Tournament Match Statistics


#1

Anyone who follows professional sports knows that stats have become a central facet of postgame analysis. Personally i watch basketball and i don’t believe modern statistics are a reliable metric for guaging performance, but they can certainly be fun to browse. In that spirit, i thought it might be cool for us to present and discuss some numbers gathered from tournament match replays.

With everything SF4 has done to revitalize our community, it seems like a good place to start. My friend omni and i were discussing Ryu’s Ultra, and i started wondering how important it was for him to be able to land it. So i looked through every match between Evo2k9 SF4 finalists Daigo Umehara and Justin Wong, since Daigo stuck with Ryu the entire time. Here’s what i found:

Daigo (Ryu) vs Justin (Abel/Balrog)
total number of rounds: 29
rounds Ryu won: 16
rounds Ryu lost: 13
total number of times Ryu lands ultra: 7
number of times Ryu lands ultra and wins round: 5
number of rounds Ryu wins without landing ultra: 11
number of times Ryu lands ultra and still loses round: 2
number of times Ryu lands ultra more than once in a round: 0

Ryu ultra setups:
antiair LP DP: 0
antiair DP FADC: 2
ground DP FADC: 3
EX fireball FADC: 1
air to air j.MP: 0
ground combo into ultra: 1

(source: http://www.ustream.tv/recorded/1840331)

Keep in mind that Justin used Abel against Daigo when they first met in Winner’s Bracket. After losing both matches and the first match of Grand Finals, he switched to Balrog. Below are the stats for each character matchup.

Ryu vs Abel
total number rounds: 8
rounds Ryu won: 6
rounds Ryu lost: 2
total number of times Ryu lands ultra: 1
number of times Ryu lands ultra and wins round: 1
number of rounds Ryu wins without landing ultra: 5
number of times Ryu lands ultra and still loses round: 0

Ryu vs Balrog
total number rounds: 21
rounds Ryu won: 10
rounds Ryu lost: 11
total number of times Ryu lands ultra: 6
number of times Ryu lands ultra and wins round: 4
number of rounds Ryu wins without landing ultra: 6
number of times Ryu lands ultra and still loses round: 2

Discuss? I dunno, i was expecting to see Ryu landing an ultra in almost every round but that clearly wasn’t the case. Maybe someone can do the same thing from Balrog’s perspective because he seemed to win via ultra finish a lot more frequently.


#2

shouldnt this be in the SF4 forums? I’m assuming you want it to be more general, but unless you give some more examples from different fighters, chances are it’ll be straight up SF4 talk, which is fine, but then it becomes a thread that should be in the SF4 forums


#3

hmm, never seen a break down like this before. This would be great to figure out someones style and you can even add more to this type of break down. looks like I got a better tool to use now.

post more!


#4

Well, i started it in FGD because i wanted the discussion to cover any fighting game that came up. But both jchensor and omni said the same thing about moving it to the SF4 forum so i got outvoted. Anyway i hope this thread develops into something interesting. Fingers crossed.


#5

I’d like to know the number of assists for the teams at SBO.


#6

That’s pretty interesting maj. Problem is, that’s a massive undertaking heh. But it would be neat, nonetheless


#7

Exactly how many times did Daigo smile? Really, the statistics is too hard to calculate. Especially with all the Japanese daigo matches too :O.


#8

This is pretty interesting actually. You could log all sorts of random stats. Stuff like

How many times Daigo whiffed low forward in Rog sweep range?
How many time fireballs Daigo would throw in a row at mid screen distance?
How many time Daigo get’s hit by naked Rog sweep?
Time he gets hit by sweep due to standing/hitting a button(i.e. jab)/walking forward/etc…?
Number of times he jumped at Justin a round?
Number of times Justin mashiejabs his way out of stuff? (LOL)
Etc…

Not sure what you could pull out of it. Maybe just some tendencies based off stats.


#9

It doesn’t have to be if we start small and focus on situational things instead of trying to build a player database. Everyone can pick some aspect of Street Fighter that interests them, form a theory, and watch a handful of quality matches to gather data.


#10

SF4 has just entered the next level… we could have an error % based on how many times someone drops a combo.

how many jump attempts were made and how many were blocked, how many were hit, how many got AA’d etc.

-dime


#11

I love the stats, I’m also a stat freak when it comes to football(the only sport I watch). Yes as Ed Ma pointed out, you can make stats about everything like how you can do in every sport. But it’s fun to pull stats from the most noticeable things like Ultras or win percentage when ahead by 50% at any point in the match. If someone paid me, I’d be the official stat guy of every major tournament.


#12

Interesting stuff. But man it’d be difficult to keep track of all the various stats, then to organize them properly… damn. You would need whole teams of stat men to do it right.


#13

I took some statistics classes in college. This is a pretty awesome concept. A great statistical program you could use, if you really wanted to get deep into this, is SPSS. It’s the number one statistical analysis software in the world. One of the college students on this forum could get a discount on the program probably. It may even be on the internet somewhere.


#14

I think an interesting stat would be to see the effect of fireballs in the match, if they caused more damage than he took from them ect. Although it would be difficult to add in the effect of the ex meter gain.


#15

There is also a free stat program called “R”. I don’t remember where its at but its free to download. Undergrad years were so long ago heh


#16

What about Justin’s stats?


#17

This is hella cool.


#18

Why not analyze all matches you have from Daigo recently? The idea is great but the sampling group way too small for the results to be representative. A character and playstyle can make too much difference, you need to iron those differences out with a bigger dataset.


#19

Fantasy SF4 anyone?