The stratagem pilot: Tyranids and Leagues of Votann (steps 44 and 44b)

6 Oct 2026, Pacific (T-713; Jordan approved the pilot on 6 Oct). Frozen snapshot 8dcd24032d, inputs hash dcd477d8a78c. Tools: tools/stratfx/ (the catalogue, our own words), tools/stratuplift.js (the uplift), tools/stratagems.js (the CP spend, the credits, the post step and the fit). Base: step 41 (tools/ledger-benchmarks/2026-10-06-step41-stable.json). Each prediction was written in reports/tiers/ledger-steps.md and committed before that step was scored. Nothing is adopted, and nothing on the site, in docs/data or in the published benchmark changes.

Detachments are named by their slugs, the ones the ledger's form ids already carry. Stratagems are named only by our ids (<slug>.<n>) and described in our own words.

In plain English

What we built.

• The catalogue. It covers every detachment the ledger knows for the two armies (10 each). For each one we list every stratagem that changes dice, as effects the dice engine already understands. Each entry records its CP cost, the phases it can be used in, which units can take it, how many units one use reaches, and our estimate of how often a condition holds.
  • Tyranids: 20 dice stratagems, 18 of them priced.
  • Votann: 26 dice stratagems, 25 of them priced.
  • Some can't be priced yet: Precision, Feel No Pain against mortal wounds only, and ignoring wound or AP modifiers. These are kept with their reason.
  • The rest (movement, reserves, objectives, mortal wounds, healing) are listed by kind only.
• The uplift. For each unit in each detachment we take its best form there. We add one use of the stratagem's effects and measure what it adds, using the ledger's own dice and field.
  • Attacking stratagems: the points the unit finishes in one activation (Burst), against the typical 2,000-point army. A fight-phase stratagem counts at the share of fights the unit's charge reaches.
  • Defending stratagems: the points saved when one average enemy unit attacks it.
• The CP economy.
  • 2 CP a battle round. In 11th edition each player gains 1 CP in every Command phase, both players' turns (core rules, Wahapedia). That makes 10 CP a game.
  • The typical list for a detachment is the army's units by their Value there, best first, up to 2,000 points. No popularity goes in.
  • Each round, each stratagem can be used once per phase. A use goes to the unit it adds most to, and uses are bought by points added per CP.
  • A unit's credit is what it receives. A CP source's credit is what its extra CP buy.
• Two versions.
  • Step 44 uses the detachments' own stratagems only.
  • Step 44b adds the core re-roll: 1 CP to re-roll one hit or wound roll, which any army can use. It was declared after step 44's scores, so it is a second look, not a clean test.

Points per CP depend on the detachment, as Jordan said. At the margin (the points the next CP would add, step 44b), the values run:

• Tyranids: from 26 points (synaptic-nexus) down to 0 (unending-swarm).
• Votann: from 22 points (needga-rd-oathband) down to 6 (farseekers).

Averaged over the 10 CP:

• Tyranids: 11 to 33 points per CP.
• Votann: 10 to 28 points per CP.

The detachments whose stratagems add +1 to wound, re-rolls, Sustained Hits or extra shooting have the highest values. The weak ones run out of dice stratagems to buy.

The CP sources are worth about what Jordan guessed.

• The Swarmlord's +1 CP a round (about 4 CP a game) buys +106 points in synaptic-nexus, +88 in invasion-fleet and +55 in assimilation-swarm. Jordan guessed "maybe an extra 100 points".
• The same 4 CP buy almost nothing (+9) in subterranean-assault. That is where the ledger puts the Swarmlord and the Hive Tyrants: it holds one once-a-game dice stratagem, so extra CP have little to buy.
• Ûthar's cheaper stratagem is worth +88 points a game in needga-rd-oathband and +70 in brandfast-oathband.

The stratagems land on a few units.

• Tyranids: the Norn Assimilator, the Swarmlord, the Hyperadapted Raveners, Old One Eye and the Deathleaper.
• Votann: the Kapricus Defenders, the Sagitaur, the Thunderkyn, the Yaegirs, the Earthshakers, the Hekaton and the Hearthguard.

Agreement with the reviewers.

• Tyranids: no change. At face value (w 1): −0.1 [−0.6, +0.3] on their three 11e lists, 82.6 → 82.5. No strength helps them by more than +0.1.
  • Their best-valued units already sit at the top.
  • The credit adds at most 8% to a top unit's Value (Deathleaper), and 12 to 15% to two mid-table ones.
  • Their CP sources earn nothing in subterranean-assault, where the ledger places them.
• Votann: a large rise, but not one we can trust. At face value, 61.0 → 70.5 (+9.5 [+0.0, +21.9]); +12.4 with the core re-roll, and +19 to +20 at w 2. Their three 10e lists and Gemini's list rise too.
  • Credits handed to random units of the army also lift Votann: by +4.7 on average at w 1 (+7.2 with the core re-roll), and by up to +16 for one seed.
  • Shuffled letters move ±8.
  • Votann have one 11e list grading 18 units, and that cannot separate the real credit from noise.
• The cross-test fails the rule stated beforehand. That rule was: both directions gain, and each beats its placebo credits.
  • Step 44: fitted on Tyranids, w is 0, so Votann gain nothing. Fitted on Votann, w is 2, and the Tyranids lose 0.1.
  • Step 44b: fitted on Tyranids, w is 0.25, and Votann gain +3.8 (the placebo credits' mean is +2.2). Fitted on Votann, w is 2, and the Tyranids lose 0.1.
  • One direction, at best.

What it means.

• Stratagems are real damage multipliers, and the pilot measures them in the points-exchange frame. One CP is worth 5 to 35 points depending on the detachment, and a CP source 10 to 100 points a game.
• Priced into Value they move the Votann a lot and the Tyranids not at all. With one Votann list we can't tell whether that Votann move is signal.
• Not adopted. The honest test needs more armies. The same catalogue for all 16 armies, measured with the 15-army held-out exam (tools/recipestep.js) like every other step.

1. Points per CP, by detachment

The budget is 10 CP a game (core). The points are what the 10 CP buy over a game, killed and saved together.

• Margin: the points the next CP would add.
• Swarmlord / Ûthar: what a 4-CP source adds there.
• Lictor / Pioneers: what the 0.5 CP (Tyranids) or 0.67 CP (Votann) source adds.

Sorted by the step 44b margin.

How to read it:

• A margin of 0 means the detachment's dice stratagems are all bought before the 10 CP run out. More CP would buy nothing we price: a generator there is worth nothing in our terms.
• unending-swarm scores low because its stratagems need ENDLESS MULTITUDE units. None of them is among the army's top units by Value, so none is in its typical list. A list built around them would score differently.
• The Hive Tyrant and the Winged Hive Tyrant have the same 4 CP a game as the Swarmlord (a cheaper stratagem each round), so the same figures apply. The Lictor, Neurolictor and Deathleaper make about 0.5 CP a game each.

2. Where the CP goes: the biggest uplifts

Step 44b. Uses are a game's uses; killed and saved are points a game; per turn per 100 is the credit per turn, per 100 points of the unit.

The single biggest uses:

• Tyranids:
  • feel no pain on the Swarmlord in the enemy's fight phase: 40 points a use;
  • feel no pain 5+ (4+ on an objective) on the Norn Assimilator: 51 points for 2 CP;
  • the re-rolls of 1 near a Synapse unit, reaching two units: 35 points.
• Votann:
  • a light vehicle shooting back in the enemy's turn: 55 points for 2 CP;
  • the Hekaton doing the same with Yield points: 44 points;
  • Sustained Hits on the Thunderkyn: 23 points a use;
  • melee re-rolls on the Yaegirs: 23 points;
  • +1 Strength and Sustained Hits on the Hearthguard: 17 points.

3. The CP sources

Step 44b. A CP source's credit is the points its CP add to its detachment's spend in a game. The post step counts it per turn per 100 of the source's points, at the army's median Kill rate.

The post step lets a unit move to the detachment where its Value plus its credit is highest. The Swarmlord's synaptic- nexus form plus its credit still doesn't reach its subterranean-assault form's Value, so it stays there and gains little. The Swarmlord's aura (enemy stratagems cost more) is left out: it acts on the enemy's budget, not ours.

4. Agreement

Pairwise accuracy, %.

• Tyranids: their three 11e lists (52 units in the army).
• Votann: their one 11e list, Auspex (22 units in the army, 18 graded).

The cross-test (w picked on one army's human letters, scored on the other):

Face value (w 1), with a paired unit bootstrap:

• Tyranids: −0.1 [−0.6, +0.3] in both versions.
• Votann: +9.5 [+0.0, +21.9] in step 44, +12.4 [+0.0, +29.4] in step 44b.

Against the prediction (written before scoring):

• Right:
  • the Tyranids within ±1.5 (−0.1);
  • the cross-test failing;
  • controls as large as the effect;
  • Hearthguard and Thunderkyn up;
  • Exocrine, Zoanthropes and Neurotyrant unmoved.
• Wrong:
  • Votann within ±3: it was +9.5, three times the range.
  • The Tyranid fit chose w 0, not 0.25 to 1.
  • The Swarmlord and Hive Tyrant did not move up. Their best detachment's CP buy nothing, which step 44b only partly fixes.

5. The disputed units

Rank among the army's ranked units, step 41 → face value (w 1, step 44b). Letters are each list's (Auspex 11e first).

Votann

• Thunderkyn: 5th → 4th (+45% Value; A, S, S). Right, small in rank.
• Hearthguard: 16th → 15th (+20%; A, A, A). Right, but far too small: still 15th of 22 against three A letters.
• Hekaton: 11th → 10th (+15%; A, S, S, S). Right, small.
• Kapricus Defenders: 13th → 6th (+82%; B, C, C). Against most letters: the shooting-back stratagem lands on them.
• Sagitaur: 15th → 8th (+52%; A, C, D, A). Mixed.
• Ûthar: 8th → 7th (+35%; B, B, A, A). About right.

Tyranids

• Hive Tyrant: 2nd → 3rd (+0.5%; S, A, S). Unmoved.
• Swarmlord: 4th → 4th (+0.4%; S, S, A). Unmoved.
• Neurotyrant: 29th → 29th (A). Unmoved: none of the dice stratagems in its detachments goes to it.
• Exocrine: 18th → 19th (A, A, A). Unmoved: the shooting stratagems that suit it (ignore cover, crusher-stampede) lose to the defensive ones on points per CP.
• Zoanthropes: 9th → 9th (S, A, B). Unmoved.

Gainers off the disputed list

• Deathleaper: 3rd → 2nd (+8%).
• Norn Assimilator: 14th → 13th (+12%).
• Ranged Warriors: 15th (+15%).

The Hearthguard and Exocrine point at the same gap. A stratagem worth 17 points a use only pays where the unit is the best place for that CP. A reviewer's "Hearthguard are great with their stratagems" is a different judgement: what the unit is worth when it does get the CP.

6. What the pilot leaves out (doubts)

• Same uses every round.
  • Every round is the same and the units never die, so a stratagem's five uses all go to the same unit. In a game, the best target is sometimes out of range or dead.
  • A unit may take two different stratagems in one phase. The spend adds their values, though they don't add exactly.
• Not in the dice.
  • A fight back in the enemy's fight phase: our fight-phase stratagems are counted in our turn only.
  • Precision, Feel No Pain against mortal wounds, and ignoring wound or AP modifiers. These are noted, at 0.
  • Core stratagems other than the re-roll: Grenades, Tank Shock, Fights First after the enemy has fought, Overwatch.
• Our own estimates, not fitted.
  • Votann's Yield-point versions at half the uses.
  • Army-rule states at half.
  • An assailed target at half.
  • A hidden target at half.
• Units of the typical list.
  • The list is chosen by Value with one copy of each unit. That is circular by design (strong units get the CP).
  • It can leave out the units a detachment is built for: unending-swarm's multitudes.
• Shooting again is generous. Its whole Burst counts, at the shares stated; the range and trigger limits stay as shares.

7. Rolling out to all 16 armies: cost and plan

Cost.

• About 160 detachments, roughly 900 stratagems, of which perhaps 400 change dice.
• This pilot read and encoded 20 detachments (102 stratagems, 46 of them dice) in about 45 minutes. That suggests 6 to 8 hours of reading and encoding, plus a review pass by a second reader against Wahapedia's 11th-edition pages, in our words.
• The numbers are cheap: the uplift takes 12 to 50 s an army, and the spend and post step take seconds.

Plan.

1. The catalogue, army by army (tools/stratfx/<army>.json), with the core re-roll. The armies with the largest detachment rosters first.
2. The uplift and the spend for every army: points per CP per detachment, and every CP source's value. This is useful on its own, as a report.
3. Price into Value only as a declared step on the 15-army held-out exam, run with tools/recipestep.js's protocol and keep rule. w is fitted on 10 armies and scored on the other 5, with placebo credits as the control. That is the honest test the two-army pilot can't be.
4. Two refinements are worth declaring with it, each as its own step:
   • a unit's rounds alive (burst.js --life's L) in place of five equal rounds;
   • a cap of one use per unit per phase.

The recommendation: roll out the catalogue (steps 1 and 2). The calculator needs it anyway (section 8). Hold pricing into Value for the 15-army step. Don't adopt anything from this pilot.

8. The same catalogue in the calculator, later

The catalogue is already in the calculator's own vocabulary: effects with side, mods and when, as docs/data/detfx.json holds them. Later, each detachment could show its stratagems as chips that start off, the way the army rules' marks do (T-554). Each chip would carry the CP cost and our short description ("re-roll hit rolls in melee, a MONSTER"), never the stratagem's name or text.

Under the same card, the 282 stratagem-name notes in docs/data/detfx.json ("Stratagems, by name only…") would be replaced by these descriptions. Not done now: nothing in docs/data changes in this pilot.

Files and numbers

• Catalogue (tracked, our words):
  • tools/stratfx/tyranids.json
  • tools/stratfx/leagues-of-votann.json
  • tools/stratfx/core.json
• Credits (tracked, small):
  • tools/stratfx/credit-<army>.json (step 44)
  • tools/stratfx/credit-core-<army>.json (step 44b)
• Uplift (git-ignored): build/ledger/strat-uplift-<army>.json
• Fits (git-ignored): build/strat/fit.json and fit-core.json
• Commands, in order:
  • node tools/stratuplift.js --army tyranids (then --army leagues-of-votann)
  • node tools/stratagems.js --spend [--core] --write
  • node tools/stratagems.js --fit [--core] --json build/strat/fit.json
• The realm's live files now include tools/stratfx/*.json (tools/ledger-inputs.js LIVE). The lock and the inputs hash are unchanged: dcd477d8a78c, verified.