TEKKEN 8 Tier List: Tournament Usage

Compare TEKKEN 8 character usage and recorded win rates from 654 games across 11 tournaments. See the method, sample and sources.

Updated 2026-10-24T04:48:27.353Z. Sample window begins 2026-04-27T04:48:27.353Z. 64 event-scoped entrants. Coverage: 52.2% of scanned games.

Recorded top-eight character selections
Rank and bandCharacterSelectionsWinsWin rateUsage share
1 (S)Sergei Dragunov1066965.1%12.7%
2 (S)Fahkumram1065955.7%12.7%
3 (S)Shaheen694565.2%8.3%
4 (S)Marshall Law643453.1%7.7%
5 (S)Zafina553258.2%6.6%
6 (S)Jack-8552850.9%6.6%
7 (S)Jin Kazama462656.5%5.5%
8 (S)Kazuya Mishima452964.4%5.4%
9 (A)Miary Zo412663.4%4.9%
10 (A)Yoshimitsu402665.0%4.8%
11 (A)King341750.0%4.1%
12 (A)Kunimitsu291758.6%3.5%
13 (A)Reina281657.1%3.4%
14 (A)Hwoarang261142.3%3.1%
15 (A)Bob201155.0%2.4%
16 (A)Armor King13646.2%1.6%
17 (B)Heihachi Mishima11872.7%1.3%
18 (B)Ling Xiaoyu10660.0%1.2%
19 (B)Leo10330.0%1.2%
20 (B)Lili8562.5%1.0%
21 (B)Kuma8337.5%1.0%
22 (B)Lee Chaolan6350.0%0.7%
23 (B)Devil Jin400.0%0.5%

What this tournament tier list measures

This page ranks observed character usage among top-eight finishers at sampled completed start.gg events. It is a tournament evidence table rather than a universal statement of character strength. Usage measures which characters were recorded in the sampled games, while win rate describes the outcomes attached to those selections. A character can be popular without being the best choice for every player, and a specialist can produce excellent results with a less common character. Read the sample and source events before interpreting a rank as a recommendation.

How the event sample is selected

The collector searches featured completed events from the last 180 days for the exact game, then selects up to six of the largest qualifying fields from the bounded discovery result. Each selected event must have at least sixteen entrants. The collection is bounded rather than an exhaustive census of every tournament worldwide. It inspects a limited number of recently completed sets and reads the reported character selections inside individual games. Events that do not report selections contribute no invented character observations, and a larger attendance figure does not compensate for missing game data.

Top-eight selection and observation units

The event standings identify entrants who finished in the top eight. A character observation belongs to one of those entrants in one reported game, with duplicate copies of the same entrant and character selection removed. An entrant using several characters across a set can contribute to several character rows. In a tag-team game, several characters may be selected for the same entrant in one game. Those observations describe team selections, so their win rate is not an isolated estimate of the individual character contribution. Entrant counts are event-scoped and do not claim to be deduplicated people across different tournaments.

Usage, win rate, and small samples

Usage share divides a character observations by all qualifying character observations in the sample. Win rate divides recorded wins by that character observations. A win is attributed by comparing the reported game winner with the entrant attached to the character selection. Missing selection data is not filled from a social post, a player reputation, or another game in the set. A row with few observations can change sharply after only a few additional games. Compare both the numerator and denominator rather than treating a high percentage from a tiny sample as strong evidence.

How ranks and bands are assigned

Rows sort by reported selection count, then by an adjusted win-rate tie-break, then by a stable character identifier. The adjusted rate uses five prior wins and five prior losses to soften small-sample extremes; the table still shows the actual recorded wins and selections. S, A, and B are equal-sized rank bands based on this ordering, not expert claims that every matchup within a band is equal. They are labels for the displayed sample, not a publisher rating or a guarantee of competitive success. Matchup knowledge, execution, team composition, and player experience remain outside this ranking.

The publication threshold

A game page qualifies only when the sample includes at least three distinct tournaments, one hundred reported games, sixteen event-scoped top-eight entrants, and qualifying selections in at least half of scanned games. The snapshot must also have been refreshed within fourteen days. These are minimum evidence thresholds, not proof that the sample represents all players or all regions. A page that does not meet them is withheld from indexing and sitemap publication. The system does not fill an empty tier list with opinion in order to publish a search page.

Patch versions, missing reports, and bias

Tournament data can reflect multiple patches across the collection window, and not every organizer records a precise game version. This dataset does not infer patch identifiers from dates or present mixed-patch results as a controlled balance study. Featured events, top-eight finishers, incomplete character reports, and the bounded recent-set selection all introduce selection bias. Mirror matches can also pull aggregate win rates toward fifty percent. Use the source links to inspect the competitive context, and avoid comparing two games as though they had identical reporting coverage or identical tournament populations.

Reproducible dates and useful follow-up

The page exposes the generation time, sample window, game counts, coverage, and source event links so a reader can assess freshness and limitations. A scheduled worker refreshes snapshots without fetching providers from a public page request. A failed refresh preserves the previous successful data, but the freshness threshold still prevents an indefinitely stale ranking from remaining publishable. Use the tier list maker for an explicitly personal ranking, the glossary for terminology, and Radar for playable combo examples. When citing this table elsewhere, include the game, update date, observed sample size, and the fact that it is a top-eight tournament usage sample.

Source events