Methodology

What "tier 1 VCT" means

Everything on this site is drawn from tier 1 VCT, the top, internationally-relevant level of Valorant Champions Tour play. The guiding rule is simple:

All internationals, plus the events that qualify teams into those internationals.

What that looks like in practice depends on the era, because Riot changed the structure of the circuit at the start of 2023.

2023–present (franchised)

From 2023 on, VCT is franchised into three international leagues (Americas, EMEA, and Pacific, with China added in 2024) that feed the global Masters and Champions events. The whole structure is tier 1 by design, so every match from these leagues and internationals is included. Nothing is filtered out.

2021–2022 (pre-franchise)

Before franchising, the circuit was an open pyramid. Anyone could enter regional Challengers through huge open qualifiers, and the field narrowed down to the teams that actually reached the internationals. Including all of it would flood the database with amateur and semi-pro teams playing in 256-team open brackets, which badly distorts every leaderboard.

So for 2021 and 2022 we curate by hand to the same "tier 1" principle. We include:

• Internationals: Masters (Reykjavík, Berlin, Copenhagen) and Champions.
• The events that qualified teams into them: regional Masters / Challengers Finals and Playoffs, the regional leagues, and the Last Chance Qualifiers (LCQs).

And we exclude the open-bracket feeder stages that sit below that level:

• Open Qualifiers and Closed Qualifiers
• Preliminary and Seeding rounds
• Anything that’s effectively an amateur open bracket (Round of 256 / 128 / 64, etc.)

Where an otherwise-included event bundled open-qualifier stages alongside its main bracket (a few regional Challengers did), we keep the Group Stage and Playoffs and drop the qualifier stages, so a kept event contributes only its tier 1 matches.

Data coverage

Not every match record carries every block of detail. Coverage is near-total for the three original leagues and the internationals; China domestic play is the gap. Share of maps carrying each block:

Rates divide by the maps that actually carry a stat, so a thin sample isn’t diluted toward zero — and where it’s too thin to mean anything, leaderboards show — rather than a number. Totals are another matter: they count only the covered maps, so a career Operator-kill or clutch total reads low for a player who spent seasons in the China league.

For China league maps, where the per-round detail block isn’t published, multi-kills, clutches and the kill matrix (Operator kills, head-to-head and opening kills) are derived from the round-by-round kill feed — 499 maps so far. On maps with both sources they agree exactly on at least 99.0% of player-maps, stat by stat.

Why it’s curated this way

Tier 1 is about a consistent strength of competition. A career rating or ACS average only means something if every map in it was played against comparable opposition. Letting open-qualifier stomps over amateur teams into the same pool inflates small-sample players to the top of all-time leaderboards, exactly the noise this scope is designed to remove.

Percentiles and peer groups

A percentile says where a figure sits in a named group of peers: “88th pct vs 64 Duelists” means the figure is better than 88% of those 64 players. It is the mid-rank: the share of peers strictly below, plus half the share tied with it, so a block of identical values shares one number.

The peer groups, and the sample each member needs:

• Players in a season: everyone with 20+ maps that season across all leagues, grouped by main role (the role they played the most maps on). A role group under 15 players widens to every role.
• Teams in a season: every team with 12+ maps that season.
• Players at an event: everyone with 4+ maps at that event, all roles. With fewer than 10 of them, nobody is ranked.
• Teams at an event: every team with 3+ maps there. With fewer than 6, the event's tables carry no bars.
• Teams on a map: every team with 5+ maps on that map in the season.

Each stat also needs enough maps that actually record it: 10 in a season, 4 at an event. China league matches record fewer detail blocks (see Data coverage above), so a China player's multi-kill or opening-duel figure can read “Not ranked” while their rating is ranked. A figure under its floor is always shown; only its percentile is withheld. It is never turned into a low percentile, because a thin sample is a coverage fact, not a skill fact.

Attack and defense stats divide by the rounds played on that side, on maps that record the side split, never by all maps. First deaths per round is lower-is-better, so its percentile is flipped: a longer bar always means better.

In a table, the thin bar under a figure is its percentile within the group the scope line above the table names. On Statistics that group is every row the table ranks after its own minimum-maps filter, not only the rows on screen. The Bars switch hides them.

The strip on a stat tile is the peers’ distribution: 16 equal bins between the group’s 2nd and 98th percentile (so one outlier cannot squash everyone else into two bins), the subject’s bin filled and the median dashed. Bar colours are five equal fifths of 0–100; colour carries the percentile and nothing else.