TafoStats: How Mew2King lost at Stage-Select

by MIOM|Tafokints

The metagame has many unspoken rules in how to pick stages in a set. Often, these assumptions about what is considered a “good” stage have gone unchallenged. We hear that Yoshi’s Story is a Marth’s Stage and that Pokemon Stadium is a Fox’s stage. The rationale seems to make sense; Marth likes close encounters and cornering his opponent, so Yoshi’s Story seems to be a great choice. Stadium has low ceilings and transformations that seem to favor Fox.

Recently, I’ve been observing Mango and some of his strengths and weaknesses extensively.  From watching some of the earlier summer tournaments, I subconsciously noticed that Mango was losing quite often on Stadium, a “Fox Stage”, and began to track his win-loss records.

Stages Mango Mew2King Mango’s Win Rate
Yoshi’s Story 15 5 75%
Final Destination 5 14 26%
Dreamland 5 10 33%
Fountain of Dreams 12 3 80%
Pokemon Stadium 2 10 17%
Battlefield 6 3 67%

Table 1: Mango vs Mew2King (2013-2014)

1. Initial Impressions

Although the sample size remains rather small, we can still gather some conclusions. The stage win-loss ratios between Mango and Mew2King are vastly different than what people would generally expect. Mango performs exceptionally well against Mew2King on Yoshi’s Story. Even though Mew2King went Sheik for a handful of these games, Mango was still dominant, showing that Marth may not have a strong advantage on Yoshi’s Story as many would think.

To no surprise, Mew2King performed really well on Final Destination. His strong 0-to-death combos on Fox and Falco give him a strong edge on that stage. Mew2King also performs well on Dreamland. The long horizontal stage along with a large ceiling, give Mew2King an ability to recover with his fantastic recovery and allows him to slow down the tempo of the game whenever Mango tries to be aggressive. In general, Mew2King thrives on stages that allow for movement.

2. Surprises

Now for the non-conventional line of thought

Mango is anemic on Pokemon Stadium. This is widely consistent with how he fares against the other members of the “big 5″ on that stage (stats will be released eventually). This seems unusual since the low ceiling should work in favor of Fox/Falco and the transformations tend to be amazing as well. So why is this the case?

A. Lack of a Top Platform

It’s no secret that Mango recovers high. On a stage with a top platform, it gives Mango a plethora of options to recover.


Image 1: Mango’s options on Dreamland

In this situation on Dreamland, Mango has several options with a double jump. He can do several options (I’ll just list three, but there are much more)

1. (Red Route) Mango can use his double jump and FireFox/Side+B onto top platform.
2. (Orange Route) Mango drops a little bit, chooses to use a jump based on Mew2King’s reaction, and Side+Bs onto ledge or onto the base stage
3. (Yellow Route) Mango can challenge Mew2King’s reaction by using Side+B immediately to the stage

Out of these 3 main options, there are several permutations that can be added as another layer of depth to Mango’s recovery. Even a player as great as Mew2King cannot cover every option that Mango has in his disposal. In contrast:


Image 2: Mango recovering on a large stage

On Stadium, the number of options that Mango can do is much more simplified because of the timing windows involved. If Mango aims for a high recovery, he’ll take a long time to land and Mew2King will be able to catch him and combo/edge guard him. As a result, Mew2King doesn’t need to cover as much space to edge-guard Mango as he would on a stage with a top platform.

B. Platforms are terrible for Fox/Falco

Youtube Clip 1: Mango getting combo’d on platforms

The platforms essentially work in Marth’s favor and act as an easy extension to combo. His u-tilts can cover everything on reaction, so Mango cannot reliably hide on the platform. Either way, the neutral game works quite well in favor of Mew2King with generous space to dash dance and grab or force Mango onto a platform and punish him with utilts and uairs.

B. Stage Transformation Advantages are Overblown

For the most part, Mew2King is very much aware of when the stage will transform. If he doesn’t like it, he’ll simply stall until the stage reverts back to normal.

3. Why Fountain of Dreams was a disaster at Evo

Many suspect that Fountain of Dreams is a terrible stage for Fox due to its high ceiling and interesting platforms that can mess up a Fox’s tech-skill immensely.  However, there are many factors that lead me to suspect that FoD is not as bad as one would expect.

A. Fountain of Dreams limits follow ups

# of Openings # of Kills Total Damage Average Hits Per Opening
Game 1 19 3 357 2.06
Game 2 17 1 267 1.65

Table 2: Mew2King’s Statistics on Fountain of Dreams at Evo 2014

It’s interesting to see how Mew2King’s punishment game works on Fountain of Dreams. He averaged an abysmal ~2.06 hits per opening in game 1 and ~1.65 hits per opening in game 2. Although I don’t have the statistics, I imagine that Mew2King has a much better follow up game on every other legal stage. Although Mew2King had 19 openings, he only landed 3 kills. In game 2, he had 17 openings, but had only one kill to show for it. It’s clear that the unique platforms play a large role into the lack of followups, but it doesn’t hurt that Mango also has stellar SDI and game knowledge to minimize punishes. In a match against a high caliber opponent, getting a kill off of a small handful of openings is crucial.

Game 1 Game 2
Stock 1 75% 156%
Stock 2 163% x
Stock 3 79% x
Stock 4 x x

Table 3: Mango’s Death % on each stock

It’s not to say that Mango lived extraordinarily long either. Mew2King was able to land hits into kills and nailed most of his edgeguards in the set. In Mew2King’s four KOs, Mango died at an average of 118%. It’s a tad high of a percentage, but certainly not abnormal relative to their other games.

B. Lack of Horiziontal Maneuverability

Mew2King’s specialty is being able to dash-dance to outspace and bait his opponents. With a thin stage such as FoD,  Mew2King doesn’t have the option to move as well as he would on Stadium, FD, and Dreamland. In the neutral, Mew2King found himself frequently in shield, at the mercy of Mango.

Youtube Clip 2: Mew2King stuck in Pressure

Mew2King is caught in this situation frequently and his escape routes are rather limited to deal with this. Wave-dashing OOS, rolling, or jumping all pose immediate risks for Mew2King and will either put him in the corner or in a very disadvantageous spot. As a result, he’s never really able to control the pace of the match since they are so close to each other all of the time.

 4. Closing thoughts

It’s important to understand your weaknesses and strengths as a player when it comes to stage choice. Mew2King thrives on larger stages, especially against Mango, and the statistics back this theory up immensely. It’s not to say Mango didn’t played well (which he did), but the set would have been much closer had Mew2King chosen stages that catered to his personal strengths. Statistics and solid metrics help back-up these claims and it’s nice to see how metrics can help people strategically.

12 thoughts on “TafoStats: How Mew2King lost at Stage-Select

  1. I think it’s simpler than that. I wasn’t playing as well as usual lately. I do understand statistics, but in this case it was more him playing better than me recently.

  2. This kind of analysis is beyond a breath of fresh air, it’s absolutely amazing. Something that’s been bothering me lately is that when people talk about stagepicking strategies, it’s almost always in the form of anecdote (Mango does really well on this stage, people say this is a good counterpick, etc etc) and it doesn’t make any sense that what should be the science of counterpicking seems more like alchemy. You, sir, are doing it right.

    • The reason that it’s anecdotal is there are a dozen reasons why someone could like or dislike any given stage. It’s not possible for a commentator to know exactly what has happened in the first 1-4 games that are influencing the decision behind the counterpick.

  3. LOL m2k just like “nah i just fucked up”

    for real though i do agree that this type of analysis, even statistics aside, is a great idea and is really great for the community. too often when i was first really new to the game and would ask for advice from better players i would get either “don’t roll” or “play better”. the latter is basically completely useless advice, and while the former tries to be slightly more informative, it just sort of gets tangled up in needless hyperbole (that is, obviously you can’t NEVER roll, it’s just usually stupid because it’s usually very obvious).

    in particular though i would be interested to see how these stats bear out for the same MU played by different players. of course you probably just went with mang0/m2k because they play, at least in theory, the MU to the highest level currently available and so provide the best candidates for such an analysis, but it would be interesting if the results are very different for other players of somewhat comparable skill levels–or perhaps similar, but different in predictable ways.

    overall the type of option analysis you did is exactly the type of information that can spur even more growth in the community and just expand the competitive discussion of melee as a whole. just those couple edgeguarding diagrams would go a long way in showing the inexperienced player or uninitiated viewer the competitive complexity of the game but likewise it’s surprising simplicity–a melee match is above all (even moreso than other fighters imo) a battle for positioning, and consists of players shifting around the stage attempting to gain a foothold against the other player by mutually guiding the gamestate through a variety of situations that, while they make look disjointed or unrelated on the surface, are actually highly recurrent. this forces players to choose one of a few options not just once, but multiple times. the distribution of these choices then constitutes what we would call “habits”. thus the task of each player is to choose carefully each time in such a way that the opponent cannot either predict or react to the chosen option, and on the contrary each player must play diligently to outpredict their opponent’s choices. of course, with the sheer range of motion available to the player in melee, there are myriad situations that would take forever to document, but i like the examples in this article because they are clean, relatively common situations, and i’m sure similar analysis could be applied to other such common situations as well.

    this is all surely stating the obvious to the best players, but not everyone can be a natural, and if we advise new players in such an articulate way as presented in this article, i think it would do a hell of a lot more good than such vague terms as “recover high” or “don’t roll”. while these might be enough to get the point across to an intermediate player, such terminology often only serves to confuse beginners and cost them a lot of valuable training time.

    sorry for the long post but i wanted to give full feedback because you definitely did your homework! great read!

  4. Statistics are awesome and we need more of them in Smash. Thanks Tafo!

  5. Very nice analysis. I would love to also get into this sort of thing.

    One question though: What did you consider an “opening”? Technically every single time that M2K was within the hitbox size of shine, it was an opening.

  6. Really great stuff. I’m no Melee player myself so I always had a lot of trouble understanding how players return to the stage without getting punished. I’m so used to just sweet spotting the edge in Brawl. This really helped me understand Melee that much more.

  7. Sick article tafo

  8. Great analysis. I do want to say though that your discussion about a lack of top platform seems a bit off. The recovery options you discuss in the Dream Land example are clearly also available on Stadium. In fact in the Dream Land example, Fox can’t even reach the top platform from where he is (and it’s worth noting that recovering any closer would be much more dangerous). It seems like a much more relevant fact (in terms of recovery) is where the lower platforms are located. On Stadium, the platforms are more than twice as far from the ledge as on Dream Land. You still have all the same options, but you can’t threaten all of them at the same time.

  9. I’m really happy you wrote this article because it’s created a ton of re-evaluation of conventional stage choices. People are finally realizing that Marth isn’t quite as dominant on Yoshi’s Story and that he has a lot of offensive tools on Pokemon Stadium instead of painting those stages as Marth and Fox auto-wins respectively. That said, I do think this particular kind of data could be confounded if it’s taken a step beyond this level of specificity. This data is taken from the right place for this analysis to be as valid as possible, but I think the stage selection system is working against us.

    Most tournament sets are best of three with bans, and most of those results are 2-0s and not 2-1s just because the better player tends to win, so you have to find players close in skill to get a 2-1 result. If a player wins 2-0, he never chose a stage. Game 1 was played on a neutral stage, and for game 2, his opponent chose the stage. So more wins will be registered for a character/player on the stages which are perceived to be their weakest.

    Mango Mew2king is a matchup for Semis and Finals, and if it doesn’t get seeded that way then I don’t know what the TO is thinking. Therefore they’re always going to play best of 5 sets. This increases the stage pool a little bit, and it takes away that ban, allowing the players to actually choose their favorite stages. You’ve even taken a data set such that both players have an equal number of wins. Narrowing the scope to this particular matchup makes your stats more valid.

    Even then, the matchups are volatile enough that momentum could have a similar effect. If Mango was having a good day and Mew2King a bad day, Mango might 3-0 or 3-1 M2K. 2 of the games Mango won in either set would be on Mew2King’s counterpicks. Both players might agree that the stages Mango won on are bad for Mango, but they’ll be recorded as wins anyway. Only sets which go to game 5 give both players an equal chance to choose their stages.

    • Quick correction, the word “would” in the last paragraph should have been “could”. It’s possible in a 3-1 set that Mango wins on the neutral stage, M2K’s CP, then Mango’s own CP.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s