Legality Tentative: MBR Official Ruleset for 2012

[EbAgItachi]

o man

he spent years studying statistics

He's also right.

 

[danieljosebatista]

Yeah he is right.

 

[Varist]

dat ***** so right.

 

[Luma]

nah he is wrong

i can win against people better then me
i can lose against people worse then me

you cant explain that

 

[Fortress | Sveet]

I've already explained what's wrong with this logic. Your "average" assumes each possible outcome is equally likely. This is not a valid assumption. Assuming equally likely outcomes is never valid if the subset under consideration can be divided into dissimilar groups. I can spell out those groups for you: sets between players of roughly equal skill, and sets between players of differing skill. As I said, 3-2's would only occur between two players of roughly equal skill, whereas 3-0's would occur between two players of different skill. The length of individual games is also heavily dependent on whether the two players have roughly equal or differing skill levels, something you have completely ignored. The majority of tournament matches are not played between two players of roughly equal skill because the majority of a given tournament's attendees do not have roughly equal skill. The three types of averages (mean, median, and mode) all take into consideration the frequency/likelihood of each possible outcome. Since you ignore the frequency/likelihood of each possible outcome, your "average" is not actually an average. Don't throw around big fancy words like "statistics" when you don't even know what an average is.

tl;dr Despite not having any basic knowledge of statistics, you are trying to argue statistics with someone who has spent years studying statistics. There is no other way to say it: you are just plain wrong.

So why are you still arguing terminology instead of just correcting my numbers? In day-to-day life the word "average" refers to the mean average. Also its simply an estimate. Its purpose is in no way to simulate reality to the degree of accuracy you need when estimating electron positioning.

In your expert opinion how much variance is your more accurate estimate from my 30 second estimate?

nah he is wrong

i can win against people better then me
i can lose against people worse then me

you cant explain that

Also this.

Ferrish, you assume 3-2s only happen between two equally skilled opponents. I completely disagree as I've 2-0'd and lost 2-3 to kels on separate occasions and that guy is miles ahead of me. You simply can't explain how sets and games go individually, it relies much more on the players' styles than anything else. My estimate had no intention of solving these problems, thus I simply made up some numbers and put them through a basic averaging function.

 

[Kal]

If you read a few pages back, I give a perfect explanation of why the counterpicks were banned.

Is this the one where you explain that a combo started a 7:45 could work while one started at 7:43 could not? I believe we've come to terms on this one, with an agreement that I'm simply going to not john, instead trying to better learn when I should start my combos, whereas you feel this warrants a ban.

I can see an argument being made for the randomness, but I've said before that I think the randomness on some of the stages (particularly Brinstar and Mute City) can be accounted for, and is also small enough (in terms of effect) to not warrant a ban.

Ferrish, you assume 3-2s only happen between two equally skilled opponents. I completely disagree as I've 2-0'd and lost 2-3 to kels on separate occasions and that guy is miles ahead of me. You simply can't explain how sets and games go individually, it relies much more on the players' styles than anything else. My estimate had no intention of solving these problems, thus I simply made up some numbers and put them through a basic averaging function.

If you want to avoid an overly complex model for predicting the results of sets, there's nothing wrong with assuming a large skill-gap will result in a 3-0. No, it's not universally true (as exemplified by your anecdote, or any other nearly anyone can provide), but it's a useful rule of thumb to avoid obfuscating the issue.

A more complex model could, for example, apply ELO ratings to every player and then indicate the possible outcomes and associated win/loss results with attached probabilities according to the difference in ELO rating. However, creating a model like this, even by someone with a heavy background in statistics (alas, my background is in Algebra), would require a much larger sample size than what we have. It also would be a pointlessly rigorous proof for an extremely simple (and somewhat subjective) issue: whether having best of five will significantly increase the duration of tournaments.

If Ferrish wanted to be more precise, he would have said "the majority of 3-2s occur between players of similar skill" or "the probability that a set is 'close' decreases as the skill-gap between players increases." Basically, get off your high-horse: your anecdotes don't disprove anything, they merely point out that Ferrish was not using precise terminology. Since, as far as I'm aware, it is not a scientific nor mathematical article he is posting, that is not something to get worked up over.

 

[crush]

In this situation a good idea for the smash community may be blame fox?

 

[Kal]

I hope Armada pulls a Bob-bomb in a clutch moment and it sparks the flames of stupid-debate on SWF. We'll have ten to fifteen threads titled "Should Peach be banned?"

 

[Fortress | Sveet]

Peach is secretly the best character in the game.

 

[Kal]

If by "secretly" you mean "obviously."

 

[Fortress | Sveet]

Kal i just saw your post to me (must have been edited in).

I think its funny that you are telling me to get off my high horse of terminology when I had made no claims about terminology and in fact he was the one nit picking mine.

Again, my model was simply a basic one to show that time will increase a lot if the TO increases the number of games. If you want to make a better model, do it. I don't think anyone is arguing this point and thus Im not really sure why they are arguing at all.

 

[Kal]

Yeah, I did it edit it in. I didn't see an appropriate way to place it in my most recent post. Sorry about that.

Anyway, my point is that, while you haven't explicitly said anything about terminology, that's what you are getting caught up on in your latter point. The fact that examples exist where a player miles ahead of another goes 3-2 does not invalidate Ferrish's point, because his point is about probabilities. It's no different than observing that a better player will usually beat a worse one. What you're doing is effectively the same as assuming that he's saying that a better player always beats a worse one. When discussing real-life probability, 0 and 1 do not come up that often, and Ferrish, having a background in Statistics, knows better than to suggest that 3-2s universally occur between players of similar skill.

No one is getting caught up on the model. The point we're addressing is this idea that the average best of five will result in a 3-2 result. The fact is, you have no better reason to believe that than any other result. However, Ferrish makes a good point: the variation in skill at most tournaments (especially national ones) is huge, and so it's arguably much more likely that a set's result would instead be 3-0 or 3-1 than 3-2.

 

[Fortress | Sveet]

... my model uses the average number of games in a set. For a best of 5 thats (3+4+5) / 3 = 4 games. My estimate used 3-1 not a 3-2.

 

[Beat!]

*awkward silence*

 

[BairJew]

This ruleset seems odd. Finally very glad RC has been removed never fully understood why that was ever a CP and YI64 was not. Brinstar also confusing slightly, and it seems all these stages in general now highly benefit spacey players now that Brin is gone they don;t have to worry about Puff/Peach players messing up a perfect 2-0 best of 3 set. all they have to do is Ban DL64 now and these Puff/Peach players are quite screwed. Now it seems the only CP left is PS which basically allows Fox/Falco mains to wipe the floor with whoever they play. THis has highly messed up the play for other player and seemingly might allow the community to heavily lean towards maining spacies, therefore, the outcome being a bracket of all spacies and occasionally the lonely marth player. It seems that more stages should be thrown into consideration to allow the prospects of every character in the top region to benefit not just the top 2. however, this being just my opinion people are bound to highly disagree. I feel that Crimson had a good point in his argument in the beginning and if such drastic changes are constantly being made to the stage selection within the community we should just find one stage such as BF and only play there. W/e though hopefully this isn't set in stone.

 

[Strife]

This ruleset seems odd. Finally very glad RC has been removed never fully understood why that was ever a CP and YI64 was not. Brinstar also confusing slightly, and it seems all these stages in general now highly benefit spacey players now that Brin is gone they don;t have to worry about Puff/Peach players messing up a perfect 2-0 best of 3 set. all they have to do is Ban DL64 now and these Puff/Peach players are quite screwed. Now it seems the only CP left is PS which basically allows Fox/Falco mains to wipe the floor with whoever they play. THis has highly messed up the play for other player and seemingly might allow the community to heavily lean towards maining spacies, therefore, the outcome being a bracket of all spacies and occasionally the lonely marth player. It seems that more stages should be thrown into consideration to allow the prospects of every character in the top region to benefit not just the top 2. however, this being just my opinion people are bound to highly disagree. I feel that Crimson had a good point in his argument in the beginning and if such drastic changes are constantly being made to the stage selection within the community we should just find one stage such as BF and only play there. W/e though hopefully this isn't set in stone.

Spacies are gonna wipe the floor with the rest of the community now.

Maybe this was done so to give Mango and PP a better chance of beating Armada in the future.

 

[BairJew]

Spacies are gonna wipe the floor with the rest of the community now.

Maybe this was done so to give Mango and PP a better chance of beating Armada in the future.

This is very true it really does inflict a giant wound on this community now that the options for other characters to stand a chance in bracket has now diminished. I was hoping for stages to possibly get added for more options instead of yet again taken away. However RC was a good decision never fully understood the reasoning for that to begin with it somewhat creates a loss for those who tried to improve their game on that stage in no avail to beat spacies.
The idea that this was done to improve Mango and PP against Armada seems very odd even though it can be seen as true. I just find it odd that Armada seemed to struggle against the Falco matchup anyways and all Mango had to do if he was in a stick was go Puff since the Peach Puff Mu is terribly in Puff's favor I doubt that this stage selection would benefit either side rather than inflict both sides when choosing Puff/Peach or any floaties for that matter. It's a very odd decision for this rule change I wish it would be thought over before drastically improving the spacey game with this choice in the community.

 

[Luma]

spacies are the best chars in melee, so ofc they have an advantage when we got a neutral stage pool

the stage pool is not meant to "balance" the game

 

[Battlecow]

I say we slash PS

If we're trying to balance the game, which we shouldn't be, then slashing PS takes the advantage away from spacies

If we're trying to pick stages that encourage a desirable metagame, which we should be, slashing PS gets rid of a janky random-filled stage

No reason to keep it other than to pander to the tastes of all the spacies mains

 

[Strife]

spacies are the best chars in melee, so ofc they have an advantage when we got a neutral stage pool

the stage pool is not meant to "balance" the game

But as it is the stage pool is benefiting the spacies so it's neutral. You don't think that it's a problem that the only cp in the game is a stage that spacies(already the best chars) are dominant on, and they removed a stage in which spacies struggles on?

 

[BairJew]

@Battlecow thats basically what I'm trying to get across PS needs to be gone.

@Luma bringing up point by Strife, it seems they kept PS which spacies wreck every character on and got rid of other stages which spacies struggle on. It seems that the decision for stage banning has been made on a bias towards spacies due to the continued shrinkage of the stage selection to accomidate spacies compared to other characters. Therefore your saying this game should not be balanced by the stage pool, if that is so then you are implying that you are Pro spacies I fear this game is taking slow turns in the wrong direction.

 

[Luma]

why is everyone talking like ps is a freewin for spacies against any other char? o.O

they have a slight advantage yeah, but nothing gamebreaking

and btw i main sheik and peach, why would i be pro spacies?

 

[BairJew]

Its not a freewin. Its just so highly in there favor in almost every mu that it can be seen as something more than a slight advantage.
Also, this is a pretty janky stage and basically every transformation the stage has would be banned. Which just adds to the need for this stage to be not in the CP selection.

 

[crush]

yeah i definitely agree with bairjew

<33 bairjew

 

[BairJew]

^ <33 crush 4 Ever

 

[Doser]

They removed Brinstar (Floaty CP) and Rainbow Cruise (Spacie CP), both of those stages were automatically banned by the vast majority of players as far as I know.

 

[crush]

<33 bairjew 4 ever

 

[Kal]

... my model uses the average number of games in a set. For a best of 5 thats (3+4+5) / 3 = 4 games. My estimate used 3-1 not a 3-2.

Not really the point. The average (mean) number of games in a set isn't necessarily 4 just because (3 + 4 + 5)/3 = 4. Games with four matches and five matches may not occur as often. You respond with the fact that your model is simple, and that we should provide a better one, and that's what Ferrish does, albeit with some vagueness: sets with three matches are most likely to occur, followed by sets with four matches, followed finally by sets with five matches. Your assumption that the average is just (3 + 4 + 5)/3 requires that sets of three, four and five matches all occur in equal proportion. That is the issue Ferrish has brought forward against your model.

Don't get caught up on my misquoting you, because, while the issue is the average you've acquired, it's also the methodology you're using to acquire the average. Supposing, for example, that half of all sets are 3-0, and the rest are evenly split between 3-1 and 3-2, then our average is (.5)(3) + (.25)(4) + (.25)(5) = 3.75. This is just a guess, of course, and the probability distribution of set length is likely different.

 

[Strong Badam]

I say we slash PS

If we're trying to balance the game, which we shouldn't be, then slashing PS takes the advantage away from spacies

If we're trying to pick stages that encourage a desirable metagame, which we should be, slashing PS gets rid of a janky random-filled stage

Guess we should ban Yoshi's Story and FoD.
Random filled stages are sooooooo bad

 

[BairJew]

^ I don't think FoD of Yoshis compares to the rock or fire transformation on PS lol

 

[tarheeljks]

One of the best ways to improve is to constantly attend tournaments. That is not the same thing as one of the main purposes for a tournament is improvement of the players.

since you said main i don't really disagree enough to argue about it, but you originally said "the" purpose


edit: shoutouts to non-uniform distributions as well wrt average # of games in a set, though i see this has been addressed already now

as far as the # of minutes a set takes, character matchups will have an effect in addition to skill gaps

 

[Fortress | Sveet]

I'd rather see FD slashed than PS.

 

[crush]

I think randall does. (Sorry for disagreeing with you bairjew <33)

Edit: oops meant to say this to bairjew

 

[BairJew]

I'd rather see FD slashed than PS.

It seems like a good idea however I'm curious on your reasoning?

I think randall does. (Sorry for disagreeing with you bairjew <33)

Edit: oops meant to say this to bairjew

And that is a good point point good sir I might agree with you on that one <3 Crush

 

[Fortress | Sveet]

Not really the point. The average (mean) number of games in a set isn't necessarily 4 just because (3 + 4 + 5)/3 = 4. Games with four matches and five matches may not occur as often. You respond with the fact that your model is simple, and that we should provide a better one, and that's what Ferrish does, albeit with some vagueness: sets with three matches are most likely to occur, followed by sets with four matches, followed finally by sets with five matches. Your assumption that the average is just (3 + 4 + 5)/3 requires that sets of three, four and five matches all occur in equal proportion. That is the issue Ferrish has brought forward against your model.

Don't get caught up on my misquoting you, because, while the issue is the average you've acquired, it's also the methodology you're using to acquire the average. Supposing, for example, that half of all sets are 3-0, and the rest are evenly split between 3-1 and 3-2, then our average is (.5)(3) + (.25)(4) + (.25)(5) = 3.75. This is just a guess, of course, and the probability distribution of set length is likely different.

How is assuming they have equal probability wrong? Do we have numbers saying otherwise?

 

[Varist]

did i just ****ing hear a consensus on slashing one of the two purely non-random stages we have?

****, first it was all the CPs, now it's our god damned final destination?

**** you guys, Kal was right, ruining this game. ruuuuuining this game.

at least put MKII back in, you slippery sloping douchenozzles

 

[Cactuar]

Lmaoooooo. Seeing some pretty salty posts.

 

[danieljosebatista]

Honestly, as far as ferrish's numbers go, his model is absolutely correct. Kal is also right in his explanation of ferrish's model. Sveet, simply saying "do we have numbers showing otherwise" does not make him wrong. In fact his model is far more comprensive than the one you provided. Not trying to offend, just stating facts. Also @ cactuar, these threads are salty of course, but the mbr should ultimately pay attention to the best arguments. Those belong to ferrish snd kal. That is my opinion

 

[FerrishTheFish]

How is assuming they have equal probability wrong?

As I (and now many other people) have said, assuming equally likely outcomes is not valid when the subset under consideration can be divided into dissimilar groups. If you agree that you can change the probability that a set will last 3, 4, or 5 games just by changing which two players are playing the set, then you agree that you cannot assume equally likely outcomes.

Some other people have also pointed out that you could probably also change the probability that a set will last 3, 4, or 5 games by changing which characters the two players use, or changing which stages the games are played on. You could also change the expected length of an individual game by changing the players, the characters, and the stages. You see how complicated this can get? Not to mention that player choice, character choice, and stage choice all influence each other. You really think you can ignore all that and just say all outcomes are all equally likely?

Not that I need numbers or a better model to disprove an invalid assumption, but I honestly suspect that I can apply a Poisson distribution to the number of individual games that the winner of the set lost (0, 1, or 2), and I can practically guarantee that I can apply an Exponential distribution to the length of an individual game (controlling for stage and character).

Do we have numbers saying otherwise?

You realize this was my whole argument, right? That you can't claim that bo5's automatically result in a 3-hour increase in tournament length without some numbers backing up that claim? As far as I'm concerned, if Kal says he runs bo5 tourneys with 20~30 entrants that finish in 4~5 hours, that proves it is possible for other TO's to do so as well.

 

[Luma]

since 20-30 guys are the same as 300....