Counter-Strike is the exception on this site: Valve publishes no per-match statistics API at all. Every figure in circulation — every ADR, every headshot percentage, every rating — was computed by a third-party match service from matches it happened to record. Two services can disagree completely and both be accurate.
There is no official endpoint that returns a player's Counter-Strike match statistics. Steam exposes profile-level information, and the game shows the player their own results, but the per-match record that every statistics page depends on is not published by Valve.
What fills the gap is match services that host or parse games and compute figures from their own pool. That produces an unusual situation: the same account can have a strong record on one service and a mediocre one on another, because the two services recorded different matches. Neither figure is wrong, and neither is complete.
The practical rule follows directly. A Counter-Strike statistic is only meaningful with the service named beside it, and comparing figures between two services is not a comparison at all. This is the single most important thing to know about the game's data, and it is almost never stated on the pages that publish it.
Counter-Strike 2 runs Premier, which issues a numeric CS Rating with colour bands, alongside the older skill groups that still apply per map in competitive. They are separate systems measuring separate playlists, and there is no conversion between them. A player can hold a high CS Rating and an unremarkable skill group on a map they rarely play.
CS Rating decays if Premier is not played, and the season rollover archives the previous rating rather than carrying it. The skill groups do not decay in the same way but are map-scoped, so an account holds as many of them as it has played maps.
The colour bands on CS Rating are the part most often misread. They are presentation, not tiers: the rating is continuous, and two players in the same band can be a long way apart. Reading a band as a rank is the Counter-Strike equivalent of reading a merged K/D.
CS Rating (Premier) plus per-map skill groups. Two systems run side by side. Premier issues a numeric CS Rating with colour bands; the older skill groups still apply per map in competitive. The two do not convert into each other. CS Rating decays if Premier is not played, and the season rollover archives the previous rating.
The share of players sitting in each tier moves every season, so no fixed percentage is printed here, because a number published once would be wrong within weeks. The order is the part that holds.
Average damage per round is the strongest single figure this game produces. Every point of damage in Counter-Strike lands on a health pool: there are no shields, no plates and no destructible cover absorbing hits, so ADR measures what it appears to measure. It also normalises per round, which removes the playtime distortion that ruins lifetime totals.
Headshot percentage is weaker than its reputation. It counts only the finishing shot, and it varies enormously by weapon: an account that buys rifles has a structurally higher figure than one that plays a lot of pistol and SMG rounds, at identical aim. Read it per weapon or not at all.
K/D in Counter-Strike suffers a specific problem beyond the usual ones: it does not distinguish an opening pick from a retake kill, and in a game where the opening duel frequently decides the round, those are not comparable events. A round-win contribution figure would be better, and some services compute one, under names that do not match between services.
The seven recurring figures, the formula behind each one, and the list of things it does not count.
K/D ratio = kills ÷ deaths
A 1.4 in a tactical shooter and a 1.4 in a 64-player mode are not comparable numbers. Read K/D against the mode first, the rank second, and never on its own.
Not included in this number
Win rate = wins ÷ matches played
Below roughly 50 matches a win rate is noise. In squad modes it is closer to a measure of who the player queues with than of how they play.
Not included in this number
Matches played = count of recorded matches
This is the number that tells you whether any of the others mean anything. Read it first.
Not included in this number
Damage per round = total damage ÷ rounds played
The most honest single number in tactical shooters, and the most misleading one in battle royales, where plate-breaking pads it.
Not included in this number
Ranked tier = a band on the game's own rating ladder
The most authoritative figure available and the one most often shown stale. Check when it was last refreshed before believing it.
Not included in this number
Peak rank = highest tier reached, per season
Useful only with the season attached. A peak from a season with different tier boundaries is a different achievement wearing the same name.
Not included in this number
Headshot percentage = headshot kills ÷ total kills
Read it per weapon or not at all. An account-wide headshot percentage mostly reports which guns the player likes.
Not included in this number
Name the service, then the sample, then the playlist. Those three conditions have to be met before any Counter-Strike figure means anything, and no presentation convention on any page enforces them.
For a single player over time, one service used consistently is the most useful setup available: the pool is at least stable. For comparisons between two players, both must be measured by the same service on the same playlist, which in practice means both play on that service regularly.
A ratio is only worth reading next to the match count that produced it. Below roughly fifty matches, every figure on this bench is noise.
Because Valve publishes no per-match statistics API. Every figure is computed by a third-party match service from the matches it recorded, and two services with different pools will produce different figures for the same account. Both can be accurate.
CS Rating is a continuous number issued by Premier. Skill groups are the older per-map bands that still apply in competitive. They measure different playlists and there is no conversion between them, so a high rating and a modest map group can coexist.
No, they are presentation. The rating is continuous and two accounts inside the same colour band can be far apart. Treating a band as a rank is the most common misreading of this system.
Yes, clearly. Every point of damage lands on a health pool: no shields, no plates, no destructible cover, and the figure is normalised per round, so it avoids both the inflation and the playtime distortion that affect other games.
Weapon mix. Only the finishing shot counts, and rifle-heavy play produces a structurally higher figure than pistol and SMG play at the same level of aim. The figure is readable per weapon and misleading per account.