What a tier list is
A tier list groups the characters of a fighting game into bands, often labelled S, A, B and so on, from those considered strongest to those considered weakest. It is a summary of how well each character is expected to do against the rest of the cast when played by people of similar skill.
The usefulness of a summary depends on how it was produced, and the production method is usually less rigorous than the tidy layout suggests.
From matchups to rankings
The raw material is a matchup table. For every pair of characters, someone estimates how the matchup plays out, often written as a ratio such as 6-4 or 7-3. The estimates come from several sources: tournament results, community votes, the opinions of high-level players and sometimes online data. None of these is precise, and different groups can reach different numbers for the same pairing.
To produce a ranking, the table is collapsed into one figure per character, usually an average of that character’s matchups. An illustration with three invented characters shows the process. On a 10-point scale:
- A beats B 7-3 and loses to C 3-7.
- B loses to A 3-7 and draws with C 5-5.
- C beats A 7-3 and draws with B 5-5.
A’s average is (7 + 3) / 2 = 5. B’s is (3 + 5) / 2 = 4. C’s is (7 + 5) / 2 = 6. The ranking is C, A, B.
Even in this tiny example, information is lost. A has a dominant matchup and a terrible one, and sits in the middle. B has no bad matchups at all, but ranks last, because it has no good ones either.
What the average hides
Spread. Two characters can share an average while having very different profiles. One may be steady against everyone, and the other may beat half the cast and lose badly to the rest. For a player, those are very different experiences.
Weighting. A simple average treats every opponent as equally likely. In reality some characters are far more popular than others. If the most popular characters are the ones that your pick struggles against, you will meet that matchup constantly, and the average underestimates its effect. If your weak matchups are rare characters, the reverse is true. Weighting by how often each character appears gives a different, usually more useful number, but needs data that is not always available.
Skill and execution. A tier list typically assumes players of equal skill who can use the character properly. A character with demanding execution may rank high because top players extract the full value, while remaining a poor choice for most players. Conversely, a straightforward character may do better in real play than the list suggests.
The ceiling versus the floor. Some rankings reflect best-case results at the highest level, and others reflect average results across all players. These are different questions, and it is easy to mistake one for the other.
Uncertainty in the numbers
Most matchup values are close estimates and not measurements. A 55-45 figure, for instance, is within the range where small numbers of games, patch changes and player skill can easily reverse the apparent order. Tournament data is also limited. A character who wins few games may simply have few players, and one with many wins may only have many entrants. Win rates from small samples can mislead, a general concern that applies to any analysis based on a handful of results, as discussed in the material on sample size.
Results from a single event are even less reliable. Brackets are random in structure, and how seeding and luck affect outcomes means that a character’s presence in the final may say more about the draw than about the character.
How lists change
A tier list is a snapshot of one version of the game. Developers adjust characters in balance patches, and a few frames of change to a key move can reshuffle many matchups. The community’s understanding also changes, as players find stronger options. A character may rise purely because new techniques are discovered with no patch at all. Old lists therefore age quickly, and a list that does not state which version it covers should be treated with suspicion.
Developers use data of their own, such as win rates by character, to guide changes, though a skewed win rate does not always indicate a problem. The topic of balance as a measurable property of a game is covered in the material on win rate balance.
Common ways to misread a list
- Treating it as a prediction for individuals. A tier list is about averages across many players, not about whether you will beat a specific opponent.
- Assuming a lower tier means unplayable. The gap between adjacent tiers is often small compared with the effect of skill and preparation, and results at events have regularly been won with characters rated modestly.
- Ignoring your own matchups. The matchups you actually face depend on your local scene and your opponents.
- Reading the order inside a tier as exact. Ordering within a band is typically close to arbitrary.
- Using it to decide a counterpick. A ranking offers little guidance on a specific pairing. The matchup table itself does that job better, as discussed in the article on counterpicks.
Reading a list sensibly
A reasonable approach is to treat a tier list as a rough guide to which characters are likely to have an easier or harder time, then check the specific matchups you care about. Notice how the list was made, which version it covers and how many results lie behind it. Enjoy it as a conversation starter, but do not let it make decisions that your own practice and results should make.