EDITORIAL NOTE First 11th-edition data is live — the ELO ladder reset with it ›
EDITORIAL NOTE

First 11th-edition data is here — and the ELO ladder resets with it, every player back to a clean slate. The opening tournament results are in and the Strength Index is beginning to take shape across its signals, but a new edition takes time to read: expect weeks, likely months, before the numbers can support real conclusions. An early signal, not a verdict.

PREVIEW · PERFORMANCE PULL The Archive's first-party tournament-data pull. Editorially curated by the Editor before commit from public Best Coast Pairings event records; faction-anonymous outputs only.
PERFORMANCE · ARCHIVE PULL

PERFORMANCE PULL

FIRST-PARTY TOURNAMENT DATA — 5+ ROUND BCP EVENTS
THE ARCHIVE'S PULL 2026-W26 captured 2026-07-05T00:53:55.584Z

PER-FACTION AGGREGATE 28 FACTIONS · 3040 GAMES · 25 EVENTS · SORTABLE

FACTION GAMES WINS LOSSES DRAWS VP DIFF TOP-3 % WIN RATE REAL WR
Adepta Sororitas 42 21 21 0 -1.3 14.3% (1/7) 50%READ WITH CARE 48.7%
Adeptus Custodes 101 48 50 3 -0.6 9.1% (1/11) 47.5% 44.3%
Adeptus Mechanicus 40 16 23 1 -3.9 14.3% (1/7) 40%READ WITH CARE 44.2%
Aeldari 68 31 36 1 -4.1 0% (0/10) 45.6%READ WITH CARE 41.1%
Astra Militarum 92 40 51 1 -6.3 10% (1/10) 43.5% 37.6%
Black Templars 37 21 16 0 -1.1 16.7% (1/6) 56.8%READ WITH CARE 55.7%
Blood Angels 66 33 32 1 -0.8 0% (0/10) 50%READ WITH CARE 52.2%
Chaos Daemons 90 47 40 3 +2.7 11.1% (1/9) 52.2% 53.7%
Chaos Knights 41 14 25 2 -1.7 14.3% (1/7) 34.2%READ WITH CARE 37.8%
Chaos Space Marines 78 44 33 1 +0.3 14.3% (1/7) 56.4%READ WITH CARE 52.6%
Dark Angels 76 30 42 4 -2.1 10% (1/10) 39.5%READ WITH CARE 51.1%
Death Guard 81 30 48 3 -2 0% (0/11) 37% 44.1%
Deathwatch 10 4 6 0 -4.4 0% (0/2) 40%READ WITH CARE 48.2%
Drukhari 25 14 10 1 -3.4 25% (1/4) 56%READ WITH CARE 42.2%
Emperor’s Children 101 61 40 0 +3.7 33.3% (4/12) 60.4% 60%
Genestealer Cults 38 18 20 0 -1.7 14.3% (1/7) 47.4%READ WITH CARE 43.3%
Grey Knights 67 33 33 1 -0.2 0% (0/10) 49.3%READ WITH CARE 54.9%
Imperial Agents 10 4 4 2 -0.4 0% (0/2) 40%READ WITH CARE 50%
Imperial Knights 139 66 68 5 +1.4 8.3% (1/12) 47.5% 50.2%
Leagues of Votann 60 30 29 1 +0.9 25% (2/8) 50%READ WITH CARE 50.3%
Necrons 169 85 80 4 +1 33.3% (4/12) 50.3% 55.5%
Orks 85 54 30 1 +5.7 55.6% (5/9) 63.5% 61.1%
Space Marines 218 98 117 3 -0.6 30.8% (4/13) 45% 49.5%
Space Wolves 63 29 33 1 -2.7 0% (0/9) 46%READ WITH CARE 49.6%
T’au Empire 159 70 86 3 +0.1 50% (5/10) 44% 51.7%
Thousand Sons 90 53 37 0 +4.9 50% (5/10) 58.9% 59.7%
Tyranids 172 91 81 0 +0.7 23.1% (3/13) 52.9% 48.7%
World Eaters 85 40 44 1 -3.8 8.3% (1/12) 47.1% 41.3%

WEEKLY DIFF FACTIONS MOVED ≥ 2 POINTS VS. 2026-W25

UP-MOVERS

  • Emperor’s Children 37.5% → 60.4% · +22.9 pts games this week: 101
  • Orks 41.7% → 63.5% · +21.9 pts games this week: 85
  • Necrons 29.8% → 50.3% · +20.5 pts games this week: 169
  • World Eaters 36.8% → 47.1% · +10.2 pts games this week: 85
  • Death Guard 30% → 37% · +7 pts games this week: 81
  • Chaos Knights 30.4% → 34.2% · +3.7 pts games this week: 41
  • Drukhari 53.3% → 56% · +2.7 pts games this week: 25

DOWN-MOVERS

  • Imperial Agents 80% → 40% · -40 pts games this week: 10
  • Space Wolves 70% → 46% · -24 pts games this week: 63
  • Deathwatch 62.5% → 40% · -22.5 pts games this week: 10
  • Leagues of Votann 70% → 50% · -20 pts games this week: 60
  • Thousand Sons 76.5% → 58.9% · -17.6 pts games this week: 90
  • Dark Angels 54.2% → 39.5% · -14.7 pts games this week: 76
  • Aeldari 60% → 45.6% · -14.4 pts games this week: 68
  • T’au Empire 51.9% → 44% · -7.9 pts games this week: 159
  • Chaos Daemons 60% → 52.2% · -7.8 pts games this week: 90
  • Astra Militarum 50% → 43.5% · -6.5 pts games this week: 92
  • Tyranids 59.5% → 52.9% · -6.5 pts games this week: 172
  • Grey Knights 55.6% → 49.3% · -6.3 pts games this week: 67
  • Adeptus Mechanicus 45.8% → 40% · -5.8 pts games this week: 40
  • Blood Angels 55.6% → 50% · -5.6 pts games this week: 66
  • Adeptus Custodes 52.6% → 47.5% · -5.1 pts games this week: 101
  • Black Templars 59.3% → 56.8% · -2.5 pts games this week: 37

METHODOLOGY INCLUSION CRITERION · NORMALISATION · PRIVACY · TOS

METHODOLOGY

VIEW SELECTOR

The three view tabs at the top of this page swap between three temporal windows of the same underlying data. Weekly shows only the current week's snapshot. Current patch aggregates across all snapshots since the most recent dataslate / balance review. Edition lifetime aggregates everything in the current edition. Patch and edition boundaries are recorded in _data/balance_patch_timeline.json; when the Editor maintains that file, view boundaries shift automatically. For factions with stable win rates the three views converge; for factions whose win rates shift inside a patch cycle the divergence between views is itself informative.

INCLUSION FILTER

Every BCP event with 5 or more rounds during the named ISO week. No player-count floor. The 4-round Rogue Trader Tournaments are excluded by the round-count threshold — players testing list concepts at smaller events don't have those results pulled into competitive measurement, but small 5-round events that Stat Check (≥24 players) and Warpfriends (≥20 players) systematically exclude do contribute. The criterion is a single boolean: round count.

FACTION NORMALISATION

BCP records sometimes use abbreviated or detachment-named army strings (e.g., Space Marines (Astartes), Sisters of Battle, Tau Empire). The Archive resolves these to canonical faction names via _data/performance_alias_map.json at pull time. Adeptus Titanicus is excluded by convention — not played in standard 40k tournaments, omitted by every comparison source.

RESULT DERIVATION

For each faction: win_rate = wins / (wins + losses + draws). The result field on each game (a_wins / b_wins / draw) is taken from BCP's pairings Win: / Loss: / Draw: labels — never inferred from VP comparison. Per-game VPs are preserved in the matchups stream for downstream analysis (mission breakdowns, blowout-vs-grindy distinction).

PRIVACY POSTURE

Source text dumps containing player names are local-only on the Editor's machine (docs/operational/pulls/ is gitignored). The two committed JSONs — per-faction aggregate (Stream 1) and per-event pairings (Stream 2) — are faction-anonymous. No player-name field appears anywhere in the published data; a grep-based assertion in the parser's test suite enforces this.

BCP TERMS OF SERVICE

The Archive's pull operates under a manual-only, factual-citation posture documented in docs/operational/performance_tos_review.md. No automated scraping. No API access. No bulk reproduction of BCP's compilation product. Every cited number traces back to a publicly visible BCP event record; the audit list above gives the per-event provenance.

SELF-CITATION TRANSPARENCY

A site that cites its own data alongside two community aggregators carries an obvious circularity concern. Three mitigations make this transparent rather than circular: (1) the 5+ rounds inclusion rule is fixed in this methodology and never changes mid-snapshot; if it ever does, a source_changelog entry of kind methodology_revision announces the change with a date-stamped diff. (2) The per-event audit list above publishes every event that contributed — readers can re-verify any faction's number end-to-end. (3) Stat Check and Warpfriends pull from the same BCP firehose; if the Archive's per-faction win rate ever drifts more than ~2–3 points from both comparison sources simultaneously, that's an audit signal the Editor (or any reader) can interrogate.

For the broader design rationale — why the Archive is one of its own sources, how the cumulative-vs-weekly window choice plays out, what's in scope for Phase 7.1.0 vs deferred to 7.1.1 / 7.1.2 — see the archive_performance_pull_design.md design doc.

REAL WIN %

Win rate excluding draws from the denominator: wins / (wins + losses). The standard WIN RATE column treats draws as one third of the trinary outcome (they pull the rate toward 50%); Real WR strips them out so you see the win-vs-loss signal cleanly. Useful when a faction has a non-trivial draw rate that's compressing its standard WR.

VP DIFFERENTIAL

Mean (VPs scored − VPs allowed) per game. Captures how dominantly a faction wins or loses, not just whether. A faction at 50% WR with VP diff +12 is winning convincingly when it wins and losing tightly when it loses — a different profile from a faction at 50% WR with +0, where every game is a coin flip on VP terms too.

TOP-FINISH RATE (TOP-3)

Percentage of attended events where the faction had at least one player finish in the top 3. The metric is at the event level — a faction with two players in the top 3 of a single event counts once. Useful editorial framing: "how often does this faction cash" independent of overall WR. The small-N chip fires when the faction has played fewer than 2 events in the snapshot window.

Events with no published standings (pairings-only pulls, where the source page only exposed the round-by-round pairings tab) are excluded from the top-finish denominator — the faction's games from such events still contribute to overall games / WR / VP differential, but the parser can't determine top-3 placement without standings positions. OverRep computes the same way: such events don't contribute top-cut share either.

MARGIN PROFILE

Four metrics surfacing the shape of wins and losses:

Together these distinguish "grinds out close wins" (low blowout %, high closeness %) from "ten-trick pony" (high blowout %, low closeness %, gives a lot when it loses).

REPRESENTATION + OVERREP

Pop share is the faction's share of all games played in the snapshot window (one game contributes one observation to each of the two factions involved). Top-cut share is the faction's share of all top-3 finishes across all events. OverRep is the ratio: top-cut share / pop share. > 1.0 means the faction punches above its play rate (over-performs); < 1.0 means it under-performs its popularity. Adapted from Stat Check's OverRep column with the Archive's own data — the methodological choice (top-3 cut, single-event-counted) is locked here so the calculation is auditable.

CONSISTENCY PROFILE

Cross-event WR σ is the standard deviation of the faction's per-event win rate across all events where it played at least 5 games. Low σ = consistent across events; high σ = swingy. VP score σ is the standard deviation of VPs scored across all games — same idea applied at the per-game level. Mirror match WR is the win rate in games where both sides chose this faction; symmetric at scale (every mirror game contributes one win + one loss to the faction tally on the A-side convention), but small samples can drift.

ROUND ATTRITION

Per-round win rate by faction. The interpretive frame: Round 1 in a Swiss tournament is mostly random pairings (no one's accumulated W/L records yet), so a faction's Round 1 WR mostly reflects performance against the general player pool. Final rounds reflect top-table competition — the faction is now facing the strongest other lists in the event. Wide positive Δ R1→final = the faction over-performs late (gets stronger as opponents get tougher). Wide negative Δ = the faction over-performs early and folds at top tables.

WIN RATE × VP DIFFERENTIAL (SCATTER)

Each faction renders as a dot at (VP differential, win rate). The chart's purpose is to surface the SHAPE of a faction's performance, not just its rank. Two factions can both sit at 52% WR; the one with VP diff +10 wins decisively, the one at +0 wins coin-flips. Both can fairly be called "above average," but they're different faction-shapes. Dots are violet (in-scope) for factions with at least 80 games observed in the snapshot; small-N factions are muted. Quadrant labels in the corners describe the four faction-shape archetypes the plane separates.

MATCHUP MATRIX

Each cell is the row faction's win rate against the column faction across all 5+ round events in this snapshot. Cells are coloured by the row faction's perspective: violet means the row faction tends to beat the column faction; gold means the row faction tends to lose to it. Cells with fewer than 5 head-to-head games are muted — the signal is too thin to read into. The diagonal shows mirror matches, where one player chose the same faction as their opponent; mirror cells use a hash pattern and surface the game count rather than a structurally-50% WR.

The matrix is structurally symmetric: a cell at row Necrons column T'au showing 54% means Necrons won 54% of its games against T'au; the mirror cell at row T'au column Necrons shows the same data from T'au's perspective (so ~46%, less any draws). The visual redundancy is intentional — readers tend to think from the faction-they-play perspective, not the opponent's.

AVG MIRROR MARGIN

Mirror match margin: when two players bring the same faction, how decisive is the game on average? Low margin (e.g. ±10 VPs) suggests the faction is list-balanced — winners and losers separated only by piloting choices. High margin (e.g. ±35 VPs) suggests list construction or matchup-specific skill swing the outcome decisively. Computed only when at least 2 mirror games are observed; otherwise displayed as —. Replaces the structurally-50% “Mirror WR” metric from the Phase 7.1.1.0 build — every mirror game by definition produces one winner and one loser of the same faction, so any computed mirror WR sits at 50% by construction and carries no editorial signal.