1/25/2016 · originally published on uselessanalysis.com

Stupid, Stupid Idea: Who's the Best Street Fighter?

When I was a kid in the 1990s, I had very strict quotas on how much I could play video games. Once I hit the limit, my father would take the Super Nintendo controllers and hide them somewhere in the house. At first, this led to great games of hide-and-seek, where I'd scour every possible location until I found the controllers and snuck in an hour of Mario Kart. This led to better hiding places, which led to longer search periods. Eventually, I cracked.

Instead of spending an hour looking for Super Nintendo controllers, I decided it'd be more entertaining to put a game into the Super Nintendo and watch the demo mode where the computer would play itself. And I'd sometimes do this for hours.

Often, the game I watched was Street Fighter II, the game that arguably defined a generation of gaming and inarguably defined the fighting genre. It has a memorable cast of characters, incredible music (including the Guile meme), and one of the most hilariously bad movies ever made. It also has a tremendous documentary about the game's impact on the video game scene that I highly recommend everyone watch.

As I'd watch the demo computer play against itself, I always wondered, "who would win a tournament if all of the characters were controlled by the computer?" I didn't know how to do the analysis 20 years ago, but I sure do now!

So today's Stupid, Stupid Idea: I watch several CPU vs. CPU matches in Street Fighter II, record who wins, and then create power rankings similar to my old OLANDIS formula for the NFL. Specifically:

• I turned on the Super Nintendo version of Street Fighter II Turbo (my favorite version, since you can play as the bosses) and didn't touch anything. This brings up demo mode, in which the game shows each of the twelve characters on their home stage fighting against a randomly selected opponent.
• I watched five full cycles, meaning I kept track of 60 matches. In some cases, I couldn't tell who was ahead when time ran out so I called it a tie. The game also randomly chooses "mirror matchups" where the same character fights himself. I just simply ignored these matchups since they don't help answer the question.
• I then ran an optimization model in order to rank the characters in order to account for both record and strength of schedule. This model allowed for ties - in fact, a tie against a good fighter is better in some cases than a victory against a weak fighter.

#12: Dhalsim (0-4 record, 0 power rating)
Dhalsim was the only fighter to go winless in the simulation, and none of his battles were particularly close. The game seemed to feel bad about it, as he showed up in the fewest number of fights overall, and the game burned two of his contests on "mirror matchups" against himself.

#11: Balrog (1-8 record, 21 power rating)
Balrog was always the easiest boss to beat in Street Fighter, and he performed poorly here as well. But he beat Dhalsim head-to-head, and that's enough to keep him out of the gutter.

#10: Chun-Li (3-7-1 record, 40 power rating)
Chun-Li beat both Dhalsim and Balrog head-to-head, but she didn't do much else. The game really wanted her to fight against the Shoto boys Ryu and Ken, as 7 of her contests were against those two. She was 0-4 against Ryu and 0-2-1 against Ken.

#9: Ryu (8-6-1 record, 55 power rating)
What? The best character in the whole series is only #9? The game seemed to give him tons of opportunities, as he showed up more often than anyone else. His 8 wins top anyone else in the simulation, but 7 of those came against Chun-Li, Balrog, and Dhalsim. And then there are those losses - 6 total, with 3 coming against Blanka. Ryu was competitive, though, and his power rating is closer to the #4 spot than it is to Chun-Li at #10. Let's chalk this up as the stunner of this article and move on.

#8: Zangief (3-4-3 record, 58 power rating)
This feels about right for Zangief - not bottom tier, but not one of the true contenders. Two of his wins came against higher-ranked fighters Ken and M. Bison, but his losing record holds him back.

#7: Vega (3-3 record, 60 power rating)
Vega didn't show up much, and didn't make much of an impact in his fights. It's very unbecoming for such a memorable and charismatic character.

#6: Blanka (5-4-1 record, 63 power rating)
Blanka went 3-0 against lower-ranked Ryu and 0-3 against higher-ranked Sagat. Looks like he's in the right place on this list.

#5: Ken (7-3-2 record, 67 power rating)
Ken's 7 wins came against Chun-Li, Balrog, Zangief, and Vega, while he lost against higher ranked M. Bison and E. Honda. Looks like he's in the right place too.

​#4: M. Bison (3-3 record, 69 power rating)
So how does M. Bison get this high up with a .500 record? It's all in the strength of schedule, which is critical when matchups are randomly determined. M. Bison defeated Vega and Ken, split two games with E. Honda, and lost against lower-ranked Zangief and higher-ranked Sagat. The Vega and Ken wins provide enough of a buffer to cover the Zangief loss and pull the Shadaloo boss into the #4 spot.

#3: E. Honda (5-3-1 record, 74 power rating)
E. Honda is a beast in this game - that's why he's on the cover! He notched high-quality wins against Zangief, Vega, Ken, M. Bison, and higher-ranked Sagat, and that helped to cover his three upset losses against lower-ranked Ryu, Vega, and M. Bison. Specifically, the Sagat win is huge because that's who's next ....

#2: Sagat (6-4-1 record, 81 power rating)
Sagat finished 6-0 against fighters ranked #4 and below and 0-4-1 against fighters in the top three. So why is Sagat ahead of E. Honda who beat him head-to-head? He has no losses as bad as E. Honda's Ryu loss. Plus, no one else had to face the #1 fighter four times!

#1: Guile (5-0-2 record, 100 power rating)
Guile is up here for one reason: he finished 3-0-1 against Sagat head-to-head. That makes him the champion of this useless tournament. Time for a coronation song!

So is this the most useless exercise of all time? Have I topped the nerdiest story ever told?

-rj