Combos

[SSBPete]

my yoshi will **** all u guys when i get back. seriously.

 

[ballin4life]

What's uair falcon kick?

 

[asianaussie]

I falcon kick edgeguard all the time. Best edgeguard ever against noobys.

I look forward to playing a good Yoshi.

What's uair falcon kick?

Epic friendlies move. U-Air, jump and Falcon Kick. Completely inferior to a continued U-Air > Up-B/jump D-Air near a ledge. It looks cool though.

 

[Blue Yoshi]

Uh... seeing as I haven't read this thread until now, here's my long delayed responses:

First off, the puff on Pikachu combo is not a true combo. Pikachu has ages to escape rest. I think I figured it out to be 20% before Pikachu can't be combo'd out of up-throw (I know it's f-throw, but I'm calling it up-throw).

Rock paper scisors actually does take skill to play well. It's not just a luck and chance game. Mindgames are definitely involved. That is... unless the opponent plays truly random (like off the results of the roll of a die).

Everyone will have their own personal playstyle no matter how good they are. Some people will prefer to extend combos as much as they can, while others will end combos early... both resulting in the kill. If someone were to be as good as Isai, he would not play every character exactly the same way as Isai. They would have their own playstyle.

 

[blaze3927]

jiggly puff has the best combos on pika, his easiest one is only three moves and its 0 to death.
(puff>pikachu :D)

 

[MiketheMike]

my yoshi will **** all u guys when i get back. seriously.

mines better

 

[SSBPete]

Uh... seeing as I haven't read this thread until now, here's my long delayed responses:

First off, the puff on Pikachu combo is not a true combo. Pikachu has ages to escape rest. I think I figured it out to be 20% before Pikachu can't be combo'd out of up-throw (I know it's f-throw, but I'm calling it up-throw).

.

thanks i'll take it off. i'll replace it with a yoshi combo.

 

[Signia]

you guys were getting into some serious game theory talk earlier. A symmetric game like rock/paper/scissors does not have optimal strategies because every option is a nash equilibrium. A nash equilibrium strategy is one where the opponent cannot change his strategy in order to improve his payoffs. An easy example of this is guarenteed combos. Obviously the best strategy is continue your combo as long as you can once you start it, (unless you put yourself in danger while doing it) because it gives you the best payoff even if your opponents knows what you're doing. That's the key for creating optimal strategies: even if your opponent knows what you will do, they cannot do anything to improve their situation if you choose this strategy.

This is less complicated than it sounds, because there aren't very many nash equilibria in smash (if there were, the game would become degenerate with very few viable strategies). Playing to nash equilibria is just playing the game not like an idiot. Grab the edge when they will only barely reach it. Do another uptilt. And another. Because there is a "right" way to play to some degree, and since the game is not symmetric, it is different than rock/paper/scissors.

And yes, you CAN choose the mathematically best choice, but, if it is not a nash equilibrium choice (the best choice no matter what), and you choose it every time, the opponent will just change their strategy (choices) to improve their payoff (counter it). All this should be obvious.

But this is what makes smash interesting: There are so many options, but, some are better than others, which keeps your decisions from feeling arbitrary, but there options to counteract those same options to keep the same strategies from being played over and over. That would be boring. This also keeps the game from being "solvable." Analog controls make every situation unique, so a complete model of how play in every situation would be hopelessly complicated. This complexity is what we call depth.

Unfortunately (or fortunately), this make theory surprisingly useless. You need theory to have any sort of game plan, but specific situations are so complex and esoteric that you need to act on instinct (experience). Which, finally, brings me to my point.

Smash is so complex that is hardly matters what the best option is --- all that matters are the options that beat what you think your opponent will do (with the exception of nash equilibria). In other fighting games it's like: ok here's the best option, here the second best that counters its counter, and maybe you'll get a third "they'll-never-expect-this-one-but-it's-risky-and-I'm-probably-better-off-choosing-either-of-the-first-two-but-it's-so-unexpected-it's-practically-guaranteed." In smash it's like "I know they'll probably want to use option 1, because it covers options 2,3,4, but I use option 5 I'm gonna have even more options covered.. but If I use option 11 it covers everything... except option 15 ... ... but they probably won't do that.. ****"

 

[asianaussie]

It's fitting that your avatar is Yagami Light, y'know.

 

[rpotts]

tl;dr

Letting your opponent escape and techchasing is a better strategy sometimes

 

[Blue Yoshi]

Smash is so complex that is hardly matters what the best option is --- all that matters are the options that beat what you think your opponent will do (with the exception of nash equilibria). In other fighting games it's like: ok here's the best option, here the second best that counters its counter, and maybe you'll get a third "they'll-never-expect-this-one-but-it's-risky-and-I'm-probably-better-off-choosing-either-of-the-first-two-but-it's-so-unexpected-it's-practically-guaranteed." In smash it's like "I know they'll probably want to use option 1, because it covers options 2,3,4, but I use option 5 I'm gonna have even more options covered.. but If I use option 11 it covers everything... except option 15 ... ... but they probably won't do that.. ****"

This is why there are no 100-0 matchups in smash 64... unlike brawl... (lol sheik vs ganon + 100 other matchups). But yeah... this is basically how higher level smash 64 is played. You try to cover as many of your opponent's options as you can. Obveously the ideal situation is to be in a position where you cover all the opponent's options, but this is very unlikely to happen due to the many options available.

 

[dch111]

Letting your opponent escape and techchasing is a better strategy sometimes

Yeah, I'm not at the level to explore this option yet, but I sometimes wonder, and am sure they exist, situations where it's better to not continue the combo. After all, many situations have techchases so easy they might as well be classed as "pseudo-combos" since they are equally inescapable. Same with attacks that don't combo precisely (couple frames off), but anything the opponent does will not work to stop it from hitting.

 

[ballin4life]

Signia good post but the nash equilibrium of rock paper scissors isn't "every move", it's the strategy of doing each move with 1/3 probability

And you kinda make it sound like all symmetric games are just like rock paper scissors which isn't true.

Anyway, there might be an "optimal" strategy in smash that defeats all other strategies. But as you said smash is way too complex for anyone to be able to know for sure which strategy is optimal (especially given that the optimal strategy will probably involve a probability distribution, and humans cannot be truly random anyway).

 

[dch111]

If I ever decide to switch to computer science, one of my side projects will be to fix and improve the CPU AI. First, things like not attempting a third jump after getting hit or upB-ing off the edge will be fixed. Then I'll improve it bit by bit, adding things like a learning capability that relies partially on probability distribution of elements of past games (probably stored as 'books' outside the rom file).

I'll end up designing two CPUs: one called level 9+ which will be basically the fixed version with learning capabilities, DI within a 'reasonable' human capability, and moves that will be well timed but occasionally be a frame or two off. This is mostly so games with CPUs won't be as ********, and more challenging. The other one will be a "TAS" level bot that has maximum input and perfect timing for everything (along with the 9+ capabilities), used more for theoretical purposes. The "optimal play" we care about probably will come more from level 9+ or the top players in the world duking it out round robin from time to time though, since the other bot would be impossible to imitate.

From this I can see there are really two types of optimal strategies, the one possible by input into the game (which, as others mentioned, is already too complex to be found anytime soon, if it even exists in the form we expect it (it can end up as a stalemate, as one side always wins, as random (e.g. tornado decides), or as a guessing game (e.g. rps)), and the more practical, more organic one that is dependent on the capabilities of the best human players. And I'm hesitant to call that the "optimal" strategy rather than the "current best" strategy. The "optimal" smash strategy for humans can never be known conclusively as long as mankind exists (lol) since it is fundamentally dependent on the maximum physical capabilities of the players themselves. There will probably be a peak, though, which will be called "optimal".

 

[SSBPete]

but once we do find an optimal, people will find ways to counter it like fox getting to the other side of the map and spamming pew pew and never letting the optimal player get close.

 

[asianaussie]

If it is optimal, it won't be counterable...so you're just saying there is no optimal.

 

[blaze3927]

the whole idea of optimal is perception and time

ness dair again and again like a trampoline WAS optimal, then i discovered that yoyu could tech roll

Jigglies and luigis dair death combo WAS optimal, then people discovered DI

Ness WAS optimal......nuff said

 

[dch111]

There is always an optimal. Whether it results in a loss or a win for one side, a random outcome, or a stalemate, however, is a different matter.

I doubt we'll ever find the first kind of optimal (the one dependent on superhuman inputs), but the second kind, the "current best", will have a peak (it's always changing, like blaze noted). Will it ultimately be two foxes running away and spamming pew pew? Judging from the play of the top players in the world, I doubt it, but who knows what the future may bring. Stayed tuned on smashboards.com to find out...

Ness WAS optimal......nuff said

Antd's Ness TAS videos still haunt me to this day, lol

 

[Signia]

Signia good post but the nash equilibrium of rock paper scissors isn't "every move", it's the strategy of doing each move with 1/3 probability

And you kinda make it sound like all symmetric games are just like rock paper scissors which isn't true.

Anyway, there might be an "optimal" strategy in smash that defeats all other strategies. But as you said smash is way too complex for anyone to be able to know for sure which strategy is optimal (especially given that the optimal strategy will probably involve a probability distribution, and humans cannot be truly random anyway).

That's exactly right, my bad. To explain the probability distribution for others, it tells you how often you should your options. It involves choosing the better options more often than the counter-to-those-options-but-not-as-good options. "How much more often" is based on how much better the "better options" are. (correct me again if I'm wrong)

Oh yeah and technically machines are not completely random either, but they get "close" enough to fool us.

edit: lolz, typo

 

[Blue Yoshi]

Computers are not random. They use an algorithm that makes results seem random. Humans have human bias / human error that prevents true randomness. The only way to truly get a random result is by rolling a die or flipping a coin.

And Fox spamming lazers will not result in a win. Eventually Fox will have to attack the opponent. But even before that, if we are going by absolute 'optimal' play, fox spamming lazers can easily be avoided by an optimal evasive maneuver. Puff, for example, can jump around them (though hard to do, it can be done).

Well, while we're on the topic of random... funny story I heard:

There was a lottery somewhere that used random numbers from a random number generator on a computer. Eventually, someone noticed that the current number pattern of the lottery numbers (at the time) was identical to a number pattern done in the past (I think a few years earlier). He bought a ticket with the numbers that were the next day's result of the past one, and they ended up being the exact results of the next one. He had won the lottery. Years later, he noticed the exact same number pattern, bought the lottery ticket again, and won again. The lottery people realized what he had done and were refusing to give him the money. The lottery winner went to court, and the ruling was that he should get his winnings, and it was the lottery people's fault for being dumb enough to use a computer for a random number generator.

Ok... back to combos

 

[dch111]

Decent pseudorandom generators are sufficiently random. You cannot, for example, predict where the next tornado will be until it appears.

Do you have a link to that story? A repetition of the same pattern should not happen annually—that's just a sign of a poorly designed generator. I'm not sure how Nintendo made theirs, but if it can be affected by the controller inputs of either player (not hard to do), then that's definitely sufficiently random. Even my TAS level AI would not be able to predict the random #, since it would have to predict its opponent's inputs, and that prediction itself is based on a probability distribution (!).

I don't know if you were implying this, blue yoshi, but I never implied that laser spam would win either. I didn't exclude it from possible optimal plays though, in the strict sense, since I also assumed that stalemates were possible (if both players agree neither will approach). That said, it is doubtful in the first place anyhow that such infinite spam would constitute any kind of optimal play. Outside of theory, stalemates are useful when people insist on playing ****ty stages. For example, cloud camping on Yoshi's Island. Much better system than going for "who's more honorable" which is a silly system when both players ought to try not to lose. In any case, since the stages and plays that promote stalemated situations have been banned in competition, it would be ultra ultra rare to have people agree to a stalemate anyway, so by no means would you have to worry about it adding any "boring" in smash.

 

[ballin4life]

Rolling a die and flipping a coin aren't really 100% random either. The laws of physics say that the result is determined by the amount of force in throwing the die/coin. Also supposedly there are some people who can manipulate their dice rolls in craps to give them an edge.

I know that some random generators use outside things like system heat to get a random distribution.

If you know all the variables, it's determined, but if you don't then it's random from your point of view.

 

[firo]

Whoops, thought this was the other combo topic...

Not to extend this conversation mindlessly, but some interesting questions have been raised.

One of the things I wonder about is whether or not is whether "perfect combos" or series of moves does exist. Meaning, a series of moves that would result in death every time it is landed, regardless of the amount of DI the other character used. If a player had perfect reaction speed, would he/she be able to always react and punish DI/techs/moves, or will the skill of the second player always matter? This can raise such questions as, if I let an opponent tech out of a series of moves, but I can always predict the tech and can continue with a unique series of moves for each scenario, is it still a combo?

I think some of this came up in the recovery topic - would anyone be able to recover to the edge if the edgeguarding player had perfect prediction, reaction time, and always did the best move?

I think some of this can be seen with Yoshi - if the player had perfect parries every time, the opponent would be forced to either use quick moves like drills, or grab. An opponent like Samus would be virtually unable to inflict any damage. I would think that matchup is as close to a 100-0 (at optimal play) as we can get right now.

 

[dch111]

Yup. That about wraps it up. Back to the topic at hand...but it is really just about Pete's combos.

 

[blaze3927]

well on combos im putting together a collection of possibly the easiest death combos around:

jiggs 3 hit on pika
http://www.youtube.com/watch?v=sXyDH5RO8gw

dk two hit on jiggs
http://www.youtube.com/watch?v=cDm1AMXW65U

 

[asianaussie]

Second one isn't a combo...however, any aerial > grab + f-throw + jump Giant Punch is.

 

[SSBPete]

Whoops, thought this was the other combo topic...

I think some of this can be seen with Yoshi - if the player had perfect parries every time, the opponent would be forced to either use quick moves like drills, or grab. An opponent like Samus would be virtually unable to inflict any damage. I would think that matchup is as close to a 100-0 (at optimal play) as we can get right now.

i thought u could parry grabs (complete invincibility frame?)

 

[blaze3927]

Second one isn't a combo...however, any aerial > grab + f-throw + jump Giant Punch is.

i mentioned in the deiscripion it isnt a combo but it is impossible to escape if done right, jiggs is just too big of target for dk

 

[asianaussie]

Button mash out.

If it were impossible to escape, it could be considered a combo.

 

[dch111]

Whoops, thought this was the other combo topic...

Not to extend this conversation mindlessly, but some interesting questions have been raised.

One of the things I wonder about is whether or not is whether "perfect combos" or series of moves does exist. Meaning, a series of moves that would result in death every time it is landed, regardless of the amount of DI the other character used. If a player had perfect reaction speed, would he/she be able to always react and punish DI/techs/moves, or will the skill of the second player always matter? This can raise such questions as, if I let an opponent tech out of a series of moves, but I can always predict the tech and can continue with a unique series of moves for each scenario, is it still a combo?

Some of the grab combos against walls are perfect combos due to not being DI-able. The tornado and pow block can also guarantee next hits. As for perfect combos where DI is possible, I have no idea. I'm guessing that if they exist, they'd probably be in open air, where the comboer can follow the target's DI no matter which direction the target goes, and get the next move in before hitstun ends. I'm pretty sure there will be a couple such perfect combos discovered, if this is looked into more, but I doubt they'd be very long due to exponentially increasing possibilities with each new step. I'd bet right now the longest perfect combo with all DI-able hits would be max 3 moves. This is all just guessing though. Only thing I know for sure is that utilt x # combos won't be possible lol

The easiest techchases (the ones on small platforms) are close to a valid continuation of a combo in every sense but the technical sense (the definition involves hitstun), which disqualifies them. I would call combos involving their usage "pseudo-combos", though there's probably a wittier name for them. I don't know about the other techchase situations (that actually involve real chasing)...but it should only be considered inescapable if it doesn't matter which way the opponent rolls (that is, not dependent on preemptive prediction, but rather seeing the first movement the opponent makes), if he rolls, or if he doesn't tech at all. And such a situation where all options don't matter like that are very rare as far as I can see outside of small platforms.

 

[blaze3927]

http://www.youtube.com/watch?v=sXyDH5RO8gw

you cant DI a throw

but how much can you really change the direction of a jiggs uair? jiggs has all the control on the and to escape death theyd hav e to di the uair like in the opposite direction to which jiggs is hitting them, if they fall anywhwere on the left the rest will mean death

 

[Lawrencelot]

i thought u could parry grabs (complete invincibility frame?)

Yeah you can (or well, sheermaddness can)

 

[firo]

Yeah you can (or well, sheermaddness can)

In that case, I don't see Samus really being able to land any moves on Yoshi except a few fairs, which won't kill him. Anything else would be predictable and/or able to be parried.

 

[TANK64]

Button mash out.

If it were impossible to escape, it could be considered a combo.

That's what he's saying. If done right, you can't escape- so it's a combo.

After all, many situations have techchases so easy they might as well be classed as "pseudo-combos" since they are equally inescapable. Same with attacks that don't combo precisely (couple frames off), but anything the opponent does will not work to stop it from hitting.

exactly

 

[DMoogle]

In that case, I don't see Samus really being able to land any moves on Yoshi except a few fairs, which won't kill him. Anything else would be predictable and/or able to be parried.

Hmmm, I dunno. Down-B is probably too slow, but charged shots are just the right speed to allow for consecutive hits.

 

[firo]

Hmmm, I dunno. Down-B is probably too slow, but charged shots are just the right speed to allow for consecutive hits.

If Yoshi can parry the charged shots, I wouldn't think Samus would be able to get a hit in. Plus, I would think that shooting a charge shot would be predictable enough that the Yoshi could get out of parrying in the first place.

 

[DMoogle]

If Yoshi can parry the charged shots, I wouldn't think Samus would be able to get a hit in. Plus, I would think that shooting a charge shot would be predictable enough that the Yoshi could get out of parrying in the first place.

Of course he can parry the shots, but there's a delay between parries, right? So it would be just like a drill, only two hits (the shot, and a follow-up attack). It probably is too predictable though.

 

[firo]

Of course he can parry the shots, but there's a delay between parries, right? So it would be just like a drill, only two hits (the shot, and a follow-up attack). It probably is too predictable though.

There is enough lag after the charge shot that the Yoshi would be able to parry the shot and any move that Samus follows up with, if I am not mistaken.

 

[SSBPete]

Yeah you can (or well, sheermaddness can)

thanks cause i've definately done it a few times.


+ i'm pretty sure yoshi can parry 2 charge shots in a row. he could probly parry two fsmash attacks in a row from samus.


and back on the topic of perfect combos, i think yoshi and ness could use nair nair fair (or in ness' case dair) when they are close to the edge.

 

[Gammelnorsk]

could jab + low angled ftilt ever work lol (FOX)

LIKE AT CERTAIN % U KNO