lightsyde]

Alright, let's do this.

I'm going to the Tiny Giant Marathon 3, and I need a car from Raleigh with me. Get that time off work or do whatever you gotta do.

I heard some Smash tourney is happening in Charlotte that weekend too. >_>

 

[Sneak8288]

Im gonna try to come but there's a mvc tourny the same day that I wanted to go to

 

[Lightsyde]

Im gonna try to come but there's a mvc tourny the same day that I wanted to go to

Lmao how you complain about PR'd players not attending tournaments and then opt to go to an MVC3 tournament instead.

Not that I don't understand having other hobbies, but in context, I feel like the pot just called the kettle ethnic things.

 

[Sneak8288]

Im retired, plus I went to almost every tournament last season. I did my part compared to most. Plus I missed the last two mvc3 tournies to go to smash tournies where the turnout ended up pretty bad so this time im leaning more away from smash since I already chose smash over it twice

 

[stingers]

i wanna go josh =]

 

[Dorsey]

My posts/point are very simple, and no one has even said a word to refute what I'm saying. It's like a debate with Obama rofl. All I get are these paragraphs telling me in essence how "arbitrary" a point system is, just because YOU TWO see no reasoning to change it up, because you think you're doing a stand-up job. If the seeding is done right in both the WB and the LB, especially after pools, then everyone is on an even playing field going into it(the seeding should no doubt come from a collaboration of tournament results between the tournament organizers). It's just the most right and fair thing to do, which doesn't make it arbitrary. As I said, any serious competitive league (that's not run by random hs/college age TO's) does it this way. The right and fair thing is to just do it by tournament results, strictly, so there is nothing questionable/debatable like there ALWAYS is when you guys do it........bottom line: there is nothing questionable/debatable if you put the effort into doing it by tournament placing, because it's what happened... you just link the results. There is no such thing as "bad data" because it's THE data. YOU JUST USE IT. And since you guys can't refute this apparently, I guess it's that our opinions differ on what the power rankings should be. I think it should be based off tournament results(placing) and who attended, which is why I am supportive of a point system. That should be the structure of the system and just because the two people who are running the PR think it's arbitrary in regard to their current way of doing things, really doesn't mean ****.

 

[Dr Peepee]

My posts/point are very simple, and no one has even said a word to refute what I'm saying. It's like a debate with Obama rofl. All I get are these paragraphs telling me in essence how "arbitrary" a point system is, just because YOU TWO see no reasoning to change it up, because you think you're doing a stand-up job. If the seeding is done right in both the WB and the LB, especially after pools, then everyone is on an even playing field going into it(the seeding should no doubt come from a collaboration of tournament results between the tournament organizers). It's just the most right and fair thing to do, which doesn't make it arbitrary. As I said, any serious competitive league (that's not run by random hs/college age TO's) does it this way. The right and fair thing is to just do it by tournament results, strictly, so there is nothing questionable/debatable like there ALWAYS is when you guys do it........bottom line: there is nothing questionable/debatable if you put the effort into doing it by tournament placing, because it's what happened... you just link the results. There is no such thing as "bad data" because it's THE data. YOU JUST USE IT. And since you guys can't refute this apparently, I guess it's that our opinions differ on what the power rankings should be. I think it should be based off tournament results(placing) and who attended, which is why I am supportive of a point system. That should be the structure of the system and just because the two people who are running the PR think it's arbitrary in regard to their current way of doing things, really doesn't mean ****.

Losers bracket can't be seeded for, but winners is seeded as accurately as possible based on PRs and a consensus among multiple people. No one ever accused the seeding of being arbitrary though.

Set counts and placings pretty much produce the same lists, so that isn't the big dividing issue you make it out to be. The effort between the two is actually probably easier if the panel just considering placings, actually. I'd have to think it out to be sure. There's plenty of effort and discussion to be done either way.

I link the results because I want you to argue how the data we had could be interpreted better using the actual data.

And yes, data by itself is not inherently "bad," but for purposes of effective PR placement the data is somewhat troubling and upsetting. Like I have said, any method this period would have resulted in more than unsatisfactory PRs.

Both ways are arbitrary to an extent, and there are 3 people on the panel. Sneak also resides on the panel.

 

[Dorsey]

Wasn't referring to manually making the seeds for the LB, was referring to a well-seeded WB automatically seeding the LB well, assuming all the number assignments in the LB correspond to the right losers in each round. Concerning your point about this being not a 'big dividing issue' and in regard to the panel's goals apparently paralleling what I'm suggesting, then I don't see why you wouldn't be for such a system to at least be the basis of it. Would make life a lot easier for the panel, and I left sneak out of it just because you two were the only panelists actively responding to my posts, not to mention he seemed at least somewhat supportive of a pt system.

 

[Dr Peepee]

Wasn't referring to manually making the seeds for the LB, was referring to a well-seeded WB automatically seeding the LB well, assuming all the number assignments in the LB correspond to the right losers in each round. Concerning your point about this being not a 'big dividing issue' and in regard to the panel's goals apparently paralleling what I'm suggesting, then I don't see why you wouldn't be for such a system to at least be the basis of it. Would make life a lot easier for the panel, and I left sneak out of it just because you two were the only panelists actively responding to my posts, not to mention he seemed at least somewhat supportive of a pt system.

effin 502.....

WB is seeded as well as objectively possible since we just go straight from the PR anyway, and use multiple opinions when there is need for subjectivity(like OOS with no PR).

Well set counts and placings are pretty similar, but set counts have been deemed as better for PR purposes due to their elimination of things such as fluke brackets. Unless you were talking about this and a point system, in which case the two systems are pretty different, and, as DJ said(http://www.smashboards.com/showpost.php?p=13309235&postcount=12810), the mathematical system cannot measure up.