Why do we play 4 stock?

[THO]

Hello,

I remember when people use to play 5 stock and then it moved to 4 when the tournament scene really solidified. Its been that way for forever. A friend of mine plays other fighters, not just Melee, and came to town for a week so we played a crap ton of melee. We were talking about why melee is the black sheep of the fighting community and at first we thought it was because of the cartoon graphics, the non-standard KO design, the totally free movement, among other things that I won't get into. But then we thought the biggest difference in the game is how long matches last. We messed around with the idea of playing 3 stock (like in brawl) but then just said **** it and went 2 stock. This pretty much matches the live count for 2D fighters and my friend liked it more. We noticed it was way easier to keep up the intensity of the match among other things.

Why don't we play 2 stock regularly? When there are multiple players there is less down time between matches, tournaments would go faster (even if every match was changed to 3 out of 5 instead of 2 out of 3), suicides would matter less in tournament (in a 3 out of 5 system), and the game would be way more stream line. I am just curious to what you guys think.

Also, give it a shot. Its fun and really does feel (tempo wise) more like a 2D fighter which is way more intense because you only have two lives. Also, we got lazy way less than in a four stock match and people can't just win by getting a stock lead and crouch canceling for a victory which really does happen a lot.

 

[Fregadero]

A little late to the party.

http://www.smashboards.com/showthread.php?t=317222

 

[Warhawk]

Hmmm did Cactuar send you here? lol

 

[choknater]

four stock just feels right to me.

 

[King5280]

four stock just feels right to me.

I concur.


.

 

[Bones0]

Maybe other fighting games should have longer matches. DERP

 

[ph00tbag]

Typically, the argumentation given is that 4 stock matches strike the best balance between meeting time constraints, and forgiving accidental self-destructs. The idea is that when a player has self-destructed, it doesn't necessarily constitute a win by outperformance on the winner's part, and so the player who self-destructed should have ample opportunity to make up for the mistake, and as the stock count increases, the impact of one stock decreases, naturally. I think, to a degree, pride plays a part, as I have yet to meet a player that really feels a sense of accomplishment when they win by SD. They'll take the win, of course, but SDs are pretty universally treated as a blemish on an otherwise solid win.

Of course, the counter-argument is that being in control of one's character enough to maintain footing on the stage is of considerable importance in a game like smash (indeed, it's the entire point). For this reason, there is no reason an SD should be less valid than a KO in determining who is the more skilled player. With that in mind, the priority of meeting time constraints takes over. Naturally, you want the first player to lose a stock to have the opportunity to adapt to their opponent, so you concede the second stock as a second chance, but there's increasingly little reason to increase the stock count, the more you increase it by. In this mindset, two stocks really is more ideal, and if you can't keep your character on stage of your own accord, then you, simply put, ain't good.

Personally, I think the latter argument holds more objective weight. And I think it's gaining traction. However, there's a lot of momentum and precedence to the former; all results since roughly 2006, have obtained within the game outlined in the first paragraph, and the second paragraph describes a very different game. By changing the rules, you effectively invalidate all previous considerations of who is good at Melee. It's not a decision to make lightly, and it will naturally meet opposition from those who don't take well to change. And that is why we still play 4 stock.

 

[TheLake]

Better questions

why are matches 8 minutes long wtf!?

 

[Spyro]

Better questions

why are matches 8 minutes long wtf!?

Yeah, you almost never see matches get even close to being timed out.

 

[stelzig]

If you think there's too much waiting between matches when you do rotation, then just lower the stock count. You do not have to follow tournament rules in friendlies. (less stocks actually means more time wasted on loading screens though so you will technically play less in the end ).

Edit: And yeah I still oppose mere 2-stocks matches alot, and it has pretty much nothing to do with SDs... But I am done discussing it unless I see it actually starting to be used in tournaments (that I might attend)

 

[Max?]

Better questions

why are matches 8 minutes long wtf!?

Agreed. 5 minute timer or gtfo

 

[Shadic]

Five minute timer just means it'll be easier to time people out.

I'd rather see 5% of matches (imaginary number) go to time at 8 minutes than see 15-20% of matches (also imaginary) go to time at 5 minutes.

 

[KrIsP!]

Okay come on, he even uses a cactus for his avi.

Spoiler

He is Cactus incarnate

Might as well bring attention back to Cactuar's thread rather than here since about every point has been brought up and they're in an actual playtesting stage and trying to avoid more theorycraft.

Edit: I still want to see 3 stock 5 minutes with Japes, Brinstar and Mute back in.

 

[Spyro]

5 mins sounds a bit extreme imo, I would go with 7 or 6 mins.

 

[THO]

Thats funny that cactuar posted that the same week my friend was over. I don't think he goes on smash boards and I didn't realize it had been posted...I should have seached the forums more thoroughly before posting. Oh Well.

Oh, Phoot, me and Malk were thinking that suicides would be less relevant because suicides typically (unless the other player just gets off) lead to a win. By making each match less valuable, it makes suicides less valuable. So, theoretically in a 4 stock match suicides matter less than in a two stock match, they really do decide the outcome when both players are evenly matched. So moving the set count up to 3/5 and decreasing the stocks to 2 will decrease the effect of a single suicide in the overall set.

 

[Kal]

THO, what you're saying is that suicides have less of an impact because the matches have less weight. This is true. However, suicides clearly have more weight, a priori, when the number of stock is reduced. So you need to provide an argument that the number of matches actually outweighs the decrease in number of stock. I think it is the case that the total number of stock in a set would actually determine the impact of suicides. Best of seven with two stock grants 28 stock total, whereas past of three with four stock grants 24 stock total. So we expect the impact of self-destructs to be similar, though perhaps slightly less with the two-stock ruleset.

I think two-stock Melee is a good idea, and should at least be tested.

 

[Deleted member]

4 stocks is what defines melee BUT 2 stocks is better for training

 

[BairJew]

5 mins sounds a bit extreme imo, I would go with 7 or 6 mins.

2 words. Puff Ditto. You fall asleep after 3 mins of watchin' that.

 

[Johnknight1]

Spoiler

four stock

just feels right to me.

...that's what she said=??? But for realZ, I agree. 4 stock Melee just fits like Shrek Cinderalla and her glass slipper.

The one thing I don't like about 4 stock matches is sometimes it takes players a while to "get their feet wet." That is where I like to play 3 stock in friendlies a lot because of the intensity it gives to the match from the get go. The whole match is intense, and it gives off a "do or die" feeling from the get go.

However, I think 4 stock gives us many things 3 stock would not. More and better comebacks, lots of momentum swings, lots of in game adjustments (such as a smart player adjusting to a player who is repeatedly doing something), and some classic clutch performances where the smarter player just "knows" what the other player is doing.

Anyways, I don't know about changing the time limit. IMO whatever prevents super camping/stalling the best without making the game worse for matches without camping/stalling, let's go with that. Maybe higher or lower could be better, but for now, 8 is (mostly) fine. If statistics prove that a different time works better, we can roll with that.

 

[stabbedbyanipple]

Thats funny that cactuar posted that the same week my friend was over.

No, what's funny is that your avatar is also a cactuar.

 

[Cactuar]

Lmao. What is this even...

 

[ph00tbag]

Oh, Phoot, me and Malk were thinking that suicides would be less relevant because suicides typically (unless the other player just gets off) lead to a win. By making each match less valuable, it makes suicides less valuable. So, theoretically in a 4 stock match suicides matter less than in a two stock match, they really do decide the outcome when both players are evenly matched. So moving the set count up to 3/5 and decreasing the stocks to 2 will decrease the effect of a single suicide in the overall set.

The issue I take, and really what I was trying to get at with my post, is that changing the number of stocks changes what "evenly matched" means. You can't use four-stock melee skill metrics on two-stock melee match-ups.

Two "evenly matched" players in four-stock melee can have a match where A is about to four stock B, but B comes back and wins by one stock. That physically can't obtain in two-stock melee. If this happened in both match-ups of a 4 stock bo3 match, B would be clearly better than A, but if only the first two stocks lost mattered, then A would clearly be better than B. Do you see what I'm getting at? Neither way of playing the game is necessarily invalid, but they require you to reconsider how you measure the skill of a player. Suicides are just part of that, and they do play a different role.

 

[stelzig]

B would clearly have made an impressive comeback and A would clearly have taken two stocks in a row. Both of these things happen constantly. To conclude that one is the better player just based on so few stocks is silly in either case, lol. If I had to go with someone based on the situation you described though, then I would certainly go with B both because he was able to take more stocks when we tested for a longer amount of time, and he was able to take multiple stocks in a row while being behind.
Extremely silly to say that the number of stocks actually changes the way the game is played anyway. Might change some people's mentality on every stock, but it will be the same god damn game, just with less stocks and possibility for more counterpicks. You need to change the timer and consider adding stages, like cactuar wants to, if you really want to change the way the game is played.

Edit: On a sidenote, B could always be salty and go back to the same stage with two stocks, and be just as likely to take 2 stocks in a row (or 2 stocks first) as A - assuming they're about even on the stage (maybe in general), considering how back and forth your example went.

Edit2: Oh, you mentioned it happened 2 games in a row. That is first of all extremely unlikely and second of all not any different stiil. Just shows that the two players were extremely close in skill. The order the stocks are taken in do not really say much. Do you really think that some players are more capable of winning 4-stocks matches than 2-stock matches because they get more time (on a stage) to adapt or something? (this could POSSIBLY be true if we actually did add a bunch of silly gimmick stages) You would also be the first (afaik) to say that the match count shouldn't be higher as the stock count went down.

 

[Strong Badam]

TTWWADI
That's The Way We Always Did It

Change is difficult to cause in a community like ours where we've been playing 4-stock for like 9 ****ing years, and in these cases, while 4 is a relatively arbitrary number and was IIRC chosen as a compromise between 3 and 5 stock rulesets on opposite coasts, there needs to be a strong case to change to a different number.
It feels "right" to many Melee players probably not because it is, but moreso because they've been playing it for so many years.

Good luck if you want to change the standard though, would be interesting to see how the game would change if a number such as 2 or 3 were used.

 

[JPOBS]

Something is fishy

 

[ph00tbag]

B would clearly have made an impressive comeback and A would clearly have taken two stocks in a row. Both of these things happen constantly. To conclude that one is the better player just based on so few stocks is silly in either case, lol. If I had to go with someone based on the situation you described though, then I would certainly go with B both because he was able to take more stocks when we tested for a longer amount of time, and he was able to take multiple stocks in a row while being behind.
Extremely silly to say that the number of stocks actually changes the way the game is played anyway. Might change some people's mentality on every stock, but it will be the same god damn game, just with less stocks and possibility for more counterpicks. You need to change the timer and consider adding stages, like cactuar wants to, if you really want to change the way the game is played.

Edit: On a sidenote, B could always be salty and go back to the same stage with two stocks, and be just as likely to take 2 stocks in a row (or 2 stocks first) as A - assuming they're about even on the stage (maybe in general), considering how back and forth your example went.

Edit2: Oh, you mentioned it happened 2 games in a row. That is first of all extremely unlikely and second of all not any different stiil. Just shows that the two players were extremely close in skill. The order the stocks are taken in do not really say much. Do you really think that some players are more capable of winning 4-stocks matches than 2-stock matches because they get more time (on a stage) to adapt or something? (this could POSSIBLY be true if we actually did add a bunch of silly gimmick stages) You would also be the first (afaik) to say that the match count shouldn't be higher as the stock count went down.

http://en.wikipedia.org/wiki/Thought_experiment

 

[THO]

@kal I thought about that. It depends on the effect of a suicide in a single match. Depending on the time in a match a suicide can have very different impacts. Some suicides lead to strong leads for player 1, others lead to strong leads for player 2, some suicides even the match or bring them relatively close to even, and some change nothing (if you are about to get jv5 stocked, the other player suiciding changes very little).

If you assume both players are of equal skill, then, a suicide, irregardless of stock number, should statistically lead to the non-suicidal player winning the match. Lets do some back of the envelope math:
1 stock: P1 suicide leads to 100% chance of victory for P2
2 stock: P1 suicide leads to 66% chance of victory for P2
3 stock: P1 suicide leads to 60% chance of victory for P2
4 stock: P1 suicide leads to 57.1% chance of victory for P2
where suicides occur at 0% on stock 1.
This asymptotically approaches a 50% chance of P2 victory for a P1 suicide as stock count approaches infinity.

So, the worst case suicide would only lead to a difference in 8.9% chance of winning a given match between 2 and 4 stocks.

Lets now look at playing that over a set.

First lets look at playing best 2/3.
If P2 suicides in the first stock of any match of the set, P1 has a 53.55% chance of winning in a 4 stock match and a 58.3% chance of winning in a 2 stock match. This isn't looking great for 2 stock.
Now lets look at playing 3/5.
If P2 suicides in the first stock of any match of the set, P1 has a 51.33% chance of winning in a 4 stock match and a 53.12% chance of winning in a 2 stock match.
Now lets look at playing 4/7.
If P2 suicides in the first stock of any match of the set, P1 has nearly a 50% chance of winning in a 4 stock match and about a 51% chance of winning in a 2 stock match.

These numbers are based on two things. The first that the system can be described using discrete random variables and the the binomial distribution with a tweak for the suicide (which I can go into more detail if you are interested in my computations).

I think I should talk more about suicides. As mentioned above, the one I am considering the worst case suicide (by my possibly flawed definition). Lets also assume that players lose based of random chances of mistakes and that a suicide is when a random mistake leads to the immediate loss of a stock. The argument then is that all players make mistakes and the melee fighting system is that there is a random chance that such a mistake leads to a stock loss without the opponent doing anything. This is an suicide. They argue that this is far to punishing a mechanic and so the effect of the possibility of SD's should be minimized.

That out of the way.
If we play 2/3 on 4 stock and player 2 makes a worst case suicide then p1 has a 53.55% chance of winning the set.
If we play 3/5 on 2 stock and player 2 makes a worst case suicide then p1 has a 53.12% chance of winning the set.
Already we see that playing 3/5 2 stock minimizes the effect the suicide.
If we play 4/7 on 2 stock and player 2 makes a worst case suicide then p1 has a 51% chance of winning the set.
4/7 nearly eliminates the effect of suicides.

2/3 on 4 stock has 12 stocks for each player
3/5 on 2 stocks has 10 stocks for each player.
4/7 on 2 stocks has 14 stocks for each player.

I'm not sure how the timing works out in terms of average match length for 4 or 2 stock, but switching to 3/5 on 2 stock seems to reduce the effect of suicides below that of playing 2/3 on 4 stock, while at the same time reducing the total stock count and most likely the time of each set.

My math could be wrong. But yeah these are the numbers I'm getting. sorry for posting so long.

 

[Kal]

This is somewhat relevant, but it has nothing to do with SDs.

Probability of the better player winning in best of two, three, five, and seven sets, with two-stock, three-stock, and four-stock Melee.

However, I don't know how you've concluded that

1 stock: P1 suicide leads to 100% chance of victory for P2
2 stock: P1 suicide leads to 66% chance of victory for P2
3 stock: P1 suicide leads to 60% chance of victory for P2
4 stock: P1 suicide leads to 57.1% chance of victory for P2

In the two-stock case, while there are exactly three possibilities (P2 takes one stock, P2 takes one stock and loses one stock, P2 loses two stock), these three possibilities are not equally likely. Instead of modeling it so confusingly, it may be better to think of it as playing an n-stock vs. (n + 1)-stock match. As n approaches infinity, the probability of P2 winning clearly approaches .5.

Even assuming the correctness of your calculations, keep in mind that a 53.55% chance of winning the set is arguably having "eliminated the effect of suicides." Anyway, I'll be happy to analyze in more depth later what your odds of winning various gametypes are given a single self-destruct and using a negative binomial distribution.

 

[Jeyfar]

Hello,

I remember when people use to play 5 stock and then it moved to 4 when the tournament scene really solidified. Its been that way for forever. A friend of mine plays other fighters, not just Melee, and came to town for a week so we played a crap ton of melee. We were talking about why melee is the black sheep of the fighting community and at first we thought it was because of the cartoon graphics, the non-standard KO design, the totally free movement, among other things that I won't get into. But then we thought the biggest difference in the game is how long matches last. We messed around with the idea of playing 3 stock (like in brawl) but then just said **** it and went 2 stock. This pretty much matches the live count for 2D fighters and my friend liked it more. We noticed it was way easier to keep up the intensity of the match among other things.

Why don't we play 2 stock regularly? When there are multiple players there is less down time between matches, tournaments would go faster (even if every match was changed to 3 out of 5 instead of 2 out of 3), suicides would matter less in tournament (in a 3 out of 5 system), and the game would be way more stream line. I am just curious to what you guys think.

Also, give it a shot. Its fun and really does feel (tempo wise) more like a 2D fighter which is way more intense because you only have two lives. Also, we got lazy way less than in a four stock match and people can't just win by getting a stock lead and crouch canceling for a victory which really does happen a lot.

People love comebacks, try that in a 2 stock match, it's impossible to call something in a 2 stock match a "comeback", with the exception of a few things like a 50% 1 stock Ganon beating a new 2 stock Shiek. Also it would be unforgiving for simple suicidal mistakes, or getting pushed off the ledge by something stupid and losing one stock very miserably.

Another tidbit of information is adaptability to the opponent. A mindgaming and reading mentality would have a hard time adapting to strong opponents in higher levels of gameplay when it is just 2 stocks. There are people that on a 4 stock match they lose 2 stocks by a beating, but then suddenly destroy their opponent easily, hence adaptability and reading. (A small example is studying your opponents teching mindsets, which is crucial for a character like Captain Falcon)

Better questions

why are matches 8 minutes long wtf!?

Look at a Peach vs Samus match, you'll understand why.

But on another note it's just nearly impossible to get to 8 minutes, I think that maybe 6-7 minutes are more than enough.

 

[SSBMLahti]

2 stocks is definitely not enough lol...

 

[Massive]

Until we've come up with some model for the average events of a smash match, all of these numbers are speculation.

The only thing we can say about theoretically equal smashers is that they both have a 50% chance of winning the match. We can infer that they will each have a 50% chance of taking the first stock from even footing, but past there is a level of randomness we have yet to develop a model to account for.

 

[Kal]

Obviously, any model will have problems. This doesn't mean we should disregard them. Instead, we just need to take our conclusions with a grain of salt. I think, until we get a better model to use, the negative binomial is a good one to base minor choices on.

 

[Bones0]

Until we've come up with some model for the average events of a smash match, all of these numbers are speculation.

The only thing we can say about theoretically equal smashers is that they both have a 50% chance of winning the match. We can infer that they will each have a 50% chance of taking the first stock from even footing, but past there is a level of randomness we have yet to develop a model to account for.

This isn't really true. Characters and players have varying levels of skill depending on how long the match has lasted and how much % each character has. Just because two players are evenly matched doesn't mean they have an equal distribution of skill. Let's say player A is notorious for his adaption, and player B is notorious for his ability to gimmick people into gimps and other traps that only work once. Player B is much more likely to take the first stock, but by the end of a 4-stock game, perhaps it evens out. If they played an 8-stock game, player A would begin winning because B would run out of tricks as the match progresses.

Then you have characters that thrive on trading or comboing or KOing. Marth may be able to combo Fox to a million %, but he tends to struggle KOing him, whereas Fox doesn't have combos as vicious as Marth's, but he has many ways of landing KOs. When Fox loses a stock, he can respawn and get a KO fairly easily compared to Marth who will still have to work really hard regardless of %. That means Fox would probably benefit over more stocks because he is able to rack up more % vs. a newly spawned Marth than vice versa. The same idea works with Peach who is really good at trading. The first stock of Falco vs. Peach is completely different from the second stock because Falco gets wrecked by trades when he is at high % and struggles to kill Peach when she is at high %.

Obviously these are just examples so the specifics aren't really what's relevant; the point is still there. The game gets changed significantly by the stock count, and viewing a 2-stock game as half of a 4-stock game is a horribly misleading perspective.

 

[THO]

@kal
My thought process was incorrect, thanks for calling me out. Let me try again and this time I will spell out where I am getting my numbers.
In a one stock match, given a first turn suicide from p2 the probability of p1 winning is 100%
In a two stock match, given a first turn suicide from p2 there are 3 possible outcomes. P1 wins the next stock, P2 wins the next stock and then P1 wins the final one, or P2 wins the next two. This results in a 75% chance of P1 winning.
In a 3 stock match, given a first turn suicide from p2 there are 10 possible outcomes, I'm won't spell them out but it is pretty simple to see them.
This results in a 68.75 chance of P1 winning.
In a 4 stock, the probability is 65.63%.

I believe this is right and is really just the probability that a player will win a majority of the possible stocks given they won the first stock.

With these new numbers:
The probability of P1 winning a 2/3 with 4 stock after the suicide is 57.8%
The probability of P1 winning a 3/5 with 2 stock after the suicide is 54.69%

Now I could be messing this calculation up.

This problem is really interesting, and if you do have time do run some calculations I would appreciate your take on the problem and, if necessary, feedback on where I went wrong (which is very possible).

 

[Kink-Link5]

You assume too much. You assume each event is equally likely to occur, when a vast number of variables make this probability extremely screwy and nearly impossible to really gather.

 

[Beat!]

2 stocks would suck imo.

I guess it could be "fun" (read: not for serious competitive games) but we already have Cactuar's alternate ruleset for "fun" purposes and it already runs 2 stocks so just play that if you want to.

 

[Massive]

This isn't really true. Characters and players have varying levels of skill depending on how long the match has lasted and how much % each character has. Just because two players are evenly matched doesn't mean they have an equal distribution of skill.

@kal
In a two stock match, given a first turn suicide from p2 there are 3 possible outcomes. P1 wins the next stock, P2 wins the next stock and then P1 wins the final one, or P2 wins the next two. This results in a 75% chance of P1 winning.

The issue is that the odds of all of those options are not exactly 1 in 3.

Percentage differences, stage differences, and character matchup differences skew the results of those events in an unpredictable way. This is what Kal was talking about.

 

[THO]

I'm am simplifying the entire problem to a flip of a coin for each stock.

Modeling this type of problem obviously gets out of hand extremely rapidly when you try to take into account other variables. It is clear that the game isn't just repeated coin flips, but that doesn't mean it can't be modeled this way. The goal of this method is to demonstrate different methods of diluting the effects of SDs by defining them as outliers to the appropriate expected consequence of a mistake. We aren't trying to model the whole game, we just want see which match/stock number favors diluting outliers better.

 

[Bones0]

The coin flip model is so far from reality that whatever conclusions you draw from it about SDing frequency is irrelevant. It's also oversimplified in the sense that the impact of SDs is only one of a million things that gets affected by reducing stock count. It's fine if you want to create a model and just take it with a grain of salt, but you'd need a mountain of salt for any model that equivocates Melee with coin flipping.

 

[Kal]

THO, I will see about writing a program to spit out a table with some relevant data. Something like a table of the probabilities that the player who is down a certain number of stock from the beginning will win, assuming he and his opponent are "equally matched."

The issue is that the odds of all of those options are not exactly 1 in 3.

Percentage differences, stage differences, and character matchup differences skew the results of those events in an unpredictable way. This is what Kal was talking about.

Well actually, I am simply pointing out that the three possibilities he mentioned are not all equally likely, even under ideal circumstances (where you basically assume every stock is a coin-flip).

These other factors obfuscate the issue. They are relevant if you want some sort of robust model but, as THO has said, we're asking a pretty simple question: how do self-destructs impact results based on the number of stock? Using a simplistic model for such a simple question is arguably ok.

The coin flip model is so far from reality that whatever conclusions you draw from it about SDing frequency is irrelevant. It's also oversimplified in the sense that the impact of SDs is only one of a million things that gets affected by reducing stock count.

Except SDing is the only thing we're discussing. We aren't saying that negative binomial with p = .5 is a robust, dynamic model of Melee. We're using it to answer a single question: how do self-destructs impact the game as a function of number stock?

It's fine if you want to create a model and just take it with a grain of salt, but you'd need a mountain of salt

Really? A mountain of salt?

for any model that equivocates Melee with coin flipping.

We're not "equivocating" Melee with anything.

Do you complain to an architect that the Euclidean geometry he is using is inaccurate because the Earth is not actually a plane? If he is designing something that will occupy 2/3 of the planet's surface area, you might. When he's designing a suburban home, you can safely make these assumptions. And what we are doing here is analogous: this model is not some attempt at modeling results of the entire game. We're trying to figure out one single thing: the impact self-destructs have on the game as a function of the number of stock.