T-524 round 2, for T-523 (Jordan, 1 Oct (Pacific), via the design session). A report, not the site: nothing here changes a published score or page. The numbers come from node tools/tiertest.js on the two panel lists below; the full working is in each army's report: Space Marines and Orks.
The question, and the short answer
Round 1 found that none of our systems agreed well with a public tier list, and that V2 agreed least. Round 2 asked two things: does a fairer yardstick change that, and does V2b (Hammer counted as damage delivered before death) fix the over-rated slow elites?
The fairer yardstick: partly. It changes the size of the gap, not the order of the systems. Comparing within an army (rank order, and letters re-cut inside the army) removes the cross-army scale problem. But on Space Marines every system still ranks units in the opposite order to the list.
V2b: no. On both panels and on list share it comes last or tied last. It makes some units better and others worse in about equal measure:
- slow elites drop (Breaka Boyz 87 → 54);
- transports' passengers and walkers rise too far (Gunwagon to S).
Best overall: V1 with its fitted weights. It's first or close to first on every yardstick: the best fit to what players take (ρ 0.174), joint best within the Space Marine army, and best on Orks' global letters. Today's letter and the plain maths come close behind it. V2 and V2b come last.
The panel
| Army | Source | What it is | Units |
|---|---|---|---|
| Space Marines | Tista Minis, "Best Space Marine Units Right Now" (22 Jul 2026) | A ranked list with a reason for each unit | 30 |
| Orks | Sprues & Brews, "Codex Orks Review (2026)" (22 Aug 2026) | A codex review with no tiers: we graded each verdict by how strong it is. Written before points, against the old rules, so worth about half a ranked list | 45 |
Both are in reports/tiers/panel/: names, our letter and our own paraphrase only. The brief asked for two more lists, from other armies and other authors. Only the Orks one could be read. The other candidates either answer a machine with a bot check, or are videos whose words need a sign-in:
- Tabletop Battles' 11th-edition Necron and Death Guard breakdowns, Goonhammer, kutami.de and warhammerguild.com (bot checks);
- Auspex Tactics' Custodes top ten and Art of War's Death Guard datasheets (videos).
Two lists are the floor T-533 needs, not the panel it describes (three to five).
The numbers
(a) Spearman's ρ between our score and the list's order within the army. Higher is better; 1 would be the same order.
| System | Space Marines | Orks | Mean |
|---|---|---|---|
| Today's letter | −0.31 | 0.24 | −0.04 |
| Maths only (Value) | −0.31 | 0.25 | −0.03 |
| V1, fitted | −0.27 | 0.22 | −0.03 |
| V1, equal weights | −0.27 | 0.16 | −0.06 |
| V2 | −0.23 | 0.12 | −0.06 |
| V2b | −0.24 | 0.06 | −0.09 |
(b) Mean distance in tiers between our letter and theirs. Lower is better. "Army" re-cuts our letters within the army by the same quotas; "global" is our published scale.
| System | Space Marines, army | Space Marines, global | Orks, army | Orks, global |
|---|---|---|---|---|
| Today's letter | 1.47 | 1.40 | 1.09 | 1.04 |
| Maths only (Value) | 1.47 | 1.43 | 1.16 | 0.93 |
| V1, fitted | 1.43 | 1.43 | 1.09 | 0.91 |
| V1, equal weights | 1.43 | 1.47 | 1.16 | 1.09 |
| V2 | 1.57 | 1.67 | 1.18 | 1.27 |
| V2b | 1.57 | 1.63 | 1.20 | 1.29 |
(c) ρ with each unit's share of the winning lists, every non-Legends unit: Value 0.113, V1 fitted 0.174, V1 equal 0.153, V2 0.127 (with T-527's Banner), V2b 0.087.
Mortarion, and Ghazghkull the other way
| System | Mortarion, global | within Death Guard | rank | Ghazghkull, global | within Orks | rank |
|---|---|---|---|---|---|---|
| Today's letter | S | A | 7 of 36 | S | S | 2 of 55 |
| Maths only | B | B | 18 of 36 | B | B | 18 of 55 |
| V1, fitted | B | A | 7 of 36 | C | C | 37 of 55 |
| V1, equal weights | A | B | 9 of 36 | C | C | 37 of 55 |
| V2 | B | D | 30 of 36 | A | A | 11 of 55 |
| V2b | C | C | 28 of 36 | A | A | 10 of 55 |
They're the two most-taken centrepieces of their armies: Mortarion is in 10 of 16 winning Death Guard lists, Ghazghkull in 15 of 21 winning Ork lists. No new system gets both right:
- V1 keeps Mortarion near the top of his army and drops Ghazghkull to C.
- V2 does the opposite: Ghazghkull's Hammer is 90, Mortarion's isn't.
Only today's letter, lifted by the meta, rates both highly, and that's the winning lists talking, not our maths.
What we learned
- The panels disagree with each other: the Space Marines list runs against us, the Orks review a little with us. With two lists, the sign of the answer depends on which army you pick, so no single number from this round should decide the redesign. T-533's standing panel needs three to five lists before its score means much.
- The same misses come back on both panels:
- Specialists diluted by averaging over targets. Eradicators, Infernus, Tankbustas and Breaka Boyz rate high on their best target and middling on the average (the archetypes, T-529; the lenses, T-531).
- What a character gives its squad beyond dice. Examples are Deep Strike for Nazdreg's Meganobz, and the Space Marine leaders (T-527 counts the dice; T-532 takes Supporting links into the live score).
- Close-range weapons that our neutral table puts beyond half range.
- V2b's delivery model is too blunt to keep as it stands. One transport assumption (Infantry rides 10" on turns 1 and 2) lifts whole armies, and slow melee units lose more than the lists say they should. Damage before death is a good idea; it needs measured movement (the north-star note's step 5) before it can carry a letter.
- Runner promotes cheap characters. V2's Runner percentile gives S to leaders that always join a squad (Chaplains are in 0 of 84 lists). This is the main reason V2 trails V1.
- The meta lift does what the maths can't. Today's letter is never the worst system, and on the centrepieces it's the only one that's right. Whatever replaces it should keep a meta signal, as the crown and its one-word read (T-523) already plan.
Where this leaves T-523 (for Jordan)
The design session's brief: stop here for Jordan. If the redesign goes ahead, the evidence points to:
- V1's composite and quotas, with the four job names. Runner stays folded into Banner until it stops promoting joined leaders.
- No V2b delivery model yet.
- The panel grown to three to five lists before any letter on the site moves. Easiest route: someone with a normal browser saves the text of a Tabletop Battles or Goonhammer unit review.
Nothing here changes the site.