Measured Differently: The Archive Enters Eleventh Edition

A new edition resets the board. Every datasheet, every points value, every habit of the last three years is up for renegotiation, and for a few months nobody — not the strongest players, not the loudest channels, not us — actually knows what the game is yet. That uncertainty is uncomfortable. It is also the most honest moment a competitive-analysis site ever gets, and we used it to rebuild the Archive around a single commitment: measure what the tournament tables actually did, and be visible about how much we still don’t know.

Here is what changed, and why.

We retired three of our four signals

The old Archive blended four inputs into one number — tournament performance, a datasheet-strength “theory” model, top-player tier lists, and an “attention” signal drawn from search and video trends. It was clever. It was also, on reflection, dishonest about what it was doing: it fused measurement (what happened at events) with opinion (what a model or a crowd expected to happen) and reported the blend as if the whole thing were measured.

A new edition made the problem impossible to ignore. There is no settled datasheet-strength model for Eleventh yet; attention data in the first weeks is mostly noise about which factions got new models, not which ones win. Folding those into a strength score wouldn’t have added signal. It would have added confidence we hadn’t earned.

So we cut them. The Archive is now performance-only. Every ranking on the site traces back to first-party tournament results — games played, at real events, with real stakes — and nothing else.

What replaced them: four facets of one honest signal

Dropping three signals didn’t mean dropping nuance. Win rate alone is a blunt instrument, so the Strength Index — the tier list — now reads a faction through four different questions asked of the same tournament data:

• Win rate — did it win games (adjusted for sample size, draws set aside)?
• Top-cut conversion — did it convert into the top finishes, or just pad the middle of the field?
• Blowout margin — when it won, did it win hard, and when it lost, how badly?
• Consistency — was it steady event to event, or a coin flip that occasionally spiked?

You weight those four however you like — the sliders are yours — and the composite re-ranks live. There is no single blessed answer, because there isn’t one: a faction that wins narrowly and often is a different proposition from one that either blows you out or bombs, and the Index lets you decide which you care about.

We show the uncertainty instead of hiding it

This is the part that matters most, and the part most tier lists skip.

Every win rate on the Archive carries its confidence interval — the honest band the true number probably sits in, wide when the sample is thin and narrow when it’s deep. A faction with twenty games and a flashy record gets a visibly wide band and a read-with-care flag, not a podium. And a faction below a minimum sample floor simply cannot be banded strong or weak at all, no matter how good or bad its handful of games looks — it sits in the middle, its number shown, its verdict withheld until the games arrive to support one.

We would rather tell you “we don’t know yet” than crown someone on a lucky weekend. In the opening months of an edition, “we don’t know yet” is frequently the truest thing on the page, and we’ve built the site to say it out loud.

The tour, briefly

• The Tier List (Strength Index) — the four-facet composite, with visible uncertainty and weights you control.
• Quantum, the data dashboard — the measurement layer underneath the tier list: per-faction win rates with confidence intervals, the full faction-versus-faction matchup matrix, scoring-posture (disposition) data, per-faction detachment breakdowns, and player ELO ratings. Slice any of it by the current week, the current balance window, or the whole edition.
• Dossiers — a performance scouting report per faction: its facet readout, its best-performing lists this window, recent form, and how its results compare to where top players rank it. Where those two disagree is usually the story.

All of it is first-party — our own aggregation of public tournament data, curated before it’s published — and all of it is faction-anonymous. We report what armies did, not who piloted them. We don’t republish anyone else’s win-rate tables; we build our own and show our working.

Measurement and interpretation are different acts

One last thing, because it’s the spine of the whole project. The dashboard measures — that’s the Archivist, and it has no opinions to sell you. What you’re reading right now interprets — that’s the Editor, and interpretation is where judgment, and the risk of being wrong, live. We keep the two visibly separate so you can trust the first and weigh the second, exactly as much as you decide it’s worth.

Eleventh Edition is young. The numbers are thin and the bands are wide, and for a while yet the most useful thing the Archive can do is tell you precisely how uncertain the picture is. That’s not a limitation we’re apologizing for. It’s the point.

— The Editor

Data is infinite; understanding it is Divine.