Stat Benchmarks by Role: Stop Judging a Cypher Like a Jett
The scoreboard sorts everyone by one number, which quietly teaches an entire playerbase to judge four completely different jobs by the same yardstick. It is the most common analytical mistake in ranked.
The core problem
Valorant has four roles with genuinely different objectives. A duelist is paid to take space, a controller is paid to deny vision, a sentinel is paid to hold ground and slow rotations, an initiator is paid to create openings. Only one of those jobs naturally generates a big number on the scoreboard.
So when a Killjoy finishes on 140 ACS and gets told to "stop being useless," the criticism is usually measuring the wrong thing.
Benchmarks by role
These are working ranges, not grades. Use them to compare yourself against other players in your role, then compare yourself against your own past games in the same role — which is the only truly clean comparison available.
| Role | ACS | ADR | KAST | Real measure of success |
|---|---|---|---|---|
| Duelist | 200–260 | 150–190 | 68–75% | First kills won vs first deaths taken |
| Initiator | 180–230 | 140–175 | 72–80% | Assists, and whether your entry survived |
| Sentinel | 160–210 | 120–155 | 72–80% | Site hold rate and flank denial |
| Controller | 150–200 | 110–150 | 70–78% | Whether executes go in cleanly |
Notice that KAST is far more consistent across roles than ACS or ADR. That is the point — KAST measures round participation rather than damage output, so it is the closest thing to a role-neutral stat Valorant has. If you play a support role and want one number to judge yourself on, use that one.
Duelist — Jett, Raze, Reyna, Phoenix, Neon, Yoru and friends
The job: take space first, create the numbers advantage, and do it repeatedly.
Judge yourself on: first kills versus first deaths. A duelist with a 1.3 K/D who never takes the opening duel is a passive player wearing an aggressive kit. See the entry fragging guide.
The failure mode: hanging back for cleanup kills. Your K/D looks great, your team plays 4v5 openings every round, and nobody can tell you why you are losing.
Utility note: your movement abilities are entry tools, not escape tools. A Jett dash saved for the escape you will not survive is a dash that did not buy space.
Initiator — Sova, Fade, Skye, Breach, KAY/O, Gekko
The job: generate information and openings so the duelist can enter and live.
Judge yourself on: assists, and specifically whether your entry duelist survived the rounds you flashed for. An initiator with high damage and low assists is playing a second duelist.
The failure mode: throwing utility on cooldown at nothing. A dart into an empty corner is not information. A flash that pops before your team is ready is worse than no flash, because it announces the push and blinds nobody.
The high-skill version: flashing on someone's timing. Saying "flashing in three" and having a teammate move on the count is the difference between an initiator and a player with flashes.
Sentinel — Killjoy, Cypher, Sage, Chamber, Deadlock, Vyse
The job: make one part of the map expensive to enter, and know instantly when a flank happens.
Judge yourself on: whether your site held, and whether the team ever got flanked without warning. Both of those are round outcomes, not scoreboard numbers — which is exactly why sentinels get undervalued.
The failure mode: playing a sentinel like a duelist. If you are dying first in a majority of rounds on Killjoy, your utility is sitting on an empty site while you fight a duel you did not need to take.
The second failure mode: setups that never change. By round eight, a good enemy team knows exactly where your trip is. Move it.
Controller — Omen, Brimstone, Viper, Astra, Harbor, Clove
The job: decide what the enemy is allowed to see, and when.
Judge yourself on: whether your team's executes go in cleanly, and whether defenders can hold their default angles against you. If your team is entering into open sightlines every round, your smokes are in the wrong places regardless of your ACS.
The failure mode: smoking the site instead of smoking the angles. A smoke on the site itself blinds your own team as much as theirs. Smokes go on the positions defenders hold from, not the space you want to occupy.
The stat trap: controllers post the lowest ACS on almost every scoreboard and are then blamed for the loss. If you main a controller, ignore your scoreboard position entirely and track your win rate on the maps you play instead.
Rounds where you did nothing — no kill, no assist, no damage, no meaningful utility, and you died untraded. More than four or five of those per game is a problem in every role, on every agent, at every rank. Everything else is role-dependent. This is not.
Filling: the honest advice
Solo queue will regularly force you off your role. Two things make that survivable:
- Have one agent per role you can play competently. Not four. One. A player with a passable Omen, a passable Sova, a passable Killjoy and a real main is far more valuable than a Jett one-trick who tilts when someone instalocks.
- Judge your fill games by the fill role's benchmarks. Going 12/14 on Omen after playing Jett all week is not a bad game if the smokes were good and the team won.
See your stats broken down by agent
FragLy pulls your real competitive matches and splits your performance out per agent — so you can compare your Omen to your Omen and your Jett to your Jett, instead of to a scoreboard average that was never meant to apply to both.
Search your Riot ID →The short version
Duelists are measured on opening duels, initiators on assists and whether their entry lived, sentinels on holds and flank warning, controllers on whether executes work. ACS and ADR bands differ by role and should. KAST is the closest to role-neutral, which makes it the best single number for anyone not playing a duelist. And the one universal benchmark is rounds where you did absolutely nothing — that stat has no excuses in any role.
Written by the FragLy team. Benchmarks in this guide are rules of thumb drawn from competitive play, not official Riot figures — use your own tracked matches as the real reference point.