In a nutshell, it feels to me like result, all the time, across all types of games, goes against the way tier-lists are shaped now(aka lots of subjective opinions)
What I would want to see is a tier-list based on numbers and performance.
That's not how it works, though.
Kuroda can win pretty much any 3s tournament playing Q, or probably half a dozen other characters.
Does that mean those CHARACTERS, in the hands of an average-good player playing another average-good player, are likely to win?
Nope. It's because Kuroda is amazing.
A tier list basically indicates the likely level of success an average-good player will have overall across a large number of situations. It tries to sum up the tools each character has vs every other character, and gives a rough indication of some sort of success-vs-skill ratio if you want to learn a character from scratch*. It does not in any way apply to a
specific match that is played, a tournament, or even a tournament season. It is only a general summary, and it assumes players of equal skill, which is almost never the case in real play.
The problem with tier lists is usually that
players try to make them based on not-nearly-enough matchup data or play experience, which is basically this entire thread (and why it is useless).
The more matches played of a given game
after the point at which the game has stopped majorly evolving, the more accurate tier lists tend to be. Although tier lists also have problems if the game has strong universal mechanics that can alter the outcome of any match, like parrying in 3s.
Some characters defy tiers (Makoto in 3s is usually listed as ?:? because it really is up to the player - if she guesses right twice you're dead no matter who you are and if she guesses wrong then bye), and some tier lists are pretty accurate. MvC2, GGXXAC, ST, etc have pretty accurate tier lists at present. But they still will not
ever predict the outcome of a single match, or even a single FT10 or FT50.
On topic:
MvC2 ranks popular teams, if you really want to do it for SG you could do the same thing, i.e. you don't have to cover s.LP assists.
* this ignores how well a character fits a given player's playstyle, which can often be a better prediction of success than how high-tier that character is.