Metagames are hard to solve

“Torture the data, and it will confess to anything.” –Ronald Coase, winner of the Nobel Prize in Economics

D.A.T.A.S.T.R.E.A.M. #01 began with the idea that we were trying to see if we could find a solution to how strong the Super Armor and Slow Super Armor secondary Spirits seem to be in the early Spirits meta. However, as the above quote demonstrates, we have to be careful with how we analyze the data as, how we cut it up can paint very different pictures.

Come with me on a journey as we look at the results of D.A.T.A.S.T.R.E.A.M. #01 and see what we can learn. The data is, indeed, instructive, but it is far from definitive.

First, before we go too far, let me say that there are going to be lots of factoids in this post. For the sake of brevity, let’s define some terms I will be using:

  • SA: Super Armor
  • SSA: Slow Super Armor
  • SA+: Super Armor or Slow Super Armor
  • AntiSA: Any Amiibo trained for the purpose of defeating Super Armor or Slow Super Armor Amiibos.

The other thing you must be made aware of is how DATASTREAM #01 was composed. All 28 Amiibo were thrust into groups of four. All four Amiibos in a group would face off against one another. The top two Amiibos in each group would advance to the playoffs.

In the playoffs, two Amiibos would face off in a best of 3 series. Each Amiibo would have 2 stocks per match. This is the first flaw in this experimental tournament. Two lives allows for a lot of variance. But we’ll discuss that later. Let’s get to the results.

The first point of data is that a SA Bowser  won the tournament. His name was TurboSatan (which is a name that makes me giggle nearly every time I said it. What’s worse than Satan? TURBOSATAN!) and he was trained by KingPingu. Congrats to him for winning! But now someone with less experience with data, someone like Johnny Newbie, might hear that factoid, overreact, and say something like:

“See! Even with people sending out Amiibo trained to defeat Super Armor a Super Armor Amiibo won! We must ban Super Armor and Slow Super Armor forever!”

Well, indeed TurboSatan played very well, but his victory is a bit hollow. A Ridley named REEEEEEEEE, trained by Supernova, was his finals opponent. The Ridley fought VERY well against SA and SSA Amiibos, but in the finals he used Space Pirate Rush right on the edge of the stage twice and killed himself with the manuevuer. Outside of those two self-destructs, REEEEEEEEE outplayed TurboSatan most of the match and likely would have won if not for that fluke.

“Oh! Well forgive me for jumping to conclusions! I didn’t realize it was a fluke! I didn’t watch the matches live on twitch.tv/splicestream when you hosted the tournament and didn’t know how the finals went. So I guess AntiSA Amiibos are viable! Looks like we don’t need to ban SA or SSA after all!”

Woah, there Johnny Newbie. You’re sort of swinging the opposite direction fast. Also, thank you for oddly shoehorning an advertisement for my stream in the middle of your sudden opinion change. Regardless, the situation is a bit murkier than that. Let’s go into the hard data itself to figure out just what went down in DATASTREAM #01

The Group Stage

First, let’s discuss how we want to break down the data. Obviously, matches between two AntiSA Amiibos isn’t that useful to our conversation as we only care about how Super Armor can be countered. For that same reason, we can ignore matches between SA+ Amiibos with the exception of matches between Bowsers armed with SSA or SA and SA+ Amiibos. The reason why this is significant will be clear once we look at the data.

SA+ vs. AntiSA
SA Amiibos had a record of 9-9 against AntiSA Amiibos.
SA Amiibos recorded 27 kills compared to 25 deaths in those matches.

SA+ Bowsers vs. AntiSA
SA+ Bowsers recorded a record of 10-4 against AntiSA Amiibos
SA+ Bowsers recorded 24 kills compared to 14 deaths in those matches.

SA+ Bowsers vs. All other SA+ Amiibo
SSA & SA Bowsers went undefeated against all other SA and SSA Amiibos, recording a record of 7-0
In those seven matches, the Bowsers had a whopping 14 kills compared to only 5 deaths.

So, AntiSA Amiibos held their own against SA+ Amiibos, though they did score slightly fewer kills. However, that even record disappears once the SA+ Bowsers are factored in. The SA+ Bowsers more than doubled up the AntiSA Amiibos.

But the surprising data to me was how much the SA+ Bowsers dominated the other SA+ Amiibos. They didn’t lose a single match? And they had nearly three kills for every death.

“Well that seals it! Super Armor and Slow Super Armor is broken and needs to be banned forever! Especially on Bowser! Too stronk! Pls nerf!”

Well Johnny Newbie, I can see why you’d come to that conclusion. However, something to remember is this was the Group Stage. The Group Stage includes everyone’s Amiibos, even the poorly conceived Amiibos. Remember, this tournament features the first large scale, concerted effort by the community to determine what is a good AntiSA Amiibo. Not every concept would work out. Besides, even as strong as SSA Bowser was, not every SSA Bowseradvanced to the playoffs (though Arklaine’s Slow Super Armor Bowser also equipped with Impact Run was hilarious to watch!).

Because the Group Stage includes everyone, including the poor concepts, let’s hold off on making any grand pronouncments until we look at the Bracket stage where only the best Amiibos had advanced.

The Bracket Stage

As the Bracket stage began, there were 9 SA+ Amiibos and 7 AntiSA Amiibos. I purposefully segregated the matches as much as I could so that we’d get SA+ vs AntiSA matches.

Also, important to note, that at this stage of the stats, I seperate TurboSatan from the other SA+ Bowsers. Why? Because every other SA+ Bowser was equipped only with SSA. TurboSatan was the only SA Bowser. I think that matters going forward.

I think it’s also worth noting that of the 9 Amiibos equipped with SSA or SA, all but 3 of them were Bowsers. The outliers? Blue’s Slow Super Armor Dedede (Escargoon), Zappy’s Slow Super Armor Cloud (Mugo), and MegaVGMaster’s Slow Super Armor Lucario (Blade). But let’s see the break down.

SA+ vs. AntiSA
SA Amiibos had a record of 2-6 against AntiSA Amiibos in the Bracket stage.
SA Amiibos recorded 10 kills compared to 13 deaths in those matches.

SSA Bowsers vs. AntiSA
SSA Bowsers recorded a record of 4-8 against AntiSA Amiibos in the Bracket stage.
SSA Bowsers recorded 14 kills compared to 20 deaths in those matches.

TurboSatan vs AntiSA
TurboSatan went 8-3 against AntiSA Amiibos in the Bracket stage.
In those eleven matches, TurboSatan recorded 19 kills to 12 deaths.

Now THAT is interesting. Suddenly, in the Bracket Stage of DATASTREAM #01, the script got flipped! It was the AntiSA’s beating up against the SA+ and the SSA Bowsers. The only SA+ Amiibo that posted a winning record against the strongest AntiSA Amiibos was the eventual champion, the lone SA Bowser, TurboSatan.

But that’s strange right? AntiSA was treading water against regular SA+ Amiibo and getting mollywhomped by the SA+ Bowsers in the Group Stage. Why the flip?

I think it’s safe to say that the AntiSA Amiibo that made it to the final 8 found a formula that worked. If you analyze the bracket itself, you’ll see that the only SA+ Amiibo to even make it to the final 4 was TurboSatan. Let’s take a closer look at those four AntiSA Amiibo that made it work.

The Sniper
Blue’s Mii Gunner- Oliver

It's times like this you gotta ask yourself one question: do I feel lucky? Well? Do ya, punk?
His mustache gives him power

Mii Gunner has proven to be a very strong Amiibo. It is the first Amiibo we’ve ever found that makes projectile spam work. Oliver in particular makes heavy use of his missiles to keep SSA Amiibos far away from him. By the time they finally get close, they have taken enough damage that they now get pushed back by the projectiles damage.

The secondary Spirits Blue chose all boosted the power of those missles. That makes sense as, clearly Oliver needs to rack up the damage fast to eat through the Super Armor before the Slow Super Armor Amiibo can touch him. In theory, a similar strategy could be done with another Amiibo that has a strong, spammable projectile game like Samus, Duck Hunt, or Mega Man (though what secondary Spirits you choose will change depending on who you select).

However, the kryptonite factor of this style of Amiibo showed itself when Oliver duked it out against TurboSatan. Since TurboSatan had regular Super Armor, he could close the distance on Oliver before the projectiles could deal enough damage. Oliver was forced to go on the defensive and fight in closer quarters than he’d like. Still though, it proved to be effective and a viable option against Slow Super Armor.

That’s not to say this style couldn’t be tweeked to defeat regular Super Armor. Perhaps if it was trained to grab a little more in close quarters? We’ll have to see if the formula can be fine tuned a bit.

The Kitchen Sink
MegaVGMaster’s Mega Man, Bass

Dear Strong Man, how do you punch with cannons on your hands. Crapfully crapfully, SmashLover13
Mega Man shows you the direction you will fly when he punches you.

I almost considered calling this archetype “The Surgeon” but that makes it sound too precise for what I feel the strength of this build is. The Kitchen Sink has a lot of versatility and throws that versatility in the face of their opponent. Let’s break this down.

Mega Man is an Amiibo many have felt had a place in the metagame but hasn’t received a lot of attention. Appropriately, MegaVGMaster finally gave a little love to the Blue Bomber and came out with a strong Amiibo.

The game plan of this Amiibo against Super Armor is a 3 step process:

  1.  Plink away from a distance with a variety of projectiles. (unlike the Sniper, these attacks tend to have very little knockback, even at high damage)
  2.  Use Throws when the  opponent is close to regain distance. (Use of the Strong Throw support Spirit is basically a must for this gameplan)
  3. Use a quick attack with lots of vertical knockback to get the kill once they are at high damage. These attacks will usually be up-tilts. (And definitely utilize the Toss & Meteor Spirit along with another Spirit that boosts your up-tilt. In the case of Mega Man, that’s the Fist Attack booster for the Mega Upper)

Mega Man works well for this game plan because he has a large variety of projectiles he can use to take small chunks of damage away from an opponent while they close the distance. Then, by training him to grab in close quarters, you negate one of the SA+ Amiibo advantages because SA+ Amiibo tend to stand there, wait to unleash a devestating charged up smash attack when you get close.

The Kitchen Sink however can ignore the long charge up attacks by either continuing to launch ranged attacks where they are safe or, since they are trained to grab, can close the distance and grab the opponent instead. Watching Bass fight against a SSA Bowser showed that as the match went on, Bowser had to stop charging up the smash attacks or it would never get them to connect. A SA+ Amiibo that can’t charge up it’s smash attacks has lost a ton of what gives them an advantage. Bass in particular started throwing so much at his opponents that he was comboing them rather masterfully.

To utilize this archetype, you’d need a character that has a healthy amount range attacks, a throw that may not get kills but definitely gets distance, and a tilt attack that comes out quick can get vertical kills. Therefore, other Amiibos that may be able to utilize this archetype are ones like Link, Palutena, or Mii Gunner. All three have a variety of projectiles to whittle away damage with and have up tilts that can get vertical kills.

As I think about it, Mii Gunner might be able to combine the Sniper and Kitchen Sink motif together to form a VERY strong AntiSA combo. Regardless, I think this archetype is interesting and would be fun to train.

The Glass Cannon
KingPingu’s Little Mac, C-Stick

ONE-PUNNNNNNNNNNNCH
Move like a butterfly, sting like a freight train

The Sniper tries to keep his distance at all times. The Kitchen Sink uses a variety of small attacks to plink away at the health meter until they hit with a big finisher. The Glass Cannon however, has a very different plan of attack that is best summed up by Iron Man:

There's only one God, ma'am. And I'm pretty sure he doesn't dress like that.

All gifs aside, look at that attack/defense breakdown on C-Stick! KingPingu knew what he wanted. To go into the negatives on any stat you have to overload one stat with Primary Spirits, then apply your secondary Spirits at the end! You can’t accidentally get negative stats. You have to intend to do it from the beginning (special thanks to Supernova for clearing up how this was achieved).

To say C-Stick is a Glass Cannon is an understatement. Stiff breezes  and harsh words can knock this Amiibo down. But the amount of damage he puts out is incredible and Super Armor Amiibo just aren’t ready for it.

Now, this isn’t to say C-Stick cleaned their clocks. Nearly all his matches came down to both Amiibos on their last life. How could they not be with a defense that’s negative? You’d need to get a little lucky when every smash attack that hits you is doing north of 50 damage. How close were C-Stick’s matches in DATASTREAM 01? Well, he likely wouldn’t have even made it out of his Round 1 fight against a SSA Bowser if the Bowser hadn’t gotten greedy.

But that’s what you get with an Amiibo like this. Highs and lows. If you’re always swinging for the fences you’re going to hit a few home runs. An Amiibo like this always has a puncher’s chance (get it?) because they can turn a match around with quick kills. However it can just as easily turn on them too. Don’t try this if what you want is consistency.

Despite nearly losing in Round 1, C-Stick made it to the final four. While it’s hard to dominate with this, you can trust that Amiibos that like to wait a long time to charge up smash attack won’t be happy if they wiff on one of their big Smash attacks and die 8 seconds into a match!

I’m not sure what other Amiibos could duplicate this style. I wonder if Mac is the only one who has the combination of quick and strong smash attacks to make something this lopsided work. Maybe Ganondorf or Ike since their swords make up for their lack of speed by giving them smash attacks with lots of range? Not sure. Needs more testing. And this style of testing should be fun because when you’re all in on damage, win or lose, fireworks are sure to fly!

The Grappler
Supernova’s Ridley, REEEEEEEEE

We all theorycrafted that someone with a good command grab would be able to beat SA+ Amiibo. But Supernova had it all come together with REEEEEEEEE.

The use of Armor Knight is probably the single best counter to Super Armor as is. For those not familiar, Armor Knight greatly increases your defense, slightly increases your attack, but also greatly reduces your speed. If your response to that is to say, “But wait, that sounds exactly like Slow Super Armor,” then let me clear up the misconception. Slow Super Armor does indeed slow you down, but it doesn’t increase your defense or attack. Super Armor increases your flinch resistance, not your defense. You’re still taking the same AMOUNT of damage from an attack, but it’s just not interupting your ability to attack.

The neat thing about Armor Knight is that by combining it with a Move Speed up Spirit, you wipe out a lot of the downside to give yourself an Amiibo that can go toe-to-toe with most SA+ Amiibo.

In case you’re wondering how to get Armor Knight… I have some bad news. It comes from only one single Spirit. The Halberd Spirit. When it comes to our hopes to fight the Super Armor menace, that’s not too reassuring.

Regardless, putting it on an Amiibo with a great command grab like Ridley made for a great contender against any SA+ based Amiibo. Again, in the finals he would have likely one if not for two ill-advised Side-B’s just barely off the edge of the stage that killed REEEEEEEEE at low percentages. If I had set the stock count to 3 per match instead of 2, REEEEEEEEE likely is the winner of DATASTREAM 01.

The biggest problem though (outside of the rarity of the Halberd spirit that grants Armor Knight) is that there aren’t a lot of Amiibos with good command grabs out there. Oh, there are a lot of command grabs that are great at interrupting but they don’t tend to give you a lot in the way of range options. The three best are Ridley’s (charge forward), Isabelle’s Fishing line (no one has experimented too much with her from what I’ve seen), and Incineroar’s Alolan Whip (Amiibo hasn’t been announced yet).

There might be an Amiibo with just a good grab game that could use a similar game plan but at the moment I’d wager once people actually start tinkering with Isabelle, we might see some interesting results.

So, what does it all mean?

We started this whole thing with a discussion about applying the Scientific Method to our Amiibo analysis and training. One of the most important things in any scientific endeavor is to figure out what your confounding factors are. It’s also important to remember that more than one thing can be true at the same time. So let’s take a look at everything and determine what we know, what we don’t know, what we think we know, and what we still want to find out.

SA+ Bowsers outclass all other SA+ Amiibos

Look. You can explain away SA+ Amiibos going 0-7 against SA+ Bowsers as possibly a statistical fluke if they were all close matches. If you flip a coin seven times, there’s a non-insignificant chance they all come up heads. But 5 kills to the Bowsers’ 14? That can’t be explained explained away.

If Super Armor or Slow Super Armor is going to be allowed in a healthy metagame, there can’t be one Amiibo that just outclasses all the others in that category. You just can’t play Rock Paper Scissors if there is also Super Rock that still defeats Scissors but also crushes regular Rock.

It’s a shame he outclasses the others so hardcore because I think there are some interesting Super Armor and Slow Super armor targets. MegaVGMaster entered a Lucario with Slow Super Armor that got into the Bracket stage (though it lost to Oliver). Lucario is an interesting choice that I personally would like to see explored more. But it can’t really be explored because as long as SA+ Bowsers are out there, it gets thrashed.

So, in my opinion, for the time being, tournaments should at least ban SSA or SA Bowser because it stifles experimentation with other kinds of SA+ Amiibo. 

There are definitely some ways to counter the Super Armor meta but it might not be enough to justify letting Super Armor in

While the Group stage showed that Super Armor is potent, there were some Amiibo that proved it could be consistantly beaten. Now, HOW consistently it can win is a topic up for discussion. If we looked at just the 4 Amiibo I highlighted (which is a highly biased sample size that only looks at the best AntiSA had to offer) their records were 22-9 against against SA+ Amiibos.  That’s a 71% winning percentage! That looks REALLY good!

However, if you dig down deeper, a different story gets told. In those matches the AntiSA Amiibo managed to get 52 kills and 40 deaths. That’s nowhere near the 71% winning percentage. We saw that SA+ Bowsers dominated the non-Bower SA+ Amiibos by nearly a 3-to-1 margin. While we wouldn’t expect such a great disparity for 22-9 but you’d at least like to see something closer to 2-to-1 on the kills to deaths there.

This means that even with the BEST Amiibos at fighting SA+ Amiibos, the SA+ Amiibos could keep it close. They very rarely lost two lives to none. If these Amiibo truly DOMINATED SA+ Amiibo, then they should have way more shutouts. But they didn’t. It may be the case that Super Armor as a mechanic prevents Super Armor Amiibos from ever being truly dominated. It’s important to find out that Super Armor can not only lose but lose big, just like every other Amiibo fighting style.

Now, given that these were only two stock matches, that’s not exactly the best sample size. If we had more lives, we might see the numbers level out to something we’d like to expect more. But it shows that even if we’re countering SA+, just putting SA+ on an Amiibo gives it a chance to win. To go back to the Rock-Paper-Scissors analogy, the game doesn’t work if it turns out Paper loses to rock almost 1/3 of the time.

We still don’t know if there is a competitive difference between Super Armor and Slow Super Armor, but I suspect their is

We only had four Amiibo get entered into DATASTREAM 01 with Super Armor. All other SA+ Amiibo were Slow Super Armor. I’d like to say the fact that the only SA Amiibo to advance to the Playoff Round was TurboSatan proves that SSA is at least equal to SA if not superior, but I believe SA’s poor showing is due to, sadly, the fact that the SA Amiibo weren’t trained as well.

On the whole, we haven’t experimented as much with Super Armor because, well, it’s so much harder to get as a Spirit. However, if we’re going to be making definitive statements about these mechanics, we need to do more thorough testing on SSA vs SA Amiibos and how AntiSA Amiibos do against them.

Super Armor punishes the newer players the most

If we get away from the cherry picked data of only the best AntiSA Amiibo, and go back to just the Group Stage, we see that AntiSA Amiibo went 9-9 against SA+ Amiibo (excluding the Bowser data as I’ve already said I think we need to ban him from having Super Armor).

While there are a lot of theories in gaming circles on what constitutes a “healthy metagame”, I think most of us can agree that part of a healthy metagame probably includes that on some level it’s accessible to newer players.

It’s clear that Super Armor has a high floor for newer players to clear. The fun a rookie Amiibo trainer had training their creative Instadrop Villager quickly gets erased when it runs into a brick wall that ignores all their attacks. We want to grow the metagame, and it’s clear to me that seeing as we are still not even two months into SSBU, that Super Armor and Slow Super Armor are not healthy for the young meta. There may be a time in the future where we can feel comfortable letting Slow Super Armor or even Super Armor in, but right now isn’t the time.

I think we were a bit hasty to dub SSA and SA a cancer and ban it. Especially because, if we are honest with ourselves, we hadn’t thoroghly tested it as a community. Very few people spent real time trying to counter SA+ Amiibos. I had a few people say, “So-and-so trained this one Amiibo to fight Super Armor and it still lost!” but that’s not really a thorough test. We need more data from a large variety of AntiSA Amiibo trained by a variety of trainers against a variety of SA+ Amiibo before we decide that it’s broken forever.

Even DATASTREAM 01 fell short of that goal but it got us closer than ever. We found archetypes that worked that we can iterate upon and see if we can apply the lessons to a larger variety. But, for now, I think we should leave the experimenting to tournaments DESIGNED to experiment this issue so people are free to play with a wider variety of Amiibo styles while we wait for the Super Armor Question to get answered.

Banning Super Armor to just heavyweights is NOT the answer

I’ve seen this suggested by some people and this is the one thing I feel the strongest about that is the WRONG answer. Why? Well, let’s ask some questions. Other than Bowser, the concensus amongst most of the trainers I’ve talked to, Super Armor and Slow Super Armor has been most problematic on Cloud, Dedede, Ganondorf, and Ike.

Okay, now look at that list and compare it to the actual weight values of characters in the game. Oh, it’s true that the 3 of the 5 problem Amibo are in the top 5 of weight… but Ike is the 12th lightest. And Cloud, the Amiibo with the second best record in tournaments that allow for Super Armor (outside of Bowser), has a weight of 100. That’s tied for 26th on the weight list, tied with Mii Sword Fighter. Heck, he’s two lighter than Mega Man and no one is complaining about him with Super Armor.

I say this to show that the problem with Super Armor has little to do with weight and much more to do with which Amiibo benefit the most from long, wind-up attacks that suddenly get to connect because Super Armor allows them to resist flinching. The fact that heavy characters have more big, heavy-hitting, wind up attacks merely coincides with their weight.

We are just scratching the surface of AntiSSA Amiibo tactics

Before this tournament we thought SSA was a free ticket to the top 8 in any tournament. Seeing as DATASTREAM was half SA+ and half AntiSA, we proved that it’s not quite true. We tried a lot of different theories on how to counter SA+ Amiibo and now that we’ve done concerted effort in this direction, I feel that we can discover more.

I want to take the lessons we’ve learned and pour them into DATASTREAM 02. I don’t have all the details yet but it will be a much more focused tournament with a broader approach to avoid some of the problems DATASTREAM 01 encountered. For 1, we’ll test against a larger variety of SA and SSA Amiibos. We’ll also do 3-4 lives in a match for sure.

DATASTREAM 02 probably won’t start until late February or early March though. My reason for delaying is the simple fact that I will need time to approach specific trainers about using specific Amiibos and I want to weight until the Ice Climbers, King K. Rool, & Piranha Plant Amiibos drop so we can include them in our results. K. Rool and Piranha Plant in particular need some testing to make sure they aren’t Bowser 2.0 since they both seem to have a lot of good wind-up attacks that could benefit from Super Armor.

These Super Armor problems don’t extend to doubles matches

This is cheating a little since this data doesn’t come from DATASTREAM but rather from my doubles tournament S.A.L.T.I., regardless, Doubles doesn’t have the same issues because in Doubles, team attack is turned on. Amiibo spend far less time at the lower damages that SA+ Amiibo benefit from.

Of the 15 teams with winning records, only two have Amiibo with Super Aarmor or Slow Super Armor. The two undefeated teams are made of a Yoshi/Inkling pair and a Pikachu/Mega Man pair. Doubles is just a different world altogether.

So if you are bummed you don’t have a good venue for your Super Armor Amiibo you put a lot of effort into training, I suggest you find them a good partner and start looking for Doubles tournament to train! Any way you slice it (or is that Splice it?…. I’m not apologizing for that word play), it’s a lot of fun and everyone should try it at least once.

Conclusion

I want to thank you all for helping me out with this research project. I implore everyone to constantly ask yourselves tough questions when you are training your Amiibo to figure out if the meta is the problem or your training is the problem. It’s the only way we’re going to improve the metagame.

I hope you all enjoyed reading this and it helps you understand the problems and potential solutions to the Super Armor meta. Regardless of what we ultimately decide, I think this little series of experiments has helped me think more in depth about the strengths and weaknesses of certain Amiibos and has definitely changed how I’m going to approach my Amiibo training. I hope it helps you in the same way!

Remember that growing the metagame, understanding it, and keeping it healthy is a community-wide task! Feel free to do your own experiments and discuss it in the AmiiboDojo Discord. And if you think I’m missing the boat on a point or two, feel free to let me know! A good scientist should always be open to honest, data-driven  critiques.

I hope to see you all in the Discord and on my twitch channel for more Amiibo action and discussion. But until next time:

Stay fit.
Keep sharp.
Make good decisions.
Splice out!

Advertisements

Post a Comment