Tier system v2, tested: Orks (2026-10-02)

T-524, for T-523 (Jordan, 1 Oct (Pacific), via the design session). Made by node tools/tiertest.js. A report, not the site: nothing here changes a published score or page.

In short

The second panel list for T-524's round 2 (T-533's panel), found and read on 2 Oct.

The list. Sprues & Brews' "Codex Orks Review (2026)" (Matt Crowther, 22 Aug 2026), an 11th-edition codex review written from a review copy. It has no tiers, so we graded each verdict by how strong it is: the reviewer's strongest praise is S, clear praise A, measured praise B, mixed C, doubtful D. That gives 45 units: S 3, A 13, B 25, C 3, D 1. Nine units the review only describes are left out. We keep names, our letter and our own paraphrase only (reports/tiers/panel/2026-10-02-orks-spruesandbrews.json).

Weaker than a tier list. It was written before points were out and mostly compares each unit with its old rules. So "better than before" counts as praise even for a unit nobody takes. Read it as one reviewer's impression, at about half the weight of a ranked list.

Every system agrees a little, and the order matches Space Marines. Today's letter, the maths and V1 come out on top; V2 and V2b come last.

Where they're higher than us:

• Nazdreg (their S, our C or D) and Meganobz (their A, our D). Nazdreg leads Meganobz, and his value is what he gives them: Deep Strike, Ignores Cover and Riled Up. The Banner method (T-527) counts his dice effects on the squad, but not the Deep Strike, which changes when the squad arrives. Meganobz are "very hard to crack", but their durability per point doesn't reach an Anvil letter.
• Tankbustas and Breaka Boyz (their A, our C or D). Anti-vehicle units. Our Hammer averages over every target, so a specialist's best matchup is diluted (the archetype lenses, T-529 and T-531).
• Every aircraft (their B, our D). The review praises them against their old rules, not against the field.

Where we're higher:

• Gunwagon (V2b S, their B). V2b gives it the full transport delivery.
• Beastboss on Squigosaur (V2 S, their B).
• Deffkilla Wartrike (Runner A, their B).

Ghazghkull goes the other way to Mortarion: V2 and V2b rate him A (his Hammer is 90), while V1 drops him to C. Today's letter, lifted by the meta, has him at S. He's in 15 of the 21 winning Ork lists, so here V2's peak-first composite reads him better than V1's.

Round 2 (2 Oct): a fairer yardstick, V2b, and Ghazghkull Thraka

V2b is V2 with Hammer counting damage delivered before death, plus a detachment credit in Banner. Each turn of a five-round game on a 44" × 60" table, the model asks whether the unit reaches the enemy with its guns or a charge:

• Start: 30" between the units. Infiltrators start 9" away, Scouts move 6" before the game, and Deep Strike arrives on turn 2 at 9.5".
• Movement: each turn the unit moves its Move (+2" with Fly; Infantry in an army with Transports rides 10" on turns 1 and 2), and the enemy closes 3". A unit still out of range advances 3.5". A gun unit stops at its longest range, or at half range when that's worth over 15% more (melta, Rapid Fire). A melee unit keeps coming.
• Damage: its ranged damage when in range (the half-range figure inside half range), its melee damage times the chance of making the charge (2D6 against the gap left), and Overwatch on 33% of enemy turns from turn 2, hitting on 6s (a quarter of a 3+ hitter's damage), or in full with Torrent.
• Survival: each turn is weighed by the chance the unit is still alive (V2's turns alive under typical fire).
• Detachment credit: a character carrying the keyword a detachment's effects need (Librarians for a PSYKER detachment) shares that detachment's best lift (rankings.json's detachment synergy) with the army's other characters carrying it, on Leader-lift's scale (up to 1.5 points). It goes to armies where 6 or fewer characters carry the keyword.

Over every unit, V2b's ρ with list share is 0.087, against V2's 0.127 and V1's 0.174.

The yardstick, three ways, on the 45 units of the public list. (a) Spearman's ρ between our score and their order within the army (higher is better). (b) Our letters re-cut by quota within Orks's 55 units, against our mapping of their sections (mean distance; lower is better). (c) Our global letters, as in round 1.

V2 to V2b on the list's units (their letter, then our numbers):

Ghazghkull Thraka under every system (Orks, 300 pts, in 15 of 21 winning lists):

V2b's Hammer for Ghazghkull Thraka: 90 → 90.

Round 3 (2 Oct): delivery modes, the bundle, and five fixes

Jordan, 1 Oct (Pacific), via the design session: the bundle rule (05:28Z), delivery modes (05:38Z) and the round-3 fixes (05:44Z). V3 is V2's four jobs with Hammer as damage delivered before death, scored under every way the unit may legally reach the fight; it keeps the best and names it:

• Walk: on the table from turn 1, 30" from the enemy. Scouts move 6" first. Infiltrators start 12" away: more than 8" from the enemy's deployment zone, with the enemy a few inches inside it.
• Deep Strike: off the table on turn 1, arrives on turn 2 9" away (the 11th-edition rule is more than 8"). It shoots at once if its guns reach. A charge from there needs 9 on 2D6 (28%, no re-roll assumed).
• Strategic Reserves: off on turn 1, arrives on turn 2 within 6" of a table edge and more than 8" from the enemy (11th-edition core rules), taken as 18" from the enemy. The 50% points cap on reserves is a list-level limit; we note it but don't enforce it per unit.
• Transport (the bundle): the unit plus its best legal Dedicated Transport, priced at both together (characters aren't bundled: they ride with the squad they join). The pair moves at the transport's Move on turn 1 and the unit is out on turn 2. The unit can't be shot for 1 round inside. The transport's own guns add to Hammer, and its hits to destroy and OC to Anvil, all per point of the pair. If it doesn't beat the unit alone, the transport isn't part of the unit's story.

The same rules as V2b after that: the enemy closes 3" a turn; a unit out of range advances 3.5"; guns stop at their range, or at half range when that's worth over 15% more; melee charges on 2D6; Overwatch 33% of enemy turns. Only turns the unit can be shot use up its life: a unit off the table or inside a transport keeps it. No terrain yet: every shot is taken to have a line (the exposure factor is 1, to be calibrated against the panel).

Who may ride what comes from BSData's transport entries at run time (tools/transports.js): the capacity, the keywords a passenger needs, the models it can't carry, and the ones that take two or three spaces. Only Dedicated Transports form bundles. With every transport allowed, Scouts chose a Land Raider Crusader and Boyz a Gunwagon, inheriting the big tank's guns. Those are units in their own right, scored alone. Allies are skipped for now.

The fixes, each switchable: (a) mortal wounds a round from abilities, our estimates read on 11th-edition datasheets (34 units, reports/tiers/2026-10-02-mortals.json); (b) a character that joins a squad (Leader or Support) and a unit with no OC can't be a Runner; (c) Battle-shock bites only below half strength, so Anvil's Leadership factor is 1 − 0.35 × (1 − the pass chance); (d) a transport's Banner counts its capacity ÷ 6, at most 2; (e) Indirect Fire is in range from turn 1. (d) lowered every yardstick, so V3 leaves it out; its row shows what it costs.

Each term on its own, against V3 with every term off (V2b's single path: ρ 0.107):

For comparison: V1 fitted 0.174, V2 0.127, V2b 0.087.

How units arrive under V3: walk 579, Deep Strike 220, Strategic Reserves 233, in a transport 129 (of 1161).

The 20 biggest gains from a transport (the unit page's line would read "Best delivered by …"):

Across the 129 bundles, the transport's own guns are 57% of the pair's damage at the median. For a passenger that does little itself (Krieg Combat Engineers, Pyrovores), the bundle's gain is mostly the transport. That is the rule as set ("when you pay for the transport you also get its guns"), but it means a weak passenger's letter partly measures its ride.

Round 1's over- and under-rated groups, and the arrival units (the public list rated the first two high and the Centurions and Terminators lower than we did):

Turn-by-turn profiles under the mode V3 chose: the distance to the enemy after moving, whether its guns reach (R), the charge chance (c), and the chance it's still alive. A dash means off the table or inside a transport.

Do the arrival-mode units move toward the public list? 19 of its 45 units arrive other than on foot under V3; their mean distance from the list's letters is 1.47 tiers under V2b and 1.32 under V3 (lower is closer).

The systems

• Current: today's letter. Value, lifted by the meta: a unit in a share s of its army's winning lists scores at least 40 + 100 × s (with 3 lists' doubt). Bands 80 / 65 / 50 / 35.
• Maths only: Value alone, the same bands.
• V1 (three domains), the first proposal: the three jobs below, composite = 0.75 × best + 0.25 × mean, letters by quota on the composite's rank (S 7%, A 15%, B 28%, C 30%, D 20%), S and A split across the jobs by their share of the winning lists' points. Banner's weights fitted (below).
• V2 (four jobs), from how a game is won (primary: OC on objectives when they score; secondary: actions and positions, which a Battle-shocked unit can't start or finish; denial: kills and screens). Hammer, Anvil, Runner and Banner, each a percentile over every non-Legends unit; the same composite (0.75 × best + 0.25 × mean of four) and quotas, S and A split by the four jobs' share of the winning lists' points.

The jobs' names are Jordan's (2 Oct) and our own words, not Games Workshop keywords: Hammer (hitting), Anvil (holding), Runner (running the secondaries, V2 only) and Banner (enabling others). A unit's tag reads like "S · Hammer" or "A · Banner".

Offense and Defense stay the published raw pair under either version (Jordan, 2 Oct: "quite solid math… more data is better"): every unit page and Rankings row keeps them with their percentile chips; the jobs are built from them and never replace them. V2 would add damage before death as a third shown number.

V2's jobs, as built here. Hammer: damage before death, the wounds it removes per turn per 100 pts (Offense's rule, the better of shooting and melee into each target, best loadout per target) × the turns it lives under typical fire (each attacker type's hits to destroy it over half the median unit's, so the median unit lives 2 turns; at least 1, at most 5). Durability counts here and in Anvil on purpose. Anvil: Defense per 100 pts × OC per 100 pts × the chance to pass a Battle-shock test (2D6 at its Leadership) × 1.15 where it heals or returns models (a returned model or a few wounds a turn restores roughly a tenth to a fifth of a unit over a game, so a modest lift). Runner: the mean of four percentiles: reach (Move, Fly, Deep Strike, Scouts, Infiltrators), units per 100 pts, a small footprint (models × a base size by type: infantry 1, mounted 2, walkers 3, light vehicles 4, monsters and battle tanks 5, aircraft 6, Titanic 12; swarms double) and Leadership. Strategic Reserves are open to nearly every unit, so they don't separate units and aren't counted. Banner: what it lends others per 100 pts: a Leader's lift on its best squad (T-062's measure), auras (up to 2), Transport, revives. Transport capacity and CP economy aren't in our data yet (BSData gives capacity as text), so a transport counts 1. V2's parts are equally weighted; nothing is fitted. Banner here uses T-527's scores for the 451 characters it scores (what each adds to its best squad per point against more bodies, tools/banner.js), mapped onto those characters' places so the order changes and their standing against other units doesn't.

V1's Anvil, as built here: the hits the unit takes to destroy per 100 pts (Defense's raw number, each of the five attacker types) × its Objective Control (OC × models), averaged over the attackers, then ranked. A product: a unit holds only if it is both hard to shift and counts on the objective (a fast tank with OC 0 or a fragile horde both score low). Ranked, a product orders units exactly as the geometric mean of the two would, so there's no weight to tune. Plain Defense stays a shown number. V1's Banner is today's four Utility parts; note board control (OC per 100 pts) appears in both Anvil and Banner, which the fit is free to weigh down.

Pool: 1161 rows (every non-Legends unit; a datasheet shared by several armies is a row in each, as on the tier list).

The job split (decides each job's S and A)

Each unit's job is its best domain; its share of the winning lists is the lists that take it (top-quarter lists, July to September, grimstat-corpus, CC BY 4.0) × its points at the smallest size. A unit taken twice in a list counts once (the lists we keep say which units, not how many).

With equal Banner weights the split is 34% / 42% / 23%.

V2, four jobs:

Banner's weights, fitted on list share

Every split of 1 into tenths over the four parts (286 in all), scored by how well the composite ranks 1105 units like their share of the winning lists that could field them (Spearman's ρ; units some 5 or more lists could field).

The fitted weights lift the composite's agreement with list share from 0.153 (equal weights) to 0.174; without Banner it is 0.126, and today's Value 0.113. Every one of these is weak: our numbers explain little of what players take, and the gain from fitting is small next to what's missing. The fit was made on list share only, so a community list can test it without circularity.

The meta-lifted S units with Value in the 60s

44 units wear S today only because the meta lifts them, with a Value in the 60s. Jordan's example, Mortarion, is among them.

Under V1: S 6, A 9, B 15, C 13, D 1. Under V2: S 3, A 8, B 10, C 17, D 6.

Ghazghkull Thraka, the acid test

V1 to V2

834 of 1161 units change letter between V1 and V2 (450 up, 384 down). Domains are V1's Hammer / Anvil / Banner and V2's Hammer / Anvil / Runner / Banner percentiles.

The 15 biggest rises:

The 15 biggest falls:

Job flips: 669 units change job between the versions (V1's Banner counts as kept when V2 calls it Runner or Banner). From V1's job (rows) to V2's (columns):

What players take that our numbers miss (a proxy for the community's reasons)

75 units are in 40% or more of their army's winning lists but get C or D under the proposal. Which of our traits they carry, against every unit's rate (lift = how much more often):

Orks, every unit

The list: Sprues & Brews, "Codex Orks Review (2026) – Warhammer 40k 11th Edition", Matt Crowther, 22 August 2026 (review copy supplied by Games Workshop) The review gives no tiers or ranks; it walks the datasheets and passes a verdict in words on most. Letters follow the strength of that verdict: S = the reviewer's strongest praise (a standout, huge output); A = clear strong praise (very good, a big upgrade, hard to shift, a favourite); B = positive but measured (better than before, handy, fun, a niche); C = mixed or a sidegrade (lost something for what it gained, pros and cons, much the same); D = the reviewer doubts it is worth taking. Units only described, with no verdict, are left out (Mek, Zodgrod Wortsnagga, Weirdboy, Painboy, Painboss, Runtherd, Mek Gunz, Warbuggies, Kill Rig). Verdicts were written before points were known and largely compare with the previous rules, so these letters are the reviewer's impression of each datasheet, not a competitive ranking.

Current: mean distance 1.04 tiers, 12 of 45 the same letter.

Maths only: mean distance 0.93 tiers, 14 of 45 the same letter.

V1: mean distance 0.91 tiers, 17 of 45 the same letter.

V2: mean distance 1.27 tiers, 12 of 45 the same letter.

V1's fitted weights on this list: mean distance 0.91 tiers with the weights fitted on list share, 1.09 with today's equal weights (the fit never saw this list).

Where they rate a unit higher than V2 does (29 units, biggest gaps first):

Where V2 rates a unit higher than they do (4):

The reasons behind the units we under-rate (V2 lower than theirs), counted:

Orks's letters: current S 14 A 15 B 11 C 13 D 2; maths S 4 A 11 B 17 C 19 D 4; V1 S 4 A 13 B 16 C 14 D 8; V2 S 3 A 5 B 13 C 17 D 17.