Mew2King's Performance at Smash Summit 3

I’ve watched a bit of smash, and I’ve always been a little curious about stats are tracked for the game. tafostats is very helpful, but unfortunately as far as I could tell, there isn’t much more detail about specific in-game statistics. Here is a little experiment that I’m trying with annotating games to see what kind of interesting numbers could be found from SSBM tournaments.

This first post follows Mew2King’s performance at Smash Summit 3 from this past year. Because this first dataset is so small, this post won’t try to draw any concrete conclusions or trends from Jason’s playing style. It’s simply a presentation of some numbers I found interesting from his games at the tournament.

Before we dive in, a little about the stats that are tracked. It was all centered around kills/deaths, and I tried to only record things that are not subjective. For example, for each death I recorded stats like the attacker/defender’s percentage, what part of the stage did they die on (left, top, right), the move used, and a couple others. I also recorded the timestamp for each death for further analysis.

Overview

Before we get into some of the smaller details, let’s go through some of the high level stats from the tournament. In total, M2K played in 5 sets, for a total of 20 games. He went 3-2 overall, and went 10-10 in his games. Figure 1 below is a table that represents his individual character vs. character set records, as well as the average game time for each matchup.

character matchup	win-loss record	avg time per game
sheik vs. falco	1-1	3:59.5
marth vs. peach	1-3	5:56.5
sheik vs. captain falcon	3-2	3:51.2
marth vs. falco	2-1	3:06
fox vs. jigglypuff	0-3	3:39.333
marth vs. fox	3-0	3:23.667

As you can see, M2K’s best record is as Marth when playing against Fox, but this is also when he 3-0’d Leffen in Winner’s Quarterfinal. Unfortunately here, most of the win-loss records are representative of specific rounds in the tournament, versus single players. You can see the Marth-Peach matchup versus Armada, the Fox-Jigglypuff from Losers’ Finals with Hbox, and the 5-game set with S2J (Sheik-Captain Falcon). In time as I catalogue and annotate more tournaments, I hope to be able to show large collections of pro character matchups.

The one interesting thing to note here is related to the average time per game. For the most part, the quick games make sense. Marth-Fox and Marth-Falco are both heavy damage hitters and can kill quickly. The sole long game with Marth-Peach also makes sense. Surprisingly, the Fox-Jigglypuff matchup was the third-fastest, coming in at 3:39.33. From what I’ve seen, Hbox’s Jigglypuff games sometimes take quite a while, but I’m attributing this to M2K pulling out Fox, a character he doesn’t play as much.

character	move	% of kills	avg defender %	# of edge guards
marth	forward smash	23.5294	140.125	0
	forward air	20.5882	108.571	4
	down air	8.8235	125	0
	up b	8.8235	110	0
	down smash	5.8824	145	0
	down tilt	5.8824	158	0
	forward tilt	5.8824	114	0
	up throw	5.8824	211.5	0
	b special	2.9412	159	0
	back air	2.9412	177	0
	neutral air	2.9412	202	0
	n/a	2.9412	36	0
	up air	2.9412	164	0
sheik	back air	45.8333	116.091	0
	forward air	29.1667	128	0
	neutral air	16.6667	113.5	1
	down smash	8.3333	106	0
fox	up smash	50	101	0
	back air	25	139.5	0
	up air	25	82	0

Figure 2 is a breakdown of all of M2K’s kills throughout the tournament, grouped by each character. Mew2King uses a variety of methods to kill his opponents, but there is a clear gap between Marth’s best attacks and the other methods he uses. The only edgeguards¹ with his Marth came from his off-stage forward air combos. The highest average defender % came from up-throws into oblivion, while the lowest non-SD kill came from the forward airs. As a note, I am not tracking combos or anything related because of the subjectivity, these are just the last-hits.

M2K’s Sheik stuck to the basics and used aerials for 92.67% of his kills. His Fox had even less variety against Jigglypuff and of the 8 kills, half of them were up smashes.

Specific Stats

Now that we’ve got some of the basic overview stats out of the way, let’s look into some more specific analysis of M2K’s play.

Gimp Kills

One of the first things I wanted to check was the number of times that M2K was gimped and the number of times he gimped others. The criteria for a gimp kill here is not too specific, the player getting killed must have less than 50 percent damage at the time of death. In total, M2K’s Marth got 3 gimps in the tournament, 2 of them with his forward airs, and one with a forward tilt. Surprisingly here, none of them came from down air punishes.

gimp kills	marth - forward air	2
	marth - forward tilt	1
gimp deaths	fox - down b	1
	jigglypuff - down b	1
	jigglypuff - forward air	1

He got gimped 3 times as well, one of them coming from a shine spike and another one from an early rest from Hbox. The lowest percentage M2K died at was 15 percent, when he got hit by a forward air from Jigglypuff near the bottom of Final Destination.

Comebacks

Next, let’s look at how Mew2King performs when he goes down by a stock or more. This was a little bit tougher to analyze and breakdown, simply because the stock count at any given time in a game is not a true indicator. Let’s say a game starts 4-4, and one player kills the other, but not before her character takes 85% damage. If you look only at the stock count, it would look like she has the lead but you miss out on all the work the second player did to bring her percent up to 85%. There are also other intangibles such as momentum that can’t be tracked with the statistics I’ve gathered.

For the purpose of this analysis, we’ll look at occurrences when the stock count goes to 4-3, 3-2, 2-1 and the attacking player has less than 50%. This analysis also does not include the inverse, when Mew2King goes up by a stock against his opponent.

In this tournament, there were 11 total situations where M2K went down by 1 stock and his opponent had less than 50%. These 11 moments happened across 7 total games. In 2 games, M2K went down 4-3, tied it, then went down 3-2, and tied it one more time at 2-2 before going down 2-1. Of these 7 games, his win percentage is 28.57%, which is not too surprising. For those two times that he was able to pull off the win, the first time² he went down by a stock was on the 4-4 even stock. If the first time he went down by a stock was at a count of 3-3 or 2-2, Mew2King was not able to comeback and pull out the win.

In both of those games where he went down by a stock and tied multiple times, his win percentage was 50%, so nothing indicative. As I collect more data, I’m curious to find out whether or not that number will change.

Damage Per Minute

The next stat I began to look into was damage per minute. My guess is that this speaks more about the character matchup than the person playing it with limited stats such as these. In the long term, I hope to establish some benchmarks on the average damage dealt per character per minute, but for now there’s not much else to do with this data.

player	character	damage per minute	minutes played
armada	peach	94.1233	23:46
hungrybox	jigglypuff	106.048	10:58
leffen	fox	112.243	10:11
m2k	fox	94.9237	10:58
	marth	114.451	43:15
	sheik	112.22	27:15
mang0	falco	126.307	17:17
s2j	captain falcon	91.2974	19:16

What’s Next

If you’ve made it this far, I hope the numbers and charts presented have been interesting and piqued your curiousity. There’s a bunch more ideas for stats that I would’ve liked to calculate if I had the time. Ideas such as streakiness, clutch factors, efficiency ratings and more have been going through my head, but I decided to get this out the door sooner for feedback.

In the near future, I’m going to keep working on annotating as many games as I can from upcoming tournaments. If you want to help out at all, let me know and maybe we can figure out a way to divide up the work to get stats more quickly and reliably!

My background is in programming, so I’ve been analyzing these datasets using Python and csv’s for most of it. I’m sure there are others that might have some pretty neat ideas for how to manipulate the data, so building an API to make all of it available is definitely something on my mind as well. Of the stats I’ve covered here, all of these analyses are generalized and coded up so it shouldn’t take long to publish more posts and reports.

For my next in-depth post, I’ll probably try to cover another player at the next big tournament. If you have any more ideas or feedback, please send them my way so I can make that one even better.

---