Ivysaur vs Crouch Canceling (WIP)

[TreK]

Soooo... Crouch canceling is one of Ivysaur's problem
Let's see what we can do about it.

First things first, if you haven't already, read this, specifically the part where he explains the difference between Crouch Cancelling and ASDI down.
tl;dr, if you're actually crouching, it's CC. Else it's ASDI down. CC>ASDI down. So basically, as long as you hit the opponent during another move, they don't have access to CC, the worst they can do is ASDI down.

Except that's not what happens in practice. Ivysaur has an unusually large amount of multihit moves, some of which allow the victim to crouch cancel after getting hit. We'll get to that in a minute.

There are three possible outcomes when somebody is crouch cancelling, or ASDIDing a move : either they simply flinch, and are able to counterattack extremely fast, or they are knocked down, and forced into a tech chase situation (which is not beneficial to us, as Ivysaur's movement speed and very high startup on average makes her a sub par tech chaser), or they are actually knocked back. If they are knocked back while crouch canceling, they will receive reduced knockback.

There are a couple factors which affect CC and ASDID. Among them, whether you are standing on a flat, a downwards slope or an upwards slope, whether your move is fresh or stale, its knockback and its angle.

But the two major ones, those that you have to care about, are the opponent's weight, and %. To give you an idea, Ivy's fair forces a tech chase on CCing Sheik at 79%, on CCing Bowser (heaviest character in the game, weight 114) at 93%, and on CCing Puff (lightest character in the game, weight 60) at 65%. It's linear, so there is a way to calculate exactly by how much you need to multiply the values I'm going to give you below, but I have not calculated it yet.

Tests were done on Sheik, who has the average weight for a PM character (90, the average being 91). I can confirm that falling speed is not a factor, I have the thresholds with Zelda tested for several moves. Zelda has the same weight as Sheik, but has a different falling speed. The thresholds were the same every time. You can find the complete list of the weights of all the characters in PM here.

Generally speaking, a higher threshold means you want to crouch cancel/ASDI down this move, a lower threshold means this move is good against CC/ASDI down.

So let's get to the tedious part now. This is a wip, doing the first half of Ivy's moveset took me about three hours of discontinuous work. If you want to contribute, here's how to test this properly. Go to training mode, so that your moves don't stale. Pick Ivysaur and pick Sheik for the cpu. Activate controlled CPU in the AI options. To test CC thresholds, hold down as Sheik. To test ASDI down thersholds, taunt as sheik, and during the taunt, hold down as sheik. It might be a little trickier with some moves (ex bair1/bair2), be a bit creative with your setups and use a variety of stages. Just make sure you don't use stages with slopes.

There are four columns, 2 for CC and 2 for ASDI down. flinch threshold means, after this percentage, Sheik does not flinch anymore, she suffers a knockdown and is forced into a tech chase situation. Knockdown threshold means, after this percentage, Sheik is not knocked down, instead, she suffers knockback.

Move|Crouch cancel flinch threshold|Crouch cancel knockdown threshold|ASDI down flinch threshold|ASDI down knockdown threshold|Commentary
jab1|N/A|N/A|N/A|N/A|jab1 has constant knockback, and will always cause a flinch on CC and ASDI down. CC gives more frame advantage.
jab2|93%|>200%|N/A|0%|on ASDI down, jab2 always produces knockback, meaning if the victim doesn't CC the first hit, the second is safe on hit
ftilt|51%|149%|51%|149%|**** this move you can actually CC it after getting hit by it why does it even exist
dtilt|123%|>200%|46%|172%|
dtilt tipper|32%|41%|17%|36%|
utilt|72%|100%|3%|15%|this move is exceptionally good against ASDI down compared to CC
dash attack sexkick|26%|59%|N/A|0%|always produces knockback on ASDI
dash attack meaty|140%|192%|140%|192%|the meaty is extremely susceptible to both CC and ASDI down.
fsmash (uncharged)|52%|131%|20%|72%|
dsmash (uncharged)|76%|>200%|23%|>200%|
usmash (uncharged)|39%|61%|0%|15%|
usmash sweetspot (uncharged)|2%|8%|N/A|0%|As expected, this is Ivysaur's best move against crouch cancel and ASDI down
nair last hitbox|27%|105%|27%|118%|you can CC in between the hits.
nair landing hitbox|118%|142%|118%|142%|you can CC in between the hits, but the landing hitbow is weaker against CC than the last hitbox
fair|121%|>200%|56%|>200%|
fair tipper|80%|104%|7%|26%|good against ASDID, bad against CC
bair1|N/A|N/A|N/A|N/A|constant knockback. will always flinch on CC/ASDID.
bair2 close hit|62%|185%|25%|87%|if they CC/ASDID'd the first hit they will CC the second hit
bair2 far hit|94%|>200%|41%|>200%|^
dair|78%|156%|29%|68%|
dair sweetspot|34%|78%|0%|22%|
uair|61%|79%|19%|32%|
uair sweetspot|26%|36%|N/A|0%|
uair belly stomp|165%|>200%|81%|132%|It's actually detrimental to CC/ASDID this move below the flinch threshold, if they do, they receive increased hitstun.

The things we can take away from this data so far :
-nair sucks against CC and ASDI down, in case you didn't already know. Opponents can actually CC the nair on hit, on reaction, and avoid most followups under 120%. However, if they do CC it, you can not-fastfall on purpose and let all hits connect, which will force a tech chase situation. It's better than the alternative, even though it's not exactly good.
-both dash attack and usmash are great moves against CC, making the gatling combo a prime answer to CC.
-dtilt tipper also wrecks CC, and even if they don't CC it it's still a great move. Conclusion : spam the hell out of it.
-Ivy's jab combo is surprisingly good : most jabs you can react to them and ASDI down after getting hit by the first jab then punish after the last jab. Ivysaur's jab2 always knocks back on ASDI down, which is a great thing to have. You can't afford to wait between jabs though, else the opponent will be able to CC and punish the jab2.
-utilt is good at punishing, but should not be used as an opener because the risk involved with being CC'd is too great.
-ftilt sucks for yet another reason.
-more to come !

Enjoy !

 

[Swann]

SCIENCE AHOY

Good **** @ TreK you are the man

The only issue with gatling is how situational it is. Literally the only consistent situation is when you're on the ledge and they are just crouching there waiting to CC you for a quick kill. Waveland->gatling is super useful.

Other than that, though, it's so telegraphed I have a hard time ever landing it. I will need to experiment more.

Dtilt is pretty much our best ground move so no surprise there.

I've been counterhit in the time between jab1 and jab2. Not sure if I'm just slow or if it actually sucks sometimes.

 

[TreK]

Yeah the reason Ivy's jab is better than most jabs in this game are is that you can't CC/ASDID it on reaction, as in, between the two hits. But it's still possible to CC/ASDID on prediction and punish between the two hits. I don't think any jab besides Zard's is imune to preemptive CC tbh.
Ivy's jab is not good against CC, it just doesn't get destroyed by it like every other jab in the game, basically.

One thing I tried with the gatling combo is to do it after a nair : the nair makes people want to CC, but if they do, and I gatling combo them, it's going to be a bad day for them. We're talking a 40% in one second+half a solar beam charge+a forced tech chase kind of bad day here.
Buuut they can shield it. So basically, what I do on characters with no fast moves is that after a nair, I make a mixup between a sh footstool to beat their shield, or a gatling combo to beat their counterattack. It's kind of a 50/50 except if they attack and I choose to footstool I'll just get out of the way safely, and if they shield and I gatling, it will do a lot of shield damage, making the next attack more likely to get me in.
I really have to go check in frame advance how much time I have after a CC'd nair, I feel like this is going to be a very important thing to know about. My region is dominated by a Charizard main, but if I'm going to travel and fight lots of faster characters, I probably won't be able to rely on this trick. With a bit of luck that is going to be weight dependant.

In any case, if tomorrow I don't get to play with my friends, I'll try to finish this data and figure out the formula which will allow us to convert the thresholds for other weights automatically.

SCIENCE !

 

[TreK]

Okay, I've completed the table above minus the special moves (to keep it short : synthesis and sleep powder are not affected, razor leaf sucks against CC/ASDID, and seed bomb, vine whip normal/tipper, and solar beam are all great against it). But let's bump this thread with more science.

I'm going to go through everything I've done so that if you think my reasoning was erroneous, we can correct it.

I have measured the effect of weight on Crouch Cancelling and ASDI down. The way I've done it is, I have measured the thresholds for a variety of moves with different knockback and angles on crouch cancelling Jigglypuffs and Bowsers, respectively the lightest and heaviest character in the game.
Each time, no matter what move I used, you can see that you can draw a line between those points, that goes through the point corresponding to Sheik's threshold. Mathematically, this means that the threshold is linear with the victim's weight, which is extremely important information :

If you want to compute this data yourself,
Jigglypuff : fair 65%, fsmash 41%, dash attack 20%
Bowser : fair 93%, fsmash 62%, dash attack 32%
Sheik : see OP

However, if you look closely, the slope is not EXACTLY the same every time.
For a threshold of 26, we get a derivative of about 0.21
For a threshold of 52, we get a derivative of about 0.36
For a threshold of 79, we get a derivative of about 0.48
As it turns out, this is also linear. Meaning, the higher the threshold on Sheik is, the steeper the slope. We'll have to take that into account too.

The ajusted line's equation here is y=0.0055x + 0.08

But then, on top of calculating the expression of the slope, we also have to calculate the offset at zero, in function of the threshold on Sheik. This is pretty easy : we just have to follow the three lines until they meet with the y axis, and not the coordinates of the intersections.

And then, we draw our final graph with thresholds on sheik in the x axis, and the intersection in the Y axis :

The equation for this guy is y = 0.5x - 7.2

SO.

The function will take this form :

T : threshold for a given character and a given move
w : weight of the victim
x : threshold value for Sheik, which you can find in the table in the OP.
A : function describing T's slope depending on x
B : function describing T's offset at zero depending on x
T(w, x) = A(x)*w + B(x)
T(w, x) = 0.0055x*w + 0.08w + 0.5x -7.2

We have a result, don't we ?
But an equation is only as good as what it can predict.
So let's check its efficiency :

Ivysaur's fair tipper on Bowser knocks him down at 93%.
It knocks down Sheik at 79%, and Bowser's weight is 118.
T(118, 79) = 93.011
This is extremely promising.

Ivysaur's fair tipper on Jigglypuff knocks him down at 65%.
It knocks down Sheik at 79% and Jigglypuff's weight is 60.
T(60, 79) = 63.17
Okay that's slightly less accurate but it's still a good indication.

Let's try it with a random character and a random move.
Ivysaur's usmash knocks Mario down at 42%
It knocks Sheik down at 39% and Mario's weight is 100.
T(100, 39) = 41.75
Alright, at this point I think we can conclude that this formula is accurate.

But, all the tests so far have been done on crouch cancel, and for the flinch threshold.
Let's first check if it still works for the knockdown threshold :

Ivysaur's utilt knocks Squirtle up at 77%
It knocks Sheik up at 100% and Squirtle's weight is 82
T(82, 100) = 94.46

Okay here comes the bad news. We're going to have to find another formula for the knockdown threshold. Let's check if we'll need a third and fourth one for ASDID.

Ivysaur's dsmash knocks down an ASDIDing Sonic at 22%
It knocks down an ASDIDing Sheik at 23% and Sonic's weight is 82
T(82, 23) = 21.223
This is good, let's double check.

Ivysaur's bair2 tipper knocks an ASDIDing Charizard down at 46%
It knocks down an ASDIDing Sheik at 41% and Charizard's weight is 110
T(110, 41) = 46.905

Alright ! The formula we've found actually works on ASDI down as well, not just crouch cancelling. This means we will only have to find two formulas, one for the flinch threshold, and one for the knockdown threshold.

However, I think I'm going to stop here for today. I will work out the knockdown threshold formula later.

 

[TreK]

Quick update before I go back to class

T : threshold for a given character and a given move
w : weight of the victim
x : threshold value for Sheik, which you can find in the table in the OP.
A : function describing T's slope depending on x. A(x) = 0.0064x + 0.0594
B : function describing T's offset at zero depending on x. B(x) = 0.52x - 8.11
T2(w, x) = A(x)*w + B(x)
T2(w, x) = 0.0064x*w + 0.0594w + 0.52x -8.11

Let's see if it's any good :

usmash sourspot knocks up a crouch cancelling Peach at 59%
it knocks up Sheik at 61%, and Peach weighs 90 (like sheik, wtf)
T2(90, 61) = 64%

dash attack sexkick knocks a CCing samus up at 62%
it knocks Sheik up at 59%, and Samus weighs 110
T2(110, 59) = 70%

fair tipper knocks jiggs up at 85%
it knocks sheik up at 104, and jiggs weighs 60
T2(60, 104) = 89%

dair sweetspot knocks bowser up at 99
it knocks sheik up at 78 and bowser weighs 118
T2(118, 78) = 98%

Dair sweetspot knocks an ASDIDing Roy up at 27%
It knocks an ASDIDing Sheik up at 22 and Roy weighs 85
T2(85, 22) = 20%

I am not pleased with those results, an 8% difference is very important imo, but at the same time, I've yet to find any error bigger than that.

See, the thing here is that I believe the ASDI down threshold formula is not linear, but logarithmic :

The dots form an arc rather than a line. I'll need another graphing tool to do logarithmic approximations, though. It's going to take a while to refine this formula. For the moment, you can consider that I have both equations, but the second one is much less accurate than the first one.

Also, the test results I've made today make little to no sense if I'm running on the assumptions I've made, which are that for a given move, only weight affects the thresholds : Samus' threshold is extremely low for a character that is much heavier than average, and Roy's threshold is actually higher than Sheik's despite being lighter.
I'll have to look into this as well. Maybe falling speed does matter after all.

Another reason I wanted to update this thread is that I wanted to make an announcement :

When I have worked out all the details and am pleased enough with my work, I will release an Android app that you will all be able to use before a tourney match to quickly get an idea of when your moves are vulnerable to CC, and when they aren't.
This is basically the end goal of this little project of mine.

The app is already 70% done, but as you can see I have a bit of work to do before I can release it