Paste a list — yours or an opponent's — and you get a structured read of it in seconds: what kills things, what holds ground, what moves, what breaks first, and where the list has nothing at all. No dice, no opinions, no forum consensus. Just the maths that is already sitting in your list, made visible.
On your own list
Find the hole before your opponent does. Most lists have one — a phase they cannot contest, a target profile they cannot hurt, a scoring plan that depends on one unit surviving turn two. The Stat Lab names it.
On an opponent's list
Read the game before it starts. What are its priority threats, when do they arrive, what anchors it, and what happens if you remove one piece.
Computed numbers versus AI-written text
Purple card means it was written by a language model. Everything else is computed.
The Stat Lab uses language models for tactical breakdowns and short summaries, and marks them clearly so you always know which is which. AI-written text can still be wrong, awkward, or miss the point. It is a summary of the analysis, not an authority — the numbers above it are the important bit.
Detachment colours
Detachments carry a colour. The colour tells you the detachment's Force Disposition — what kind of army it is built to be. It is a category, not a rating.
- DisruptionDeny, redirect, dictate where the game happens.
- Take and HoldOccupy ground and refuse to be moved off it.
- Purge the FoeKill things; the plan is attrition.
- Priority AssetsControl specific high-value points.
- ReconnaissanceSpeed, coverage, board presence.
- Not recordedNo disposition detected on the roster. Never inferred.
Reading a partial analysis
Sometimes a list will not fully resolve — an unusual export format, a unit we cannot match, a datasheet newer than our data. When that happens we tell you, on the result, rather than quietly scoring a partial list as if it were complete.
You will see a data confidence figure showing how much of the list we recognised, and anything we could not read is listed by name. Units we could not resolve are excluded from the maths rather than guessed at, and any part of the score we could not measure is left neutral — missing data never helps or hurts your score. It just makes the result less complete, and we show you how much less.
If confidence is low, treat the score as provisional and check what did not resolve first.