The first dispatch built on the Archive’s own numbers. Treat what follows as a starting position, not a trend — and as a working demonstration of why the window you measure decides the meta you see.
A note before the numbers. This is the first week the Archive has run its own performance pull, the first ingest of three outside aggregators, the first attention snapshot, and the first pair of practice tier lists gathered into one place. Everything here is a first capture. There is no “last week” to compare against, so there are no movers, no deltas, no trends — only a baseline. That is worth saying twice, because the instinct with a fresh table of win rates is to read a story into it. The honest story this week is the table itself.
The week on the table
The Archive’s Week 21 pull covers six events — Alamo GT, the Alberta Classic, Conquest Midseason, Las Crónicas Chilangas, Mayhem 5, and ObSec War Calls — and twenty-eight factions across roughly 2,850 games. The inclusion rule is simple and public: every event of five or more rounds, no player-count floor, compiled by hand from Best Coast Pairings records.
One faction is having a different week from everyone else. Chaos Space Marines sit at 59.0% across a healthy 442-game sample — more than two points clear of the next faction and a full ten points above the field average of 48.8%. Behind them the upper tier is recognisable: Deathwatch at 56.7%, T’au Empire at 55.0%, Necrons at 53.2%, Drukhari at 52.5%. Genestealer Cults and Imperial Agents both post high rates as well, but on 47 and 31 games respectively — far below the 80-game line where these numbers begin to mean much, so set them aside.
The more important shape is the middle. Seventeen of the twenty-eight factions fall inside a single eight-point band, 45% to 53% — the bulk of the roster clustered around even. The bottom is as clear as the top: Space Wolves (38.9%), Blood Angels (39.5%), Aeldari (41.0%) and vanilla Space Marines (41.4%, and on the largest sample in the set at 382 games) are the week’s strugglers. This is a meta with sharp edges and a soft, crowded middle.
Cross-source pulse — mind the window
The Archive is not the only group counting, and the others do not agree with it — or with each other. That sounds alarming and mostly is not. The four sources are measuring different things.
The Archive’s pull is a single ISO week. Stat Check’s is a four-week cumulative reaching back to the Defiler dataslate. Warpfriends’ is last week’s post, cumulative across the whole dataslate. 40kstats is edition-cumulative — around 280,000 games stretching back toward the start of the edition. Lay the same faction across all four and the disagreement is really a disagreement about time.
Necrons show what agreement looks like: 51% to 54.5% across every window, a spread of three and a half points — whatever the question, the answer is “comfortably above even.” Chaos Space Marines show the opposite. The edition-long number is 46.9%; Warpfriends’ dataslate window has them at 52.8%; Stat Check’s four-week window at 57%; the Archive’s single week at 59%. That is not four sources contradicting each other — it is one faction whose recent form runs well ahead of its edition average, with each shorter window catching more of the climb. Whether the 59% holds is a question only next week can answer.
And then there is Thousand Sons, which is the genuinely confusing one. Stat Check has them at 62% — a top-of-the-meta number. The Archive’s week has them at 47.8%. That is a fourteen-point gap on the same faction, and it is too large to wave away with “different windows.” It might be a faction that has cooled sharply since the early dataslate; it might be sampling; it might be both. The correct response to a spread that wide is not to pick the number you prefer — it is to hold low confidence and wait. The Archive’s own methodology writing has a name for this: when independent samples diverge this far, the divergence is the finding.
Anecdotes from the noise
A few things stood out this week. All of them carry the same asterisk — one week, first capture — so read them as observations, not verdicts.
The Defiler dividend, unevenly paid. The current balance dataslate is, on the Archive’s timeline, the “Defiler legalization” patch, and the Defiler is a chassis five factions can field — Chaos Space Marines, Thousand Sons, World Eaters, Death Guard and Emperor’s Children. So the obvious read is “Chaos rose.” The obvious read is also incomplete. All five drew something from the datasheet change, but the table says they did not draw it equally: Chaos Space Marines are at 59.0%, while Emperor’s Children sit at 51.4%, World Eaters at 50.0%, Death Guard at 49.9% and Thousand Sons at 47.8% — strung right across the field average. The same unit, folded into five different army rules and detachment structures, pays five different dividends. A shared buff is not a shared outcome, and the spread across the Defiler factions is a better story than the CSM number alone. The usual asterisk still applies — one week, one source.
Deathwatch, quietly everywhere. The week’s most consistent overperformer is not the headline faction. Deathwatch land in the top three of all four sources — 54% to 63% depending on window — and both practice tier lists rate them A or better. The samples are moderate rather than huge, but four independent reads pointing the same direction is the strongest signal the data offers this week.
T’au win without the noise. T’au Empire post 55% on the table and sit near the very bottom of community search interest — only Genestealer Cults rank lower among competitively-active factions. They are the quiet competence of the current meta: winning steadily while the conversation happens elsewhere.
Custodes are sliding. In early March, the Veizla tier list placed Adeptus Custodes in its top bracket. The Breaking Heads list from this month drops them to the lower-middle, and the table agrees — 45.6%, the wrong side of the field. Across the span of the dataslate this is the clearest fall the practice data records.
All eyes, no wins. Space Marines drew more than eight million YouTube views inside the attention window — comfortably the most-watched faction — and posted 41.4% on the table across the largest sample in the pull. Attention is not performance, and Space Marines are this week’s proof.
Attention — handle with tongs
The Week 20 attention snapshot ran cleanly: search interest and video activity for all twenty-nine factions, no source failures. It is also, this week, a low-resolution instrument, and I would rather say so than dress it up. Google Trends interest is bunched near the top of its scale for most of the roster, and the YouTube video count — one of the two inputs — is capped at fifty for twenty-six of twenty-nine factions, which means it barely separates anyone. Treat this week’s attention reading as mood music, not measurement. The one clear note in it is that Chaos has the room’s attention: Emperor’s Children and Chaos Space Marines trail only the perennially-massive Space Marines in video views. And with a single snapshot on file there is, again, no trend to read — only a first mark on the wall.
Practice — two voices, one of them old
The practice layer carries two expert tier lists so far, against a locked set of five sources, so it is early here too. The recent one is Breaking Heads’ list from 15 May, framed explicitly by its authors as the final 10th-edition tier list before the edition turns. It puts four factions in its top bracket: Chaos Space Marines, Astra Militarum, Thousand Sons and Emperor’s Children. The second list, from Veizla, is from early March and now carries little weight by design — the practice adapter decays old observations, and a March list is mostly faded.
Two things in the Breaking Heads list are worth a moment. It rates Astra Militarum top-bracket while the table has the Guard at 46.9% — a real theory-versus-table gap, the kind the Archive exists to surface rather than resolve. And it leaves Grey Knights in the lower-middle, which the 45.4% win rate agrees with — consistent with the durability problem this column’s recent cover-rules piece walked through.
The digest
Goonhammer’s Competitive Innovations continues its weekly cadence, the latest instalment — “The Island of Dr. Filer, pt.1” — posted 20 May. Even the recap titles cannot stop punning on the Defiler. It remains the most reliable weekly read on the competitive scene, and the Archive links to it rather than reproducing it.
How to read this week
Everything above rests on first captures. There are no week-over-week movers, because there is no prior week on file. The four performance sources measure four different windows of time and should not be averaged casually. Two faction rates — Imperial Agents and Genestealer Cults — sit on samples too small to trust. The attention signal is compressed. The practice layer is two lists deep, one of them stale. None of that makes the week unreadable; it makes it a baseline, and the value of a baseline is everything that gets measured against it.
Editor’s note
This is the issue the Archive was built to produce, and it is deliberately modest. We have a starting line and very little else — and the temptation in this hobby is always to narrate the starting line as if it were the race. Next week brings the first deltas; the week after, the first hint of a trend. Until then, the most useful thing the data says is “here is where everyone stood.”
Data is infinite, understanding it is Divine — and understanding takes more than one week.
— The Editor