To the people still discussing the premises and methodology behind a tier list as opposed to individual placings on it, I really want some serious consideration/discussion/criticism of the claim I've been making about what a tier list SHOULD be:
First of all, it doesn't take into account abstract, ill-defined, irrelevant things like "potential" or "how good the character is overall", either. The first thing misses the point completely (which is to give an accurate theoretical representation of the CURRENT high level metagame), and the second doesn't really mean anything. Too many people say things like "Character X has A, B, and C good properties; he should be higher." The thing is, those good properties don't even mean anything in a vacuum; they only mean anything in the context of: do they help this character win against the other characters in the game?
Another common claim is that it's "based (mostly) on tournament results".
The tier list is (and by is I mean should) only be based on tournament results in the sense that tournament results are used to CONFIRM it and make sure that generally your evaluations of characters are accurate. It fundamentally has nothing to do with results in the same way that scientific theories are not defined by experiments, but verified by them.
If it's 'based off tournament results', then there's no need for a special back room of highly knowledgeable people to make the list; anyone who's willing to run the numbers could have made it. Ankoku was doing for brawl (his 'character rankings list'), and the first words in his thread were: IF YOU THINK THIS IS A TIER LIST, YOU ARE A MORON.
UmbreonMow has it as "which character is most likely to win a tournament?", and that sounds pretty good. This question pretty much only makes sense for players of equal skill level, and it's only ineresting for players at the "highest" skill level, so those are universally accepted premises.
Which brings me to my main point: The tier list is (by is I mean should) fundamentally be based on character matchups, and matchups alone. Among players of equal skill, matchup values between characters can be determined in the form of probabilities, and from those probabilities the "character most likely to win a tournament" can be determined.
Tier list ranks would be assigned based on expected winning percent (calculated by an average of the matchup values of the character in question weighted by how much each character makes up the metagame).
This model is good because it not only makes a list that serves the intended function "which character is most likely to win a tournament?", it's also more concrete than just "list the characters from best to worst". It means that the fine-tuning can go to evaluating the individual character matchup values correctly, and that's where the knowledge/experience with the game comes in.
tl;dr:
-tier list shouldn't be based on abstract qualities, potential, or pure tournament results
-tier list should however mostly reflect tournament results
-tier list should be based purely on character matchups, because character matchups are what determines if a character wins or not
You take the challenge? If so go up to him at Event52 cuz hes banned and cant post,
