Meta Knight Officially Banned!

[Flayl]

Are you unable to figure out that the current data for winnings is done with bad math?

What's the bad math?

 

[Battousai780]

What's the bad math?

His data presents MK as overcentralizing the metagame because he wins most of the money. But with this current system of keeping the data, is it really MK that wins this money? The system inflates the winnings of characters that are used even slightly more. Unless a player actually uses MK 50% of the time in a tourney and their other main the other 50% of the time, how is it fair or truthful to state that MK wins half the money?

 

[Flayl]

His data presents MK as overcentralizing the metagame because he wins most of the money. But with this current system of keeping the data, is it really MK that wins this money? The system inflates the winnings of characters that are used even slightly more.

You realize that actually has no bearing on my question right

If you secondary MK, you count as a MK player (for the 11%)

so again

where's the bad math?

 

[stingers]

Are you unable to figure out that the current data for winnings is done with bad math?

The math is accurate. You've gotta realize that there's no possible way we could check each set to see who used which character and make decisions off that. And even if we could, how would we use it?

If we're only interested in money, only the characters that you used in losers finals, winners finals or grand finals (with normal top 3 pay out systems) would count, as they're the only matches that have any effect on how much money you win. If we counted every match, in winners and losers, what if someone used their Ness in all early round matches, but when it came to finals, they only went MK. How would you interpret that? I wouldn't be surprised to find out that the MK usage in later rounds of the tournament bracket is way higher than in earlier rounds. Which would only INCREASE the % of money won if we had the capability to determine character usage in individual sets. So...lol

 

[Battousai780]

You realize that actually has no bearing on my question right

If you secondary MK, you count as a MK player (for the 11%)

so again

where's the bad math?

^^^ Highly inaccurate. Having MK as a secondary or being a MK player says nothing about the actual truth of how many times he's used in a tourney and how much money he wins. The only way to figure these things out is to acquire accurate data. Without accurate data, don't try to validate your arguments with bias. I'm merely calling for more accurate data after exposing why the system is biased, and calling for an end of using this biased data to support your position.

 

[stingers]

The data's not biased. The data's simply data. Data cannot be biased. Your interpretation of it certainly can be, but assuming John#s isn't tampering with the data in any way, we can assume it's just actually what's going on, that we can determine to the best of our ability.

 

[Flayl]

^^^ Highly inaccurate. Having MK as a secondary or being a MK player says nothing about the actual truth of how many times he's used in a tourney and how much money he wins. The only way to figure these things out is to acquire accurate data. Without accurate data, don't try to validate your arguments with bias. I'm merely calling for more accurate data after exposing why the system is biased, and calling for an end of using this biased data to support your position.

Dude you don't get it

I'm talking about MK PLAYERS. The over 50% of money is won by MK PLAYERS. That includes mains, secondaries and even tertiaries.

Those 11% MikeHaze was talking about? MK PLAYERS.

Please someone tell me they know how to do basic math before going on about accurate data

 

[Battousai780]

The math is accurate. You've gotta realize that there's no possible way we could check each set to see who used which character and make decisions off that. And even if we could, how would we use it?

If we're only interested in money, only the characters that you used in losers finals, winners finals or grand finals (with normal top 3 pay out systems) would count, as they're the only matches that have any effect on how much money you win. If we counted every match, in winners and losers, what if someone used their Ness in all early round matches, but when it came to finals, they only went MK. How would you interpret that? I wouldn't be surprised to find out that the MK usage in later rounds of the tournament bracket is way higher than in earlier rounds. Which would only INCREASE the % of money won if we had the capability to determine character usage in individual sets. So...lol

The argument that actual data is harder to acquire is not an argument proving why the current data is true or should be used. If we know it's wrong, it's wrong. We don't just use it because we having nothing better. In the issue about the Ness: it doesn't matter how we interpret it. I don't care. All I'm saying is that we should at least know the truth. I don't care what you interpret from that data, I just want that accurate data first.

 

[Battousai780]

The data's not biased. The data's simply data. Data cannot be biased.

It's how he got the data that's biased. derp

Dude you don't get it

I'm talking about MK PLAYERS. The over 50% of money is won by MK PLAYERS. That includes mains, secondaries and even tertiaries.

Those 11% MikeHaze was talking about? MK PLAYERS.

Please someone tell me they know how to do basic math before going on about accurate data

If I use Marth 90% of the time in a tourney I think Marth played a greater role in me winning a tourney than whomever I picked for the rest of the 10% of matches.

 

[Ussi]

its not like he excludes tourneys from being in the data (unless they are like not following a proper ruleset)

 

[stingers]

If I use Marth 90% of the time in a tourney I think Marth played a greater role in me winning a tourney than whomever I picked for the rest of the 10% of matches.

Thats the thing lol. That's why it matters how you interpret the data. Because someone might say your marth had more effect, however if that 10% was losers, winners, and grands and you used MK the whole time...that says a completely different story.

 

[Flayl]

It's how he got the data that's biased. derp



If I use Marth 90% of the time in a tourney I think Marth played a greater role in me winning a tourney than whomever I picked for the rest of the 10% of matches.

holy **** you simply do not know basic math

look i'll even use your own example:

let's say you won $100 with Marth and Ganondorf

John doesn't know the ratio of usage, so he assumes 50%, 50%

He puts $50 for Marth and $50 for Ganondorf in the full split category
He puts $100 for Marth and Ganondorf in the no split category

Now let's do it with your ratio of usage
He puts $90 for Marth and $10 for Ganondorf in the full split category
He puts $100 for Marth and Ganondorf in the no split category

Do you get it now? The no split category doesn't care how much Ganondorf you used, because it's how much money you as a PLAYER won

 

[Ripple]

I honestly don't know where mike got 11%. Anyone care to explain?

 

[Flayl]

http://www.smashboards.com/showpost.php?p=13576034&postcount=558

 

[Dr. Tuen]

The data's not biased. The data's simply data. Data cannot be biased. Your interpretation of it certainly can be, but assuming John#s isn't tampering with the data in any way, we can assume it's just actually what's going on, that we can determine to the best of our ability.

Ah, I'd argue against this point. Data can be biased when an uneven portion of the data is missing or replicated. We have both issues: when money is multiplied across multiple characters to 'represent' their winnings, we have the replication issue. When tournaments that may display MK's dominance (or lack thereof) are missing in bulk, we have the missing data issue.

===

Then theres no way to accurately measure the data. No Johns If the best method I have for making measurements is still innacurate, that doesnt suddenly make it accurate.

This is true. Though I'd argue that the inaccuracies here were as a result of various transformations performed on the data (money multiplication for secondary representation). I think the raw data may still be OK, the analytical method needs to be changed. For this to be portrayed accurately, the analytical method must not depend on information we do not have (i.e. the money split problem can't be solved without percentage-of-use data for every player involved).

 

[stingers]

Fair enough, I'm certain you know far more about statistics than me lol.

Based on the information we have though, we've got it all down.

 

[da K.I.D.]

I was going to quote like 15 posts and split them into paragraphs and answer them all individually. But then I saw this.

http://www.smashboards.com/showthread.php?t=313154

Yes, we have Concentrate, of course, but you all missed the fact that 3 other MK banned tournies happened this weekend:

http://www.smashboards.com/showthread.php?t=313159
http://www.smashboards.com/showthread.php?t=313063
http://www.smashboards.com/showthread.php?t=313039

Behold the results.

10 Characters in the top 8, 2 repeats. (both of the repeats were atomsk.)
9 characters in the top 8, 1 repeat.
10 Characters in the top 8 ZERO repeats.
EIGHT characters in the TOP 3. THIRTEEN characters in the top 8. 3 Repeats.

If that kind of diversity doesnt tell you that banning MK was the right move, then you wont ever be convinced. Because thats all the proof anybody should need

 

[John12346]

I honestly don't know where mike got 11%. Anyone care to explain?

http://www.smashboards.com/showpost.php?p=13576034&postcount=558

Those numbers are inaccurate, I had to recheck them.

http://www.smashboards.com/showpost.php?p=13549487&postcount=512

 

[Flayl]

K.I.D. it hasn't even been a month yet -_______-

 

[Battousai780]

ITT: explaining easy to understand points 5 times and still being misunderstood

 

[Grim Tuesday]

I was unaware that opinions of the newb majority should dictate how people who actually play competitively should be affected. If "broken" is not criteria to ban a character and it boils down to the majority's preference, this game is a joke. If everyone decided to ban Ness because he was too small and hard to hit, making him not fun to play against, we shouldn't just cave in and do it. I'm sorry, but being salty because better players beat you with MK is not criteria for banning him. I can't just rally people to ban Diddy because I think his bananas are gay or DDD cause chaingrabs make me mad.

This is the peak of irony coming from someone who supports a LGL.

The point is that he didn't get the 11% from you because the way you present your data is biased. Just like he said, it is not fair to evenly distribute tournament winnings among a players characters. What I gathered from Mike is that even if used in 1 set, the money is split evenly among all characters used in a tourney period. How is this not biased? Why don't you actually look at the % usage of the character and dibby up the money based on that percentage? If a character like MK is used the most, as he is because he is the best, (which happens in all games), the ammount of money he seems to win is incorrect and extremely biased.

I hear that this is the only way to measure the data? What bull**** excuse is that? Of course this is not the only method to analyze the data. It's so highly inaccurate which can easily be determined by the criteria of how the numbers are found.

Me providing a better method to calculate winnings is not necessary to determine that the current method has down syndrome. What I'm merely saying is that if we know that the current data is horribly skewed, no one should be able to use it in order to provide validity to their argument.

Also, the idea that a new set of data would show MK as being more broken is bias theory. How do you know what the data holds before you take it?

-edit-

Answer me this question: If your method is alright because in a set a person needed MK to win making the distribution of money even, how do you know that he couldn't have won otherwise with a different character? The only non bias is to show winnings based on the percentage of character usage in all matches. <--- Not hard when most TO's keep a bracket and can record this data. However, if you haven't recorded this data, you can't use it to say why MK is broken.

Confound Battousai, he drives me to drink.

 

[John12346]

I'm just gonna leave this here because this thread isn't moving at 9999mph, plus people are complaining about my charts here too anyway: http://www.smashboards.com/showpost.php?p=13578522&postcount=695

Alright, as I said, I need to rock this world! Time to defend my charts!

(For those of you wondering - yes, I edited this post out of my previous post because it got lost to the next page.)

Okay, so the argument is that we don't know how much money each character has truly won, because it's possible that some characters were used less than other characters, correct? With that in mind, I went through my character breakdowns, and performed a process I officially dub "Reduction."

Here's how it works:
Step 1: Find a player.
Step 2: Determine which character he or she has won the most money with.
Step 3: EXCLUDE all other characters who have not won as much money for a player as X%{0%, 25%, 50%, 75%, 100%} of the player's top character(in the case of ties, just leave the characters alone).
Step 4: Repeat Steps 1 - 3 for all money winning players.
Step 5: Add up the breakdowns for each character for a new modified "Cash Won(No Split)" for every character.
Step 6: Add up those values from Step 5 to create a modified amount of money to divide by. Note that I need to divide by modified versions of "No Split" because I can't calculate how the sum for "Full Split" will decrease with this function.

The drawback for Part 6 is that the percentages displayed are not the actual success percentages, but that's not important in the context of this project, because we want to observe how much the values change over various Reduction sessions, rather than compare how high the values actually are, so don't worry about that.

Alright, so what the hell did I exactly do? In order to counter Mike's statement that my data cannot reliably show how successful a character is, I went ahead and made a graph that shows what happens when you begin to tear away less successful secondaries from players. We can find out exactly which characters are being carried, and which are not. Of course, we cannot determine such a thing off one data point, so I went ahead and used increments of 25% so we can observe how the changes occur.

Putting it simply, the left side of the graph indicates a general idea of success when all players' characters are considered, and the right side of the graph indicates a general idea of success only when all players' top characters ONLY are considered. And of course, the space between them shows how success for each character changes as you remove players' secondaries(100% Reduction is a flawed approach, mind you, as it does occasionally remove some players' legitimate characters).

- A line with a positive slope indicates that a character is LESS reliant on secondaries in order to win money, and the value set for the character's "Average" category on my charts is likely an underestimation.
- A line with a negative slope indicates that a character is MORE reliant on secondaries in order to win money and the value set for the character's "Average" category on my charts is likely an overestimation.

Oh looky there, MK has a positive slope.

 

[Dakpo]

ಠ_ಠ

ಠ‿ಠ

ಥ‿ಥ

 

[Grim Tuesday]

John#s is ****.

 

[Cygnet]

(I know some people have raised this argument before, but I'm not sure what people said about it, so....)

John's chart is based on totally unbiased data (as data is by nature), but its interpretation suggests that MK is the one winning money and not the people playing MK. It suggests that without MK, these people would not be making this kind of money because the chart simply attributes all of that money to "MK" even though we all know that player skill is a strong factor also, and admittedly, that would be hard to include quantifiably a chart.

(The prime example is the Socal tournament where Tyrant, Tearbear, and other dominating players who mained MK continued to dominate without MK. Although this probably also happens at other MK-banned tourneys, just like the tourney Mikehaze referenced in his video.)

So, does the chart have any significance? The data is real, but it's missing a VERY important confounding variable: individual player skill. There's obviously a strong correlation between MK and money, but people are already assuming causation from that. And for all those Stats/Science people out there, you should know that correlation in no way means causation.

 

[Eon the Wolf]

A good way to test that Cygnet? Let the ban run its course. If player skill is as crucial as you say it is, then the data in the coming months will show other characters SKY ROCKET in earnings, and if you're right...those 'pro MK players' will become 'pro diddy' or 'pro marth' or 'pro pikachu' or pro this or that. IF player skill is responsible for the large majority of money earned, then without MK, we should see similar climbs in money earning of a significant amount for one or more characters in the coming months. If its not a drastic increase in any one or two characters in terms of money earned...then it proves that MK himself was responsible for the drastic boost of his money earned compared to money earned by other characters.

 

[Battousai780]

ITT people don't realize that you get conclusions from data and not data from conclusions.

 

[Eon the Wolf]

...That's what I'm suggesting.
The data we have so far SUGGESTS that MK is vastly superior, and we've all agreed that while player skill is no doubt an important factor, it is one that's hard to quantify.
Anti ban is proposing that it is more due to player skill than MK alone.
Ergo, I propose a compromise;
For the first couple months after the ban's official start, we wait and see.....
If there's a significant rise in one or more characters and their money gain %s so as to illustrate a significant increase in gains in comparison to other characters, then the player skill argument now has some data to back up their conclusion, as that means that new characters have risen to be as 'superior' as MK is "now". As a result, the 'MK is vastly superior' argument will then be in doubt. Doubly so if, after bringing MK back, MK doesn't regain/keep as significant a money gain % as he has now, which would presumably be due to the evolved metagames of other characters as a result of the forced ban/character change

IF, however, the money gain % changes are more averaged out after the ban (IE no one or two characters or w/e are SIGNIFICANTLY above the rest), or rather not as drastic, then that is evidence to point towards MK himself being a significant factor in the large % of money being earned by MK...

Bam. An easy, long term test to showcase how much of an effect player skill will have on character profit returns. If it's based on player skill as much as anti ban likes to say, then surely within 3-6 months of the official start date for the ban, there will be significant meta game improvements and/or drastic increases in profit %s of total money earned among all characters so as to show whether or not it's the character, or the player skill, most affecting profit returns.

 

[Budget Player Cadet_]

who picks MK when they don't need him to win lol

I just thought I'd point out Ramin's MK: he's beaten Leon fairly well for a while now, but now he uses MK against him not because it's necessary, but because he's sick of it, IIRC.

 

[Elyssa Xey Hexen]

Whenever I see Budget Player Cadet_ I always expect a free cup full of rage to be served.

I was disappointed this time around.

 

[C.J.]

I just thought I'd point out Ramin's MK: he's beaten Leon fairly well for a while now, but now he uses MK against him not because it's necessary, but because he's sick of it, IIRC.

From my conversations w/ Ramin it seems like he picks MK vs Leon because it's just easier.

 

[Budget Player Cadet_]

From my conversations w/ Ramin it seems like he picks MK vs Leon because it's just easier.

That is also possible; he did really wreck Leon's **** at Archer OSTS with MK.

@Xeylode: sometimes I am calm and collected. Other times I just **** all over my keyboard.

 

[Mr-R]

I just wanted to try out MK and when I kept winning why bother changing ?

 

[Spelt]

And soon enough Leon's gonna switch to MK too!

 

[Orion*]

Or don't, and start your own tournament scene with MK legal.

Most tournaments still have MK legal tho

Did you even read what he posted?

LMAO
funniest **** ever

Cut it out, all of you.

I think he's serious though that's the best part

and fox has bad matchups iirc

No...
Even ones. Yeah. But he's good on all stages, with no bad MU's.

The data's not biased.

While I don't think they changed anything in the data, and I trust them both as a accurate source of information. I do think the Presentation of the data is biased

http://www.smashboards.com/showpost.php?p=13576034&postcount=558

Wow... LOOOOOL
/done

Ah, I'd argue against this point. Data can be biased when an uneven portion of the data is missing or replicated. We have both issues: when money is multiplied across multiple characters to 'represent' their winnings, we have the replication issue. When tournaments that may display MK's dominance (or lack thereof) are missing in bulk, we have the missing data issue.

===



This is true. Though I'd argue that the inaccuracies here were as a result of various transformations performed on the data (money multiplication for secondary representation). I think the raw data may still be OK, the analytical method needs to be changed. For this to be portrayed accurately, the analytical method must not depend on information we do not have (i.e. the money split problem can't be solved without percentage-of-use data for every player involved).

Yeah but who are we going to find to run through everything like that lol

I was going to quote like 15 posts and split them into paragraphs and answer them all individually. But then I saw this.




10 Characters in the top 8, 2 repeats. (both of the repeats were atomsk.)
9 characters in the top 8, 1 repeat.
10 Characters in the top 8 ZERO repeats.
EIGHT characters in the TOP 3. THIRTEEN characters in the top 8. 3 Repeats.

If that kind of diversity doesnt tell you that banning MK was the right move, then you wont ever be convinced. Because thats all the proof anybody should need

Like. There's not even a metagame // tierlist for that **** yet. When brawl came out everything was ALLLL over the place. Don't get me wrong I do think there will be more diversity with MK gone. But just wait on it, things will devolve.

(I know some people have raised this argument before, but I'm not sure what people said about it, so....)

John's chart is based on totally unbiased data (as data is by nature), but its interpretation suggests that MK is the one winning money and not the people playing MK. It suggests that without MK, these people would not be making this kind of money because the chart simply attributes all of that money to "MK" even though we all know that player skill is a strong factor also, and admittedly, that would be hard to include quantifiably a chart.

(The prime example is the Socal tournament where Tyrant, Tearbear, and other dominating players who mained MK continued to dominate without MK. Although this probably also happens at other MK-banned tourneys, just like the tourney Mikehaze referenced in his video.)

So, does the chart have any significance? The data is real, but it's missing a VERY important confounding variable: individual player skill. There's obviously a strong correlation between MK and money, but people are already assuming causation from that. And for all those Stats/Science people out there, you should know that correlation in no way means causation.

The only data that was actually analyzed, and went into this detail was ignored on a very large scale by the community.

http://allisbrawl.com/blogpost.aspx?id=123595

I just wanted to try out MK and when I kept winning why bother changing ?

I don't blame you, Leon just wasnt adapting LOOL

edit:

If Leon switched to MK my day will be MADE. please. PLEASE. no more marth. tired of this bull**** idc even if I won the last 2 sets.

 

[Mr-R]

And soon enough Leon's gonna switch to MK too!

I didn't ''switch'' to mk, why is it that when I use snake/zss/falco/link in tourney people don't say stuff like that but when I use mk once '' I switched to him''

 

[Ripple]

I didn't ''switch'' to mk, why is it that when I use snake/zss/falco/link in tourney people don't say stuff like that but when I use mk once '' I switched to him''

because I don't think the US has seen anything but your marth and link.

 

[C.J.]

Ramin, your falco is booty. ZSS/Snake/Link/Marth are legit though.

But we all know you're an MK main.

 

[Cygnet]

@Eon the Wolf: You have a good point, that DOES sound like a good way to test out the true effects of the ban.

Although the longer the trial period lasts or the longer that MK stays banned, the harder it will be to overturn that decision, kinda like how a lot of people didn't want changes at this point in the metagame because it was already ~4 years in. It won't be 4 years in this case, but the community may be even less receptive than they are now after a few months.

And the money distribution of MK winnings to other characters assumes that every, or at least, a significant number of MK mains will learn a new character and reach the same level of prowess at that character as they had been at MK (to reduce confounding variables), which is not the case, because learning a whole new character can take a lot of time, and some MK mains may quit outright. Some may also just use many characters and distribute their winnings among them, making the effect potentially a lot more subtle.

@Orion: Thanks for showing me that link! (Makes me wonder why other people don't cite that as a resource more often.)

For people less willing to swim through the text, a lot of people think that MK usage is low at low levels and high at high levels, suggesting that using MK may increase winnings or that at high levels of play, MK is much more polarizing than at lower levels. The analysis shows that this is NOT true and that there is no statistically significant spike in MK usage at higher levels.

Note that this does not mean that MK usage is low at any level, it just means that MK usage does not increase as skill level increases.

(People should look at this!)

EDIT: Whatever MK banned tourneys are happening right now are kinda like trial periods too. Clearly, at the frequently cited Socal tourney where Tyrant played Dedede, Dedede did make a lot more money than that character would make usually.

Also the money distribution effect Eon the Wolf mentions could be diluted across multiple characters (cause not all MK mains will all flock to one character) and the effect might not be statistically significant in that case.

 

[Budget Player Cadet_]

I didn't ''switch'' to mk, why is it that when I use snake/zss/falco/link in tourney people don't say stuff like that but when I use mk once '' I switched to him''

SO MUCH THIS; thank you Ramin. ^_^