Why do we play 4 stock?

[zZz]

2-3 should be the standard

 

[Wizzrobe]

I think 4 stock is fine.
Just cause other fighting games have shorter matches that does not mean that melee should.

 

[THO]

@kal
if you don't have time that is cool. I just wanted to know if I was getting something close to the right answer. I appreciate your feedback.

 

[Kal]

I pretty much never let a math problem go, so I will probably have something concrete posted by the end of the week.

 

[Kink-Link5]

Even a coin flip is not an accurate model because you have to assume each of the events is likely equal to occur. You never specified a fair coin being flipped, just as it isn't specified what makes each possibility within the aftermath SD's equally likely. If the events can be concluded as such, your numbers have merit, and if not, they don't. That's all there is to it.

 

[Jeyfar]

Probabilities are all wrong. There is not a single certain way of knowing which player will win. Remember Melee has a random chance that, even though it can be found in other games, Melee has a random element unlike any other game. A single mistake can be mortal in Melee, not only suicides, but a missed L-cancel can lead to demise. Take for example Ganondorf vs Fox. Fox has match-up advantage, but a missed Punishment by Fox can lead to big trouble, dthrow-->Gerudo dragon (side b)-->thunder drop(dair)--> tech chase thunder drop--> skull crusher(fair) which in all equals death for Fox (read: free stock). And a missed L-cancel by Ganon leads to shine ****. All in all you could probably get to approximations by doing a formula based on the following:

P1 Skill + - Character matchup dis/advantage + - stage dis/advantage = certain point divided by P2 points.

Let's say P1 has a high skill level (on a scale of noob, average, high and pro skill all worth form 10 to 40 respectively in multiples of 10) so he has 30 points. So he is a Falco vs a Ganondorf, which has a 30% advantage by character alone, but unfortunately for the Falco the stage is Yoshi's Story, so he has now a - 10. So we have:

30+30-10=50
As for the Ganon we have: High skill as well
30+0+10=40
(character advantage being a difference in match-up so Ganon has 0 advantage by char)
so 50/40=1.25 which equals a 25% higher chance of the Falco winning.

So um it's just an approximation but I think it's close.

 

[Kal]

Even a coin flip is not an accurate model because you have to assume each of the events is likely equal to occur. You never specified a fair coin being flipped, just as it isn't specified what makes each possibility within the aftermath SD's equally likely. If the events can be concluded as such, your numbers have merit, and if not, they don't. That's all there is to it.

Why do people find it useful to come in and mention that our simplistic model isn't complex enough?

For the record, a coin-flip generally refers to a fair coin, unless otherwise specified. It's sort of useless to refer to something as an unfair coin-flip, since you may as well just go ahead and assign the probability sans analogy.

We are also making the assumption that the events are equally likely. No, this isn't accurate. That doesn't make deductions useless. If being totally accurate is really important to you, we'll officially say they apply to Falco dittos on Final Destination. Except not really, because we're not trying to come up with obnoxiously complex models to answer simple questions and make small decisions.

Probabilities are all wrong. There is not a single certain way of knowing which player will win.

Isn't that what probabilities are for in the first place? Anyway, we're not trying to come up with any sort of robust model for predicting the outcome of sets. We're just trying to figure out the impact of suicides as a function of number of stock assuming a negative binomial distribution with p = 0.5 (i.e., that your chance of taking a stock is fixed at 50%).

 

[Jeyfar]

Isn't that what probabilities are for in the first place? Anyway, we're not trying to come up with any sort of robust model for predicting the outcome of sets. We're just trying to figure out the impact of suicides as a function of number of stock assuming a negative binomial distribution with p = 0.5 (i.e., that your chance of taking a stock is fixed at 50%).

My apologies I didn't know. As for the function. I suck at functions :/

 

[Johnknight1]

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.

I actually use to play a lot of 5 and 6 stock. Heck, I still play up to 20 stock matches. About four years ago I started playing 4 stock and it felt perfect. With 5 or 6 stocks it felt like too much. However, I wouldn't mind attending a 3 stock tournament on a common basis.

2 stocks would suck though. No comebacks, very little "feel" for each game, not many in game adjustments, and not as potential for epic mind games happening. Plus I don't want a match that could end in 1 minute and then spend a longer time on the character select screen. The shortest 4 stock match I've ever seen (rage quits aside) was 2 1/2 minutes.

4 stock matches allows for the matches to be just shot enough without being too short, just long enough without being tedious (like baseball), just long enough for lots of big in game adjusts, not too short enough where everyone has to totally change their style mid-game on a regular basis, just long enough for epic comebacks, just short enough for blowouts to not be too common, and just short enough to where huge blowouts aren't completely boring.

If you guys want 3-stock matches to become standard, have some tournaments with those rules, put up matches, and let's see what it's like.

 

[Kal]

My apologies I didn't know. As for the function. I suck at functions :/

Don't worry about it. I like your idea of assigning values based on player skill, the character matchup, and stage choice, in order to calculate the probability of winning. I just don't see a way to implement it meaningfully without making a lot of arbitrary decisions.

 

[Jeyfar]

Don't worry about it. I like your idea of assigning values based on player skill, the character matchup, and stage choice, in order to calculate the probability of winning. I just don't see a way to implement it meaningfully without making a lot of arbitrary decisions.

Yes that is rather troublesome, there is no surefire way to know how it a suicide could affect the differences between 2-3-4 stock matches, the only way is adding a 50% effect on the match if you add suicides to the equation, but functionally or exponentially, it is quite difficult. As you said it is arbitrary and this is not pokemon with a catch rate formula (check this **** out http://bulbapedia.bulbagarden.net/wiki/Catch_rate ). So in the end we would have to base ourselves in approximations and i have no divine idea of how to do an equation like that

 

[Kal]

There are some easy (but simplistic) ways to do so. Do you know what a Binomial Distribution is?

 

[greenlink]

mathboards

 

[Jeyfar]

Not quite, but I do know it would present statistical data in which it would demonstrate the effect of the desired knowledge. In an exponentially increasing (or decreasing) function. But I have no idea on how to do it. But in the end I really do not know (other than the simple math I did earlier) how to get the possible results of a 4 stock smash and how suicides (which I still think affect at a fixed rate) affect the outcome.