Residuals 1: Zoanthropes, Raveners, Deathleaper (2026-10-02)

Residual analysis on round 9 (T-562): from the community's verdict to the quantity or weight that would place the unit, and what else that change moves. A report: no tracked code changed; what-ifs ran in a scratch copy of tools/tiertest.js (round 9's run, every unit re-scored). Rules read on Wahapedia's 11th-edition pages, 2 Oct, in our words. Lists: grimstat-corpus (CC BY 4.0), 23 Tyranid lists, July to September.

Baseline (round 9): Tyranid ρ with Auspex 0.494, Space Marines −0.099, Orks −0.118; Tyranid list share ρ 0.453; every unit's list share ρ 0.077.

Summary

Engine findings (section 4): every charge is played 2" too long for the 11th edition; Synapse's 3D6 Battle-shock never reaches Tyranid Leadership.

1. Zoanthropes (90 pts, 3 models; Auspex S, 16 of 23 lists; ours C, Hammer 66)

a. Engine check

b. The raw row beside its peers (Hammer peers)

Damage: enemy points destroyed per 100 pts over the game, best way in, against GEQ / MEQ / TEQ / light / heavy. Hits to destroy: small arms / special / plasma / elite / anti-tank, then mortal.

Biggest differences. Per point, Hive Guard out-shoot the Zoanthropes on every target (+20% into light vehicles) and take twice the small-arms fire to kill; only plasma favours the Zoanthropes (4+ invulnerable). The community and lists say the opposite (S, 16 lists against A, 1): the edge isn't in their own dice.

c. What the winning lists do with it

23 units in 16 lists: 16 at 6 models, 7 at 3; 3 lists take three or four units. A Neurotyrant leads them in 10 (it is in 15 of the 16 lists). Most taken with Biovores (15), the Neurolictor (13), Raveners and Von Ryan's Leapers (10). Detachments named: Subterranean Assault (4, ours too), Ambush Predators with Vanguard Onslaught (3).

d. Hypotheses and what-ifs

1. Size (consensus right, for a reason we don't measure). We score every unit at its minimum size; the lists field Zoanthropes at 6. Per point 6 models deal 5% less, but Hammer counts damage before death with absolute survival (round 5), and 6 models live about twice as long: Hammer 66 → 89 (both: 92). Applied to every unit with a size count (S1), Hive Guard (Auspex A) also go to A; Tyranid ρ +0.058, Marines +0.045, Orks −0.026, all inside one SE. Marine sizes rest on 2 lists.
2. The leader (consensus right, measured but credited elsewhere). Node Lash is worth 8 points of the Zoanthropes' score; the Context rule hands it to the Neurotyrant. Alone it moves no letter and no ρ.
3. Terrain (consensus right: Psychic ignores cover). Zoanthropes gain 6.5 and pass Hive Guard (72 against 68), but 146 letters move elsewhere: real, small, blunt.
4. Popularity (consensus partly driven by a pairing). 15 of 16 Zoanthrope lists carry the Neurotyrant, which can lead only them, Tyrant Guard or Neurogaunts, and they are the cheapest Synapse unit. Hive Guard, better per point on our numbers, are neither and appear once. The S may price the pair: Z1a's 8 points; X1 (section 4) lifts their Runner 50 → 94.

e. Verdict: missing raw quantity (unit size)

Recommend scoring each unit at its most-listed size (its minimum where no list takes it), every army. Zoanthropes move C → A (89.2), which is the move needed (Auspex S; A within one). To be accepted: the Tyranid ρ up at least 0.05 (S1: +0.058) with neither clean panel falling more than 0.05 (S1: Marines +0.045, Orks −0.026), on a Marine size count from more than 2 lists.

2. Raveners (125 pts, 5 models; Auspex S, 12 of 23 lists; ours D, Runner 55)

a. Engine check

b. The raw row beside its peers (Runner peers)

Biggest differences. Raveners have the best reach and the most hits to destroy of the four, yet lose on the points gate (×0.73), Leadership (8+ on 2D6, bottom percentile) and units per 100 pts (34). Their damage per point is the group's lowest.

c. What the winning lists do with it

16 units in 12 lists, always 5 models; 4 lists take two. Hyperadapted Raveners lead them in 9 of 16 and are in 10 of the 12 lists. Most taken with Biovores (11), the Zoanthropes, the Neurolictor and the Hyperadapted Raveners (10). Detachments: Ambush Predators with Vanguard Onslaught (5), Subterranean Assault (4); we pick Synaptic Nexus.

d. Hypotheses and what-ifs

1. Re-deploying beats a cheap body (consensus right; wrong weight). Death From Below puts the unit on any objective each round and out of fights it doesn't want. Runner credits reach 100, then the points gate (full at 75 pts, zero at 260, borrowed from King of the Hill) takes 27% because a 125-pt unit "shouldn't" be spent on objectives; one that re-deploys every round isn't spent. R3 is this report's strongest lever: two units move, both towards their panel letters, and the Tyranid list-share ρ rises 0.06. Caveat: arrival data is Tyranid-only, so the clean panels can't test it.
2. The pack, not the squad (consensus right, critique #1). Scored with their leader the Raveners are a C Hammer; Runner dies at 290 pts. The panel may grade the led pack.
3. The Deep Strike charge (engine). A 7+ instead of a 9+ lifts the Raveners' raw Hammer 16% (95 → 111), but every melee unit rises with it, so their percentile moves 2.
4. Synapse Leadership (missing quantity). Real, and Raveners reach A, but one army's Leadership rule alone lifts 26 Tyranids (Ld percentile 0 → near 100) and ρ falls to 0.466: only for every army at once.
5. Popularity. 10 of 12 Raveners lists carry Hyperadapted Raveners (Auspex S), which can lead only Raveners. Part of the list share is the leader's.

e. Verdict: wrong weight (Runner's points gate on re-deployers), with a missing quantity behind it

Recommend gate 1 for a unit whose rules re-deploy it every round. Raveners D → B (86.0); Auspex S, so B is within two, the Red Terror lands within one of its A. To be accepted: arrival written for every army's re-deployers first, then the Tyranid ρ ≥ 0.53 with both clean panels within ±0.02.

3. Deathleaper (80 pts; Auspex A, 7 of 23 lists; ours S, Runner 98)

a. Engine check

b. The raw row beside its peers (Runner peers)

Biggest differences. The three hunters tie on Runner (97 to 98): Deathleaper leads on survival (66th percentile against the Lictor's 42), the Lictor on cheapness, the Neurolictor on Banner. Lists rank them the other way: Neurolictor 28 copies, Lictor 16, Deathleaper 7.

c. What the winning lists do with it

7 lists, one each. All 7 also take a Neurolictor; 6 the Zoanthropes, Raveners and Biovores. Detachments: Ambush Predators with Vanguard Onslaught (3), Subterranean Assault (1, ours).

d. Hypotheses and what-ifs

1. The ruler (consensus right). Deathleaper is a 98th-percentile Runner on our measures, but S at 95 gives S to 247 units, a fifth of the field. On round 9's proposed cuts it is A, as Auspex has it.
2. Survival saturates for a Runner (consensus right). Past about three turns, toughness buys a hiding unit no more objectives. D1 drops Deathleaper below the Lictor (the list order) but moves no Tyranid letter and no ρ.
3. Cheapness (the Lictor's case). D2 puts the Lictor top but halves every unit's list-share ρ (0.077 → 0.058) and lifts support characters: rejected.
4. Popularity. A one-copy character can't match a unit lists double (11 take two Neurolictors); the panel may follow that. The Neurolictor's Battle-shock engine (Banner 31) is the one difference we measure.

e. Verdict: consensus right, the ruler wrong

Recommend adopting round 9's proposed cuts (S ≥ 98.79). Deathleaper (97.93) becomes A and stays there while it is more than 0.86 under the cut. To be accepted for all units, the cuts must not cost exact matches on the clean panels; on Tyranids they cost 5 (18 → 13, against 8.4 and 7.4 by chance), so the cuts need the panel test, not this unit.

4. Engine findings

1. Charges are 2" too long for the 11th edition (bug). 11e engagement range is 2" (core rules 03.04), so a charge needs the gap minus 2. pCh (V4) and pChg (V5) take the full gap, and scoring.md says a Deep Strike charge needs a 9 (28%); from just over 8" it needs a 7 (58%). R2 fixes both: Tyranid ρ 0.496, Marines −0.136, Orks −0.123, 46 letters. It should be fixed on rules grounds, not as a fix for any one unit.
2. Synapse's 3D6 Battle-shock is never credited (gap). Only the parked --armyfx pilot applies an army rule to Leadership, so every Tyranid's Leadership percentile sits at 0 to 6. X1 shows the size: 26 Tyranid letters, ρ 0.466. It needs every army's Leadership rules at once.
3. BSData extras (no effect today). Stealth on Raveners, Deep Strike on Deathleaper, the Lictor and the Neurolictor aren't on the 11e sheets (Deep Strike comes from Ambush Predators). D3: 0 letters. Worth a note to the data's maintainers.
4. The best single detachment. Lists pair detachments (Ambush Predators with Vanguard Onslaught most); we pick one, and it is rarely theirs. Not tested here.

5. Method notes

• What-ifs re-score every unit and compare with round 9 from the same process (its JSON reproduces exactly).
• Most-listed sizes from 23 Tyranid, 21 Ork and 2 Space Marine lists; 5 Tyranid, 7 Ork and 5 Marine units change size.
• Scratch: the tool copy, hooks and cases in the session's scratchpad (res1/).