00 · Introduction & Methodology

Why this document exists

We are three players who want to become a team. The difference between those two things is not aim — it's defined roles, understood deeply, and synergy built on purpose. This book is the shared reference for that project: it says, in writing and in numbers, who each of us is, what our job is, how we fit together, and how we get better. It is meant to be argued with, updated, and lived by.

It is deliberately honest. We celebrate what the data says we're good at, and we name what's costing us rounds — because you can't improve a weakness you're not willing to look at.

What makes this different from Leetify / CSStats

Public trackers measure each player in a vacuum — your stats, ranked against the ladder. Useful, and we use them for individual scouting. But they structurally cannot answer the questions that turn three players into a team:

  • When Knox entries, does KingG actually trade him — within two seconds?
  • Whose flash sets up whose kill?
  • When we go a man up, do we close the round or throw it?
  • Do we win because we plant, or in spite of it?

Those are relational and situational questions. They require parsing our own demos and computing metrics across our specific five-man (well, three-man core). That's what the pipeline does, and it's the entire reason this analysis is possible.

Methodology

  • Source data: our own FaceIt pug demos (.dem.zst), parsed with a Dockerized Go pipeline built on demoinfocs-golang v5. Code: ../pipeline.
  • Scope: we are a 3-stack inside 5-man pugs — three of five slots are ours, the other two are random teammates each game. Every "team" metric below is computed for our side (the team our core trio is on that round). The two randoms are part of our side's man-counts but are not profiled.
  • Sample (this edition): 50 demos, 1,125 valid rounds (Aug 2025 – Jun 2026), spanning Nuke, Dust2, Inferno, Mirage, Overpass, Ancient, Cache. Warmup/knife rounds excluded. This is the most-recent ~50 co-played matches — all that FaceIt still retains as downloadable demos.
  • Sides: computed per-event, so halftime side-swaps are handled automatically.
  • Honesty rule: where a number is noisy or a metric has a known limitation, we say so inline rather than letting it read as gospel.

Sample-size honesty

1,125 rounds is a statistically serious sample for aggregate and per-side (T/CT) metrics — opening-duel counts are in the hundreds, so those splits are trustworthy. Still thin: per-map deep-dives (the pipeline computes per-side, not yet per-map) and the rarest situations (4v5 retakes by bombsite). Per-map analysis arrives in a later edition. Note also that two prior small-sample numbers corrected when we went from 9 → 50 demos (retake win % and man-advantage conversion both came down) — a reminder of why sample size matters and why this book is built to be re-run.

Glossary — how every metric is defined

Precise definitions so nobody argues about what a number means:

Combat

  • K/D — kills ÷ deaths.
  • ADR — average damage per round, using capped damage (over-kill excluded, the way HLTV/Leetify count it).
  • KAST% — share of rounds where you got a Kill, Assist, Survived, or were Traded. Measures round-to-round contribution beyond fragging.
  • HS% — headshot kills ÷ kills.

Roles

  • Opening duel — the first kill of a round. Opening kill = you won it; opening death = you lost it (died first).
  • Opening → round-win % — of your opening kills, how many preceded your team winning the round. (Are your entries meaningful, not just early?)
  • AWP share — % of your kills made with the AWP.
  • Trade kill — a kill that avenges a teammate's death within 5 seconds.
  • Traded death — your death that a teammate then avenged. (High = you die in supportable positions — a good thing for connected play.)
  • Util damage — damage dealt with HE/molotov/incendiary (capped).
  • Enemy / team flashes — enemies vs. teammates you blinded. Team flashes are a discipline cost, not an achievement.
  • Flash assist — your flash directly set up a teammate's kill.

Conversion & spacing (team-level, our side)

  • Plant rate — of our T rounds, how often we get the bomb down.
  • Post-plant win % — of our plants, how many we then win.
  • Retake/hold win % — of rounds the enemy plants on us (CT), how many we still win.
  • Man-advantage conversion — of rounds we reach a numbers-advantage, how many we win.
  • Deaths while numbers-up — our deaths that happen while we're already ahead in numbers. Texture, not a verdict — a clean trade counts here too — but high volume flags over-aggression.
  • Bomb-carrier pickups — who picks up the bomb. (We watch that the entry isn't carrying it into first contact.)

A note on numbers and people

Stats describe tendencies over a sample; they do not define a player on any given night. We use them to find leverage — the smallest change that wins the most rounds — not to rank each other. Read accordingly.