2026-10-05. What the game's points charge for each perk, holding fixed what our dice engine says a unit does. Hedonic pricing in the economist's sense: every datasheet's points regressed on its attributes (the blind appraisal's idea 1, reports/tiers/2026-10-05-blind-appraisal.md section 4). Not the 5 Oct "hedonic" report, which fitted the ledger's dials to tournament popularity; nothing here reads popularity. A report and a tool (tools/pricelist.js): no ledger, setting, benchmark, data file or frozen input changed. Frozen inputs eb3590a2aa (hash 30f0056e6103), BSData 374f505. Our own numbers; unit, keyword and rule names only, never rules text.
In plain English
- The game's points are very predictable from what a unit does. Over 829 datasheets of the 16 armies with a ledger, our engine's numbers (damage into five stand-in targets, durability against five attackers, OC, Move) plus 73 perk variables explain 88.4% of the variation in log points within an army, on armies the fit never saw (leave one army out). The engine's numbers alone explain 85.0%; every perk together adds 3.3 points of R². With the ledger's own Kill, Soak and OC in place of the rankings' numbers (586 units): 87.4%.
- What the designers charge for a perk, at equal damage, durability, OC and Move (90% bootstrap range; every price is shrunk toward zero, so these are conservative): Fly +10% (+7% to +14%); Lone Operative +10% (+2% to +16%); command points rule +8% (+5% to +12%); Indirect Fire +7% (+1% to +14%); own arrival rule +6% (+3% to +10%); Stealth +5% (+1% to +9%); leader dice effects +5% (+1% to +9%); Transport +4% (+1% to +8%); Blast +4% (+2% to +7%).
- What they discount: Support −13% (−20% to −7%); Feel No Pain −10% (−14% to −5%); transport capacity −5% (−10% to −1%). Unit types carry their own prices: Titanic +35% (+25% to +44%); Aircraft +14% (+8% to +22%); Towering +14% (+0% to +31%); Mounted +5% (+0% to +9%); Epic Hero +4% (+0% to +7%).
- Most perks are nearly free once the dice are counted. 27 of 73 perk prices have a range that excludes zero, and 13 of those are under 5% either way. The engine's numbers carry the price: their collinear group (damage, durability, OC, the statline that makes them) moves a unit's points by +63% per standard deviation, against +4% for movement and arrival, +3% for weapon keywords and +8% for the unit-type keywords.
- Under-priced units do not get better letters. On Tyranids the designers' residual alone (cheaper than the pattern ranks higher) orders the five lists' pairs 50.9% right, chance level (Track B's ledger 77.7%; the reviewers' own ceiling about 89% against their consensus, 84% one to one); ρ with the consensus 0.01. Added to the ledger at equal weight (fixed in advance) it costs 11.2 points (67.0%, paired bootstrap SE 3.9 pts); as a log term (Value per point × the mispricing) 76.2% (−1.7 pts ± 1.6 pts). The other 15 armies agree: residual alone 49.2% on average against the ledger's 58.6%, above 55% in 5 armies and below 45% in 6, so not even a consistent sign. The residual against the ledger's own capabilities does no better (Tyranids 53.6%).
- Why, probably: a residual mixes true mispricing with what the designers price and our engine doesn't see (Gargoyles' scoring and screening read as "over-priced") and with what our engine overcounts (the Toxicrene's and Tyrannocyte's damage on paper read as "under-priced"). The reviewers seem to reward the first kind and see through the second; the points alone can't tell them apart.
- For T-662's family dials (1× = what the engine credits now): the designers price weapon keywords about as the dice do (the family together 1.12×, 0.84 to 1.44; Blast 3.42× (1.97 to 5.13) and Anti-X 3.06× above; Lethal Hits 0.08× and Sustained Hits 0.44× below, with wide ranges) and Feel No Pain at 0.23× (-0.12 to 0.58) of its dice durability. The ledger credits Fights First +144% of a carrier's Value; the designers charge −8% (15 carriers; a dial of -0.06× on the increment, 0.38× on the carrier's Value). Lone Operative the other way: the ledger +1.2%, the designers +10% (8.13×, 1.97 to 13.53). Support (measured): +53% on the 63 units it moves, the designers −3% for leaders, auras and Synapse. Arrival and Stealth have no separable ledger increment (the charge model and Soak measure them), so no ratio: the designers' price is an additive credit (Deep Strike +2%, Scouts +1%, Stealth +5%). Use these as the centre of a prior, not as settings: where the ledger's large credits (Fights First, support) disagree with the points, the ledger's credits are the ones tuned on the reviewers' letters, so the dial search has to weigh the two.
- The biases (section 9): the designers price for every kind of game, not only tournaments, and answer the previous balance pass; points move every few months and come in steps of 5; they price whole datasheets, so perks that travel together (a Monster's Toughness, wounds and points; a Leader's Character keyword) can only be priced jointly (section 4); and "at equal capabilities" means equal by our engine, so whatever it misses shows up as a price.
1. The data and the fit
- Rows: docs/data/rankings.json at the frozen commit, the 16 armies with a ledger: 1071 rows; 186 are a datasheet already counted under another army (the Legends tanks Space Marines and Death Guard share, the Brood Brothers' Astra Militarum datasheets) and 56 are Crucible datasheets (built from wargear with its own points, which neither their price nor the engine's numbers include: left out), so 829 datasheets, 243 of them Legends (kept, with a flag). 0 rows without a record.
- Price: log(points) at the minimum size (rankings.json's basis) and log(points per model) there. Log points have SD 0.59 overall and 0.57 within armies; 7.2% of the variance lies between armies, and it is taken out: every army is its own price level, so a perk is priced against the units of its own army.
- Capabilities (a): rankings.json's engine numbers (tools/build_rankings.js on the frozen docs/ev.js) made absolute (× points ÷ 100), so no capability is divided by the price it explains: wounds removed into GEQ, MEQ, TEQ, light and heavy vehicle (ranged plus melee, its best legal loadout into each; log 1 + x ÷ the median), the melee share, hits needed to destroy it against five stand-in attackers (log), total OC (log 1 + models × OC), Move.
- Perks (b): 74 variables from diagnose.js's list (reports/tiers/2026-10-04-variables.md) read from the datasheet record, the effects files and tools/carry.json, never from points, in the families of section 2. Weapon keywords are read on the default loadout. Left out: anything made from points (points per wound, OC per 100, the size steps' prices), names, and army-only keywords (a sub-faction's own keyword is a name in effect).
- Fit: ridge regression on within-army deviations, columns standardised, penalty λ = 100 chosen by leaving one army out (pooled squared error on the left-out army's deviations; path in the JSON). A lasso the same way (α 0.002) says which perks survive selection. Ranges: 500 bootstrap resamples of units within armies (5th to 95th percentile), and the 16 fits that each leave one army out ("Armies" in section 2).
- Per model: the same residuals give R² 96.2% on log points per model (per-model points vary more than unit points, so there is more to explain and a larger share is explained); refitted on per-model points, 95.2%, with perk prices rank-correlated 0.71 with the unit fit's (median difference 2.4%, the largest 17.8%; ridge isn't invariant to the shift of log models' coefficient by one). Each doubling of the minimum size changes the price at equal capabilities by +1.7%: the designers charge for bodies through what the bodies do.
- Without Legends (586 datasheets): left-out R² 88.3%; the perk prices' rank correlation with the full fit 0.81.
| Fit (left out by army) | R² within army |
|---|
| capabilities + perks, unit's points | 88.4% |
| the same residuals, points per model | 96.2% (refitted per model: 95.2%) |
| capabilities alone | 85.0% |
| perks alone | 81.9% |
| in sample (all 16 armies, for reference) | 90.6% |
| the ledger's capabilities + perks (586 ledger units) | 87.4% (the rankings' on the same units 88.3%; the ledger's capabilities alone 84.1%) |
R² by army, left out (the fit never saw that army)
| Army | Datasheets | Unit's points | Per model | Capabilities alone |
|---|
| Tyranids | 57 | 92.1% | 97.5% | 89.7% |
| Emperor's Children | 23 | 95.2% | 98.8% | 93.8% |
| World Eaters | 35 | 94.6% | 98.7% | 91.9% |
| Space Marines | 158 | 91.4% | 96.4% | 89.1% |
| Astra Militarum | 135 | 90.9% | 96.5% | 87.7% |
| Death Guard | 49 | 90.0% | 96.9% | 90.4% |
| Dark Angels | 19 | 88.9% | 94.9% | 85.6% |
| Space Wolves | 41 | 88.6% | 95.9% | 83.1% |
| Thousand Sons | 38 | 88.5% | 97.2% | 84.3% |
| Orks | 70 | 87.5% | 94.7% | 87.2% |
| Adeptus Mechanicus | 38 | 84.6% | 96.2% | 78.2% |
| Agents of the Imperium | 46 | 76.1% | 93.9% | 68.0% |
| Leagues of Votann | 22 | 69.7% | 95.5% | 70.6% |
| Drukhari | 46 | 68.2% | 95.1% | 58.3% |
| Genestealer Cults | 25 | 48.9% | 91.6% | 18.9% |
| Blood Angels | 27 | 3.0% | 87.4% | -36.5% |
Blood Angels (27 datasheets, SD of log points within the army 0.27) and Genestealer Cults (25 datasheets, SD of log points within the army 0.31): the lowest R². Their spread is smaller than the 16 armies' 0.57, so there is less to explain and a few units decide it.
2. The price list: each perk, holding capabilities fixed
A perk's price is how much more (or less) a datasheet with it costs than one without, at the same engine numbers and other perks, within its army: exp(β) − 1 for a flag, per step for a number. Range: the bootstrap's 5th to 95th percentile. "Armies": the lowest and highest price over the 16 fits that each leave one army out. "Lasso": kept at the lasso's left-out α. "Ledger caps": the price with the ledger's capabilities in place of the rankings' (586 units; rank correlation of the perk prices 0.66). Bold: the range excludes zero.
Movement and arrival
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| Deep Strike | 158 | has it | +2.3% | −0.6% to +5.2% | +1% to +3% | yes | +2.6% |
| Scouts | 59 | has it | +1.4% | −1.8% to +4.7% | +0% to +3% | yes | +2.8% |
| Infiltrators | 40 | has it | +2.9% | −2.5% to +8.1% | +0% to +5% | yes | +3.2% |
| Fly | 139 | has it | +10.0% | +6.7% to +13.8% | +8% to +11% | yes | +9.6% |
| own arrival rule | 75 | has it | +6.3% | +3.0% to +10.1% | +5% to +7% | yes | +7.1% |
| own extra movement | 172 | has it | +1.1% | −1.4% to +3.3% | +1% to +2% | yes | +2.2% |
Defensive rules
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| Stealth | 46 | has it | +4.9% | +0.6% to +9.3% | +4% to +8% | yes | −1.4% |
| Lone Operative | 31 | has it | +9.8% | +2.4% to +16.3% | +5% to +15% | yes | +11.2% |
| Feel No Pain | 62 | a 5+ | −9.5% | −13.6% to −5.3% | −11% to −7% | yes | −9.4% |
| own defensive dice effects | 27 | has it | +4.6% | −1.2% to +9.8% | +4% to +6% | yes | +2.7% |
| heals or returns models | 58 | has it | +3.2% | −0.3% to +7.2% | +1% to +4% | yes | +3.1% |
Statline
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| Toughness | 829 | +1 | +1.1% | +0.6% to +1.6% | +1% to +1% | yes | +2.0% |
| armour save | 812 | one step better | +0.4% | −0.5% to +1.8% | −0% to +1% | – | −0.3% |
| invulnerable save | 359 | one step better | −1.6% | −2.4% to −0.6% | −2% to −1% | yes | +0.7% |
| wounds per model | 753 | ×2 | +2.4% | +1.4% to +3.3% | +2% to +3% | yes | +1.1% |
| Leadership | 436 | one step better | +5.3% | +2.7% to +8.0% | +4% to +6% | yes | +7.3% |
| Damaged bracket | 196 | has it | +4.6% | +2.4% to +7.9% | +4% to +7% | yes | +8.7% |
Objective
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| OC per model | 750 | +1 | +0.4% | −0.4% to +1.3% | +0% to +1% | – | +0.3% |
| Battleline | 38 | has it | +2.0% | −2.1% to +6.6% | −0% to +4% | yes | +3.2% |
| objective rule | 71 | has it | −3.2% | −7.2% to +1.0% | −5% to +1% | yes | +0.2% |
| command points rule | 87 | has it | +8.2% | +4.8% to +11.5% | +6% to +9% | yes | +7.4% |
Weapon keywords
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| Lethal Hits | 99 | has it | −2.8% | −7.5% to +1.9% | −4% to −1% | yes | −2.3% |
| Sustained Hits | 156 | has it | −1.2% | −3.4% to +1.4% | −3% to +0% | yes | −2.7% |
| Devastating Wounds | 133 | has it | −0.3% | −3.0% to +2.3% | −1% to +1% | – | −0.2% |
| Anti-X | 133 | has it | +2.4% | −0.2% to +5.6% | +1% to +3% | yes | −0.8% |
| Precision | 69 | has it | +0.7% | −2.8% to +4.3% | −1% to +2% | yes | −1.8% |
| Indirect Fire | 31 | has it | +7.3% | +1.4% to +14.1% | +4% to +9% | yes | +10.6% |
| Torrent | 83 | has it | −0.8% | −4.7% to +3.3% | −2% to +0% | – | −2.5% |
| Blast | 146 | has it | +4.4% | +1.8% to +7.3% | +4% to +5% | yes | +7.0% |
| Melta | 28 | has it | −0.4% | −3.7% to +2.8% | −2% to +0% | – | +3.1% |
| Twin-linked | 154 | has it | +1.5% | −1.1% to +4.5% | +1% to +2% | yes | +3.1% |
| Ignores Cover | 83 | has it | +0.5% | −3.8% to +4.4% | −1% to +2% | – | +2.4% |
| Psychic weapon | 57 | has it | +4.2% | −0.2% to +9.3% | +3% to +5% | yes | +2.1% |
| Heavy | 68 | has it | −0.5% | −4.4% to +3.2% | −3% to +1% | – | −0.7% |
| Assault | 74 | has it | −2.7% | −5.9% to +1.0% | −4% to −1% | yes | −0.1% |
| Pistol | 200 | has it | −0.3% | −2.6% to +2.4% | −1% to +0% | – | −0.9% |
| Lance | 22 | has it | −3.5% | −9.5% to +2.0% | −5% to −2% | yes | −6.6% |
| Hazardous | 63 | has it | −1.3% | −4.9% to +2.3% | −2% to −0% | yes | −1.7% |
| Rapid Fire | 167 | has it | −1.8% | −4.0% to +0.8% | −2% to −0% | yes | +0.8% |
| Extra Attacks | 64 | has it | −0.1% | −3.9% to +3.9% | −1% to +1% | – | +2.3% |
| One Shot | 8 | has it | +1.5% | −5.6% to +9.6% | +0% to +3% | – | +4.8% |
Leader and support
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| Leader | 182 | has it | −3.1% | −5.8% to +0.3% | −4% to −2% | yes | −8.1% |
| datasheets it leads | 184 | its carriers' median (1.39) | +1.4% | −1.4% to +4.1% | +0% to +2% | – | −0.1% |
| Support | 32 | has it | −13.1% | −20.2% to −6.9% | −16% to −10% | yes | −22.1% |
| Aura | 95 | has it | −2.6% | −5.8% to +1.3% | −4% to −1% | yes | −1.6% |
| leader dice effects | 118 | has it | +4.7% | +0.7% to +8.5% | +3% to +6% | yes | +8.1% |
| Transport | 87 | has it | +4.4% | +0.7% to +8.4% | +4% to +5% | yes | −8.5% |
| transport capacity | 54 | its carriers' median (2.56) | −5.2% | −9.6% to −1.0% | −7% to −4% | yes | +6.4% |
| Synapse | 18 | has it | +3.6% | −3.0% to +10.3% | +0% to +5% | yes | +1.8% |
Melee and other rules
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| Fights First | 12 | has it | −8.2% | −17.3% to +2.6% | −13% to −3% | yes | −8.1% |
| Deadly Demise | 338 | has it | +0.8% | −2.5% to +4.3% | −1% to +3% | – | +2.8% |
| mortal wounds rule | 83 | has it | +2.9% | +0.0% to +6.0% | +2% to +4% | yes | +2.7% |
| Battle-shock rule | 93 | has it | −1.8% | −4.5% to +1.2% | −3% to −1% | yes | −3.5% |
| hinders enemies | 53 | has it | −0.9% | −4.6% to +3.1% | −4% to −0% | yes | −0.9% |
| better Overwatch | 24 | has it | −2.5% | −8.8% to +4.3% | −5% to +2% | yes | −1.2% |
Keywords
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| Character | 278 | has it | +2.5% | −0.5% to +5.1% | +1% to +3% | yes | +3.7% |
| Epic Hero | 104 | has it | +3.5% | +0.1% to +7.1% | +2% to +5% | yes | +10.3% |
| Monster | 59 | has it | +1.5% | −3.4% to +5.8% | −1% to +3% | – | +3.5% |
| Vehicle | 281 | has it | −3.7% | −6.8% to −0.3% | −5% to −2% | yes | −3.4% |
| Infantry | 408 | has it | +1.2% | −1.9% to +3.8% | +0% to +2% | – | −3.0% |
| Walker | 54 | has it | −5.7% | −9.2% to −2.0% | −8% to −4% | yes | −6.5% |
| Mounted | 62 | has it | +4.8% | +0.5% to +9.4% | +3% to +5% | yes | −2.6% |
| Beast | 15 | has it | +6.3% | −2.2% to +14.3% | +4% to +11% | yes | −1.7% |
| Swarm | 4 | has it | +6.2% | −4.2% to +18.4% | +5% to +9% | – | −12.5% |
| Psyker | 60 | has it | +1.8% | −2.5% to +6.1% | −0% to +3% | – | +2.0% |
| Titanic | 36 | has it | +35.0% | +24.5% to +44.0% | +30% to +46% | yes | +44.0% |
| Towering | 6 | has it | +14.0% | +0.2% to +30.8% | +6% to +23% | yes | +8.6% |
| Aircraft | 34 | has it | +14.3% | +8.0% to +21.8% | +12% to +18% | yes | +11.4% |
| Daemon | 59 | has it | +1.9% | −1.9% to +7.1% | +0% to +4% | yes | −3.7% |
| Terminator | 31 | has it | +1.2% | −4.7% to +6.2% | −1% to +4% | – | +1.6% |
Size
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| models at minimum | 250 | ×2 | +1.7% | +0.8% to +2.6% | +1% to +2% | – | +6.8% |
| can grow | 200 | has it | −8.5% | −10.8% to −5.8% | −9% to −8% | yes | −10.9% |
| mixed statlines | 45 | has it | −2.9% | −7.2% to +1.7% | −4% to −0% | yes | −2.3% |
Legends
| Perk | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| Legends | 243 | has it | +4.9% | +2.5% to +7.3% | +3% to +6% | yes | – |
Capability
| Capability | Carriers | Step | Price | Range | Armies | Lasso | Ledger caps |
|---|
| damage into CHAFF | 822 | one SD within army | +2.6% | +1.3% to +4.1% | +2% to +3% | – | – |
| damage into MEQ | 822 | one SD within army | +5.3% | +4.2% to +6.7% | +5% to +6% | yes | – |
| damage into TEQ | 822 | one SD within army | +5.9% | +4.8% to +7.0% | +6% to +6% | yes | – |
| damage into LV | 822 | one SD within army | +5.5% | +4.5% to +6.4% | +5% to +6% | yes | – |
| damage into HV | 822 | one SD within army | +4.2% | +3.3% to +5.2% | +4% to +5% | – | – |
| melee share | 816 | one SD within army | −1.3% | −2.4% to −0.1% | −2% to −1% | yes | – |
| durability vs small | 829 | one SD within army | +4.3% | +3.3% to +5.2% | +4% to +5% | yes | – |
| durability vs special | 829 | one SD within army | +5.2% | +4.1% to +6.2% | +5% to +5% | yes | – |
| durability vs plasma | 829 | one SD within army | +5.0% | +3.8% to +6.1% | +5% to +5% | yes | – |
| durability vs elite | 829 | one SD within army | +5.8% | +4.5% to +7.1% | +5% to +6% | yes | – |
| durability vs at | 829 | one SD within army | +7.4% | +6.3% to +8.3% | +7% to +8% | yes | – |
| total OC | 750 | one SD within army | −1.6% | −2.7% to −0.1% | −2% to −1% | yes | – |
| Move | 794 | one SD within army | −1.0% | −2.2% to +0.3% | −1% to −1% | yes | – |
With the ledger's capabilities the price per standard deviation is: Kill into hordes −0.5%, Kill into infantry +8.1%, Kill into armour +12.1%, Kill into characters +4.7%, Soak against shooting +11.6%, Soak in melee +7.7%, OC (ledger) −2.7%, Move −1.0%.
Notes on the larger prices: Titanic, Aircraft and Towering are a handful of very large units at the top of the price range, where a log-linear fit is least sure, so part of their premium may be the curve rather than the keyword. "Support" is BSData's Supporting link (a datasheet BSData lets support another unit): cheap for what its own dice do. Feel No Pain's discount is against the durability the dice already give it (section 7). "Can grow" is a squad that can be taken larger: its minimum size is priced lower per body than a fixed-size unit with the same numbers.
3. The families jointly
"Carriers' price": over the datasheets carrying any flag of the family, the median of what its flags together add (exp Σ β x − 1). "One SD": the spread of the whole family's contribution (its numbers included) across datasheets within an army, as a price: how much of a unit's points the family moves. Ranges from the bootstrap.
| Family | Flags and counts | Numbers | Carriers | Carriers' price | One SD |
|---|
| movement and arrival | Deep Strike, Scouts, Infiltrators, Fly, own arrival rule, own extra movement | – | 423 | +5% (+2% to +8%) | +4% (+4% to +6%) |
| defensive rules | Stealth, Lone Operative, own defensive dice effects, heals or returns models | Feel No Pain | 140 | +5% (+1% to +8%) | +3% (+2% to +5%) |
| statline | Damaged bracket | Toughness, armour save, invulnerable save, wounds per model, Leadership | 196 | +5% (+2% to +8%) | +9% (+7% to +10%) |
| objective | Battleline, objective rule, command points rule | OC per model | 176 | +2% (−2% to +6%) | +3% (+2% to +4%) |
| weapon keywords | Lethal Hits, Sustained Hits, Devastating Wounds, Anti-X, Precision, Indirect Fire, Torrent, Blast, Melta, Twin-linked, Ignores Cover, Psychic weapon, Heavy, Assault, Pistol, Lance, Hazardous, Rapid Fire, Extra Attacks, One Shot | – | 741 | −0% (−1% to +2%) | +3% (+3% to +5%) |
| leader and support | Leader, datasheets it leads, Support, Aura, leader dice effects, Transport, transport capacity, Synapse | – | 373 | −1% (−3% to +2%) | +3% (+2% to +5%) |
| melee and other rules | Fights First, Deadly Demise, mortal wounds rule, Battle-shock rule, hinders enemies, better Overwatch | – | 494 | +1% (−3% to +4%) | +2% (+1% to +3%) |
| keywords | Character, Epic Hero, Monster, Vehicle, Infantry, Walker, Mounted, Beast, Swarm, Psyker, Titanic, Towering, Aircraft, Daemon, Terminator | – | 822 | +2% (−1% to +4%) | +8% (+7% to +9%) |
| size | can grow, mixed statlines | models at minimum | 222 | −9% (−11% to −6%) | +3% (+2% to +4%) |
| Legends | Legends | – | 243 | +5% (+3% to +7%) | +2% (+1% to +3%) |
4. Collinear groups, priced jointly
Variables whose within-army deviations correlate at |r| ≥ 0.6, joined through any member. Their single prices trade off against each other (a Monster's Toughness, wounds and durability rise together, and with the points), so read the group: "one SD" is the spread of the group's combined contribution, which is well identified even where its members' prices aren't.
| Group | One SD of the group | Members' single prices |
|---|
| damage into CHAFF, damage into MEQ, damage into TEQ, damage into LV, damage into HV, durability vs small, durability vs special, durability vs plasma, durability vs elite, durability vs at, total OC, Toughness, armour save, wounds per model, Damaged bracket, OC per model, Deadly Demise, Vehicle, Infantry, models at minimum, can grow | +63% (+60% to +66%) | damage into CHAFF +3%; damage into MEQ +5%; damage into TEQ +6%; damage into LV +5%; damage into HV +4%; durability vs small +4%; durability vs special +5%; durability vs plasma +5%; durability vs elite +6%; durability vs at +7%; total OC −2%; Toughness +1%; armour save +0%; wounds per model +2%; Damaged bracket +5%; OC per model +0%; Deadly Demise +1%; Vehicle −4%; Infantry +1%; models at minimum +2%; can grow −9% |
| Leader, datasheets it leads, leader dice effects, Character | +2% (+1% to +4%) | Leader −3%; datasheets it leads +1%; leader dice effects +5%; Character +2% |
| Psychic weapon, Psyker | +1% (+0% to +2%) | Psychic weapon +4%; Psyker +2% |
| Transport, transport capacity | +1% (+0% to +2%) | Transport +4%; transport capacity −5% |
| Torrent, Ignores Cover | +0% (+0% to +1%) | Torrent −1%; Ignores Cover +0% |
5. Under- and over-priced units, by the designers' own pattern
Each unit's points against what the other 15 armies' pattern predicts for a unit with its numbers and perks, its own army's price level kept ("pattern"). Negative: under-priced (it costs less than the pattern says). What the pattern can't see (an aura's worth, what a leader adds, scoring and screening, a rule the effects files only note) lands here too, so a large residual is a question, not a verdict. "Ledger caps": the residual with the ledger's capabilities (rank correlation with the main residual over the ledger units 0.85).
Tyranids (every datasheet, most under-priced first)
| Unit | Points | Pattern | Residual | Ledger caps | Letters (auspexFull, hivemindFull, secondFull, maelstrom, astrategas) | Consensus | Ledger rank |
|---|
| Lictor | 60 | 91 | −34% | −35% | SASSS | 4.8 | 3 |
| Pyrovores | 40 | 54 | −26% | −22% | A··BB | 3.3 | 30 |
| Deathleaper | 80 | 107 | −25% | −29% | ASABD | 3.4 | 6 |
| Tyrannofex | 170 | 222 | −23% | −6% | SAAAS | 4.4 | 14 |
| Haruspex | 125 | 159 | −21% | −20% | BBAAB | 3.4 | 13 |
| Toxicrene | 120 | 151 | −21% | −16% | C··DD | 1.3 | 35 |
| Tyrannocyte | 80 | 99 | −19% | −21% | C··DD | 1.3 | 48 |
| Von Ryan's Leapers | 55 | 68 | −19% | −21% | SBABC | 3.4 | 2 |
| Carnifexes | 90 | 109 | −17% | −11% | CCCCC | 2.0 | 20 |
| Tyranid Warriors with Ranged Bio-Weapons | 60 | 73 | −17% | −21% | CCCDC | 1.8 | 27 |
| Hive Guard | 80 | 95 | −16% | −10% | ABACC | 3.0 | 28 |
| Trygon | 140 | 166 | −16% | −15% | BCBBA | 3.0 | 12 |
| Zoanthropes | 90 | 106 | −15% | −12% | SABBB | 3.6 | 15 |
| Screamer-killer | 125 | 145 | −14% | −7% | BBBCB | 2.8 | 18 |
| Exocrine | 135 | 156 | −13% | −10% | AAASS | 4.4 | 9 |
| Venomthropes | 55 | 63 | −13% | −10% | BDBCD | 2.0 | 36 |
| Barbgaunts | 55 | 61 | −10% | −8% | CCBCD | 2.0 | 29 |
| Tyrant Guard | 80 | 89 | −10% | −2% | BBACC | 2.8 | 40 |
| Neurogaunts | 45 | 49 | −8% | −9% | BCCCC | 2.2 | 42 |
| Mucolid Spores | 30 | 32 | −6% | +1% | DDDFD | 0.8 | 51 |
| Tyranid Prime with Lash Whip | 75 | 80 | −6% | −6% | SAS·D | 3.8 | 17 |
| Neurolictor | 80 | 84 | −4% | −5% | SSS·S | 5.0 | 11 |
| Termagants | 60 | 62 | −3% | +9% | BBCBA | 3.0 | 38 |
| Tervigon | 160 | 165 | −3% | −8% | BCCCD | 2.0 | 41 |
| Maleceptor | 180 | 184 | −2% | −10% | SAAAA | 4.2 | 34 |
| Tyranid Warriors with Melee Bio-Weapons | 75 | 77 | −2% | −6% | BBBBB | 3.0 | 24 |
| Old One Eye | 140 | 141 | −1% | +13% | BBABC | 3.0 | 21 |
| Ripper Swarms | 30 | 30 | −1% | +18% | BCBDD | 2.0 | 50 |
| Hive Crone | 170 | 171 | −0% | −5% | DDDFD | 0.8 | 37 |
| Broodlord | 80 | 80 | −0% | −1% | AAAAB | 3.8 | 25 |
| Mawloc | 135 | 135 | +0% | +16% | BBCBC | 2.6 | 33 |
| Winged Tyranid Prime | 65 | 64 | +1% | −4% | CBBCB | 2.6 | 43 |
| The Red Terror | 130 | 128 | +2% | +1% | AAS·· | 4.3 | 22 |
| Norn Emissary | 250 | 246 | +2% | −5% | SBSAA | 4.2 | 31 |
| The Swarmlord | 210 | 204 | +3% | +6% | SSA·A | 4.5 | 4 |
| Norn Assimilator | 250 | 239 | +5% | +8% | SBSBC | 3.6 | 19 |
| Genestealers | 75 | 72 | +5% | −5% | AAAAB | 3.8 | 23 |
| Biovores | 60 | 57 | +6% | +5% | SSSSS | 5.0 | 1 |
| Psychophage | 110 | 100 | +10% | +17% | BBABC | 3.0 | 44 |
| Hive Tyrant | 195 | 176 | +11% | +13% | SASAA | 4.4 | 5 |
| Raveners | 125 | 113 | +11% | −1% | SBABA | 3.8 | 16 |
| Winged Hive Tyrant | 185 | 165 | +12% | +16% | ABBBB | 3.2 | 8 |
| Harpy | 185 | 164 | +13% | +4% | DDDFD | 0.8 | 46 |
| Parasite of Mortrex | 70 | 62 | +13% | +15% | BBCCD | 2.2 | 47 |
| Hormagaunts | 70 | 59 | +20% | +6% | SABAS | 4.2 | 10 |
| Hyperadapted Raveners | 165 | 138 | +20% | +11% | SAAAS | 4.4 | 7 |
| Neurotyrant | 130 | 108 | +21% | +42% | A··AB | 3.7 | 39 |
| Hierophant | 810 | 573 | +41% | +18% | DCCD· | 1.5 | 32 |
| Gargoyles | 80 | 55 | +45% | +41% | BBBAS | 3.6 | 45 |
| Harridan | 610 | 413 | +48% | +34% | DCCD· | 1.5 | 26 |
| Sporocyst | 145 | 97 | +49% | +45% | DDCD· | 1.3 | 49 |
| Spore Mines | 55 | 34 | +59% | +103% | DDDFD | 0.8 | 52 |
The other armies (four each way, Legends left out)
- Space Marines. Under: Impulsor 70 (pattern 102, −31%); Aggressor Squad 80 (pattern 108, −26%); Invader ATV 60 (pattern 77, −23%); Techmarine 55 (pattern 69, −20%). Over: Thunderhawk Gunship 840 (pattern 481, +75%); Lieutenant with Combi-weapon 95 (pattern 56, +71%); Desolation Squad 180 (pattern 125, +44%); Stormraven Gunship 280 (pattern 196, +43%).
- Adeptus Mechanicus. Under: Cybernetica Datasmith 25 (pattern 40, −37%); Servitor Battleclade 65 (pattern 99, −34%); Sydonian Dragoons with taser lances 60 (pattern 73, −18%); Sydonian Dragoons with radium jezzails 55 (pattern 66, −16%). Over: Archaeopter Transvector 145 (pattern 97, +49%); Archaeopter Fusilave 160 (pattern 108, +48%); Kataphron Breachers 150 (pattern 102, +47%); Archaeopter Stratoraptor 185 (pattern 132, +40%).
- Agents of the Imperium. Under: Inquisitorial Chimera 60 (pattern 88, −32%); Rogue Trader Entourage 75 (pattern 105, −28%); Inquisitorial Agents 50 (pattern 68, −26%); Sisters of Battle Immolator 90 (pattern 103, −13%). Over: Vindicare Assassin 110 (pattern 63, +76%); Inquisitor Kroyle 100 (pattern 65, +54%); Navigator 60 (pattern 42, +44%); Watch Master 95 (pattern 73, +30%).
- Astra Militarum. Under: Krieg Command Squad 60 (pattern 90, −33%); Ogryn Bodyguard 40 (pattern 58, −31%); Taurox Prime 75 (pattern 106, −29%); Nork Deddog 60 (pattern 83, −28%). Over: Ursula Creed 85 (pattern 50, +69%); Commissar Yarrick 120 (pattern 75, +60%); Aegis Defence Line 145 (pattern 95, +52%); Manticore 150 (pattern 117, +28%).
- Blood Angels. Under: Death Company Marines with Bolt Rifles 80 (pattern 110, −27%); Death Company Intercessors 75 (pattern 102, −27%); Death Company Dreadnought 150 (pattern 198, −24%); Death Company Captain 70 (pattern 89, −22%). Over: Sanguinary Priest 75 (pattern 44, +69%); Lemartes 100 (pattern 83, +21%); Commander Dante 125 (pattern 117, +7%); The Sanguinor 120 (pattern 122, −1%).
- Dark Angels. Under: Ravenwing Darkshroud 70 (pattern 90, −22%); Sammael 105 (pattern 128, −18%); Ravenwing Black Knights 75 (pattern 90, −17%); Belial 75 (pattern 87, −14%). Over: Lion El'Jonson 265 (pattern 203, +31%); Azrael 140 (pattern 112, +25%); Ravenwing Dark Talon 200 (pattern 180, +11%); Deathwing Terminator Squad 165 (pattern 149, +10%).
- Death Guard. Under: Malignant Plaguecaster 60 (pattern 85, −30%); Lord of Poxes 65 (pattern 82, −21%); Typhus 100 (pattern 120, −17%); Nurglings 45 (pattern 53, −14%). Over: Poxwalkers 65 (pattern 51, +27%); Plaguebearers 115 (pattern 92, +25%); Miasmic Malignifier 105 (pattern 88, +20%); Defiler 300 (pattern 251, +19%).
- Drukhari. Under: Cronos 55 (pattern 90, −39%); Talos 75 (pattern 115, −35%); Wyches 90 (pattern 118, −24%); Wracks 60 (pattern 77, −22%). Over: Solitaire 115 (pattern 68, +69%); Lelith Hesperax 80 (pattern 58, +38%); Troupe Master 75 (pattern 58, +30%); Incubi 90 (pattern 74, +21%).
- Emperor's Children. Under: Sorcerer 55 (pattern 70, −21%); Maulerfiend 120 (pattern 143, −16%); Chaos Rhino 70 (pattern 82, −14%); Lord Kakophonist 70 (pattern 81, −14%). Over: Lord Exultant 90 (pattern 69, +30%); Daemon Prince of Slaanesh with Wings 205 (pattern 174, +18%); Daemon Prince of Slaanesh 170 (pattern 149, +14%); Seekers 85 (pattern 75, +13%).
- Genestealer Cults. Under: Goliath Truck 75 (pattern 107, −30%); Goliath Rockgrinder 120 (pattern 167, −28%); Patriarch 80 (pattern 107, −26%); Locus 35 (pattern 47, −25%). Over: Nexos 60 (pattern 36, +65%); Clamavus 50 (pattern 35, +44%); Biophagus 50 (pattern 36, +37%); Benefictus 75 (pattern 57, +31%).
- Leagues of Votann. Under: Brôkhyr Iron-master 70 (pattern 101, −31%); Buri Aegnirssen 85 (pattern 110, −22%); Einhyr Champion 65 (pattern 83, −22%); Ironkin Steeljacks with Melee Weapons 75 (pattern 93, −19%). Over: Arkanyst Evaluator 70 (pattern 53, +33%); Hekaton Land Fortress 250 (pattern 191, +31%); Cthonian Beserks 95 (pattern 74, +28%); Hernkyn Yaegirs 90 (pattern 75, +20%).
- Orks. Under: Runtherd 10 (pattern 31, −68%); Deffkoptas 80 (pattern 113, −29%); Gorkanaut 325 (pattern 433, −25%); Zodgrod Wortsnagga 50 (pattern 62, −20%). Over: Big'ed Bossbunka 135 (pattern 75, +81%); Squighog Boyz 140 (pattern 90, +56%); Rukkatrukk Squigbuggies 85 (pattern 61, +40%); Warboss 100 (pattern 75, +34%).
- Space Wolves. Under: Iron Priest 50 (pattern 65, −23%); Logan Grimnar 100 (pattern 125, −20%); Wulfen with Storm Shields 100 (pattern 112, −11%); Wolf Guard Battle Leader 65 (pattern 70, −7%). Over: Wolf Scouts 90 (pattern 64, +42%); Grey Hunters 165 (pattern 128, +29%); Blood Claws 135 (pattern 112, +20%); Wolf Priest 70 (pattern 59, +19%).
- Thousand Sons. Under: Maulerfiend 120 (pattern 155, −23%); Daemon Prince of Tzeentch with wings 170 (pattern 218, −22%); Screamers 80 (pattern 99, −20%); Flamers 65 (pattern 80, −19%). Over: Sorcerer 95 (pattern 61, +55%); Infernal Master 100 (pattern 65, +54%); Scarab Occult Terminators 180 (pattern 138, +30%); Pink Horrors 115 (pattern 93, +23%).
- World Eaters. Under: Bloodletters 90 (pattern 117, −23%); Lord Invocatus 100 (pattern 122, −18%); Master of Executions 60 (pattern 71, −16%); Chaos Rhino 75 (pattern 87, −14%). Over: Chaos Spawn 95 (pattern 74, +28%); Khârn the Betrayer 115 (pattern 96, +20%); Eightbound 125 (pattern 104, +20%); Khorne Berzerkers 160 (pattern 134, +20%).
6. Do under-priced units get better letters?
The yardstick behind Track B's 77.7%: tools/ledgerstep.js shape() per list (tiers S to D, pairs in different tiers, a tie in the score wrong), the mean over the lists. Scores: "Residual", −residual alone (under-priced ranks high); "Over-priced", the opposite sign; "Blend", z(log Value at Track B) + z(−residual), equal weights, fixed before looking; "Log blend", log Value − residual (Value per point times the designers' mispricing, also fixed before looking); "Cheaper", fewer points ranks high (a baseline). The reviewers' ceiling on Tyranids: about 89% against their consensus, about 84% one to one.
Tyranids, the five lists
| Score | auspexFull | hivemindFull | secondFull | maelstrom | astrategas | Mean |
|---|
| Track B ledger | 80.4% | 81.0% | 77.3% | 74.5% | 75.3% | 77.7% |
| Residual alone | 53.0% | 54.5% | 57.5% | 46.1% | 43.5% | 50.9% |
| Residual (in-sample fit) | 51.5% | 53.3% | 55.6% | 45.6% | 43.1% | 49.8% |
| Residual (ledger caps) | 56.3% | 56.6% | 58.0% | 50.1% | 47.0% | 53.6% |
| Over-priced | 47.0% | 45.5% | 42.5% | 53.9% | 56.5% | 49.1% |
| Ledger + residual (blend) | 69.1% | 71.2% | 72.1% | 62.2% | 60.2% | 67.0% |
| Ledger + residual (ledger caps) | 70.5% | 72.4% | 71.1% | 63.8% | 63.1% | 68.2% |
| Ledger + residual (log blend) | 79.0% | 79.5% | 76.7% | 73.4% | 72.4% | 76.2% |
| Cheaper | 43.7% | 44.3% | 42.3% | 42.7% | 37.7% | 42.1% |
- Paired unit bootstrap (400 draws of the 52 units), against the ledger: blend −11.2 pts ± 3.9 pts (90%: −17.9 pts to −5.1 pts); log blend −1.7 pts ± 1.6 pts (90%: −4.2 pts to +0.9 pts); residual alone −27.6 pts ± 6.5 pts (90%: −38.3 pts to −16.8 pts).
- The blend's weight scanned (in sample, so optimistic; the fixed weight 1 is the honest number): w 0.25: 75.4%, w 0.5: 71.3%, w 1: 67.0%, w 2: 61.0%, w 4: 57.0%. No weight beats the ledger alone.
- Spearman with the consensus (S 5 … F 0) over the 52 graded units: residual 0.01 (ledger caps 0.07), ledger 0.72; residual with ledger 0.23.
- The Tyranid table above shows why: the most under-priced by the pattern include S units (Lictor, Tyrannofex) and D units (Toxicrene, Tyrannocyte) alike, and the most over-priced include S to B units whose worth is scoring, screening or support (Gargoyles, Hormagaunts, Hyperadapted Raveners, Neurotyrant) beside the D-rated Sporocyst, Harridan and Spore Mines.
Every army with a panel (Track B's settings; the other armies are never tuned)
| Army | Lists | Units | Ledger | Residual | Over-priced | Blend | Log blend | Cheaper | ρ residual–consensus |
|---|
| Tyranids | 5 | 52 | 77.7% | 50.9% | 49.1% | 67.0% | 76.2% | 42.1% | 0.01 |
| Space Marines | 4 | 84 | 60.2% | 51.6% | 48.4% | 58.2% | 60.7% | 53.6% | -0.00 |
| Adeptus Mechanicus | 1 | 34 | 67.4% | 67.4% | 32.6% | 68.0% | 68.6% | 41.9% | 0.39 |
| Agents of the Imperium | 1 | 29 | 63.7% | 36.3% | 63.7% | 52.0% | 60.5% | 36.3% | -0.32 |
| Astra Militarum | 1 | 72 | 69.2% | 60.7% | 39.3% | 68.5% | 70.4% | 25.9% | 0.25 |
| Blood Angels | 1 | 20 | 56.5% | 33.9% | 66.1% | 45.2% | 55.6% | 33.1% | -0.38 |
| Dark Angels | 1 | 93 | 51.6% | 43.7% | 56.3% | 46.8% | 49.8% | 44.4% | -0.22 |
| Death Guard | 1 | 36 | 48.1% | 54.6% | 45.4% | 48.3% | 47.7% | 50.0% | 0.11 |
| Drukhari | 1 | 37 | 62.3% | 45.4% | 54.6% | 50.7% | 59.9% | 44.0% | -0.07 |
| Emperor's Children | 1 | 23 | 68.1% | 55.5% | 44.5% | 57.6% | 64.4% | 50.3% | 0.12 |
| Genestealer Cults | 2 | 90 | 51.8% | 56.3% | 43.7% | 53.8% | 52.6% | 35.1% | 0.14 |
| Leagues of Votann | 2 | 22 | 61.9% | 35.2% | 64.8% | 39.2% | 53.7% | 29.8% | -0.32 |
| Orks | 1 | 55 | 64.4% | 51.0% | 49.0% | 60.9% | 63.2% | 38.6% | 0.02 |
| Space Wolves | 1 | 21 | 49.3% | 64.3% | 35.7% | 55.0% | 52.1% | 47.1% | 0.13 |
| Thousand Sons | 1 | 35 | 40.8% | 40.8% | 59.2% | 41.1% | 41.7% | 42.1% | -0.22 |
| World Eaters | 3 | 30 | 63.7% | 42.0% | 58.0% | 53.2% | 61.0% | 57.3% | -0.27 |
| Mean of the other 15 | | | 58.6% | 49.2% | 50.8% | 53.2% | 57.5% | 42.0% | -0.04 |
Most of the other armies have one list each (a pairwise SE of 4 to 6 points), so single armies swing; the mean is the reading. Neither sign of the residual carries the reviewers' order, and adding it to the ledger lowers every mean.
7. The family prices on the dial scale (a starting point for T-662)
T-662's dials read 1× as "what the engine already credits". Three cases:
(a) Perks the ledger credits with a rule multiplier or a switch. The ledger's credit: what the perk adds to a carrier's Value now (Track B; the toggle shown; Value before ÷ after − 1, the median over the carriers in the 16 armies' ledgers, with the Tyranid median beside it). The designers' price: what a carrier costs above an equal unit without the perk (for support, its leader, aura, Synapse and Support flags together, over the same carriers). If a dial scales the perk's increment, the designers' dial is price ÷ credit; if it multiplies the carrier's whole Value, (1 + price) ÷ (1 + credit).
| Perk | Toggle | Carriers (moved) | Ledger's credit: median; on those moved; Tyranids | Designers' price | Dial, increment | Dial, multiplier |
|---|
| Fights First | rule kind first (× 2.64 on melee Kill) to 1 | 15 (14) | +144.2%; +145.8%; +126.3% | −8% (−17% to +3%) | -0.06× (-0.12 to 0.02) | 0.38× |
| Lone Operative | rule kind lone (× 1.93 on Soak) to 1 | 31 (31) | +1.2%; +1.2%; +0.8% | +10% (+2% to +16%) | 8.13× (1.97 to 13.53) | 1.08× |
| Support (leaders, auras, Synapse, Support) | supportMeasuredW 0 (the measured support credit off) | 263 (63) | +0.0%; +52.8%; +10.6% | −3% (−5% to +1%) | -0.05× on the units it moves; the median carrier has no credit | 0.97× |
| heals or returns models | soakfx 0 (own heals and turns off the table on Soak) | 52 (2) | +0.0%; +1.8%; +1.6% | +3% (−0% to +7%) | not meaningful (it moves 2 units) | 1.03× |
Synapse, command points, heals, buffs and debuffs take × 1 under Track B's supportMeasured (their worth is the measured support credit, the third row). Extra movement, mortal wounds and Overwatch take × 1 at Track B: the ledger credits them nothing, the designers +1%, +3% and −2%; both say "small". The soakfx row moves almost no unit outside Tyranids (the exposure and own-heal fields are Tyranid data so far).
(b) Perks the dice engine plays (weapon keywords, Feel No Pain). The engine's credit: the change in the engine's numbers when the perk is taken off (docs/ev.js, frozen; the default loadout into the five targets; for Feel No Pain the factor it puts on hits to destroy), priced at the designers' own rate for those numbers; the direct price is what they charge on top (negative: less than the dice give). The dial, 1× = the dice as the designers price dice, is (credit + direct) ÷ credit, the median carrier's. Rankings' neutral table plays no half range, cover, charge or character target, so Melta, Rapid Fire, Lance, Ignores Cover, Precision and Indirect Fire get no dice credit there and no dial.
| Perk | Carriers | Engine's credit | Direct (on top) | Total | Dial (range) |
|---|
| Feel No Pain | 62 | +13.9% | −9.5% | +3.1% | 0.23× (-0.12 to 0.58) |
| Lethal Hits | 99 | +3.1% | −2.8% | +0.2% | 0.08× (-1.56 to 1.65) |
| Sustained Hits | 156 | +2.1% | −1.2% | +0.9% | 0.44× (-0.57 to 1.68) |
| Devastating Wounds | 133 | +4.6% | −0.3% | +4.2% | 0.93× (0.31 to 1.48) |
| Anti-X | 133 | +1.2% | +2.4% | +3.6% | 3.06× (0.82 to 5.59) |
| Twin-linked | 154 | +3.7% | +1.5% | +5.2% | 1.42× (0.68 to 2.23) |
| Torrent | 83 | +4.2% | −0.8% | +3.3% | 0.79× (-0.19 to 1.75) |
| Blast | 146 | +1.8% | +4.4% | +6.2% | 3.42× (1.97 to 5.13) |
| Heavy | 68 | +3.9% | −0.5% | +3.4% | 0.88× (-0.17 to 1.79) |
| Extra Attacks | 64 | +5.5% | −0.1% | +5.3% | 0.98× (0.26 to 1.71) |
| Melta | 28 | +0.0% | −0.4% | −0.4% | not meaningful (the table is never in half range) |
| Rapid Fire | 167 | +0.0% | −1.8% | −1.8% | not meaningful (the table is never in half range) |
| Lance | 22 | +0.0% | −3.5% | −3.5% | not meaningful (the table never charges) |
| Ignores Cover | 83 | +0.0% | +0.5% | +0.5% | not meaningful (the table has no cover) |
| Precision | 69 | +0.0% | +0.7% | +0.7% | not meaningful (no character target in the table) |
| Indirect Fire | 31 | +0.0% | +7.3% | +7.3% | not meaningful (the table has line of sight) |
| Weapon keywords as one family (Σ over carriers and keywords) | | | | | 1.12× (0.84 to 1.44) |
(c) Not meaningful as a ratio. Arrival (Deep Strike +2% (−1% to +5%), Scouts +1% (−2% to +5%), Infiltrators +3% (−3% to +8%), own arrival rule +6% (+3% to +10%)) is measured inside the charge model's lands and threat, and Stealth (+5%) inside Soak (tools/matrix.js plays it as cover): no toggle isolates their increment, so there is no 1× to divide by. If a family dial is wanted for them, the designers' price is an additive credit, in points per 100. Fly (+10%) is a keyword the ledger reads through Move and the charge model. The statline numbers are the engine's own (durability); their direct prices say how far the designers' curve bends from ours, not a dial. Unit types, size and Legends are not perks a dial should scale.
The families on the dial scale, in one table
| Family | Designers' price | On the dial scale | Meaningful? |
|---|
| Movement and arrival | carriers +5%; Deep Strike +2%, Scouts +1%, Fly +10% | – | no: measured inside the charge model; an additive credit |
| Defensive rules | Lone Operative +10%, Stealth +5%, Feel No Pain +3% in all | Lone Operative 8.13× (increment) or 1.08× (multiplier); Feel No Pain 0.23× | yes for those two; Stealth no (inside Soak) |
| Weapon keywords | the family −0% at the carriers on top of the dice | 1.12× (0.84 to 1.44) | yes, for the keywords the table plays |
| Fights First | −8% | -0.06× (increment) or 0.38× (multiplier) | yes; 15 carriers, a wide range |
| Leader and support | −3% | -0.05× on the units the credit moves | partly: the median carrier has no credit |
| Objective, OC | Battleline +2%, command points +8%, OC per model +0% | – | no: Hold and Score are lines with weights, not perk dials |
| Synapse | +4% (18 carriers, Tyranids only) | – | no: held at 1 under supportMeasured; one army's carriers |
How to use these: as the centre of a prior for each family dial (say log-dial ~ Normal(log of the designers' dial, 0.35)), never as values to adopt. A dial far below 1× here (Fights First, support) says the points don't pay for what the ledger credits; the ledger's larger credits were kept because they moved the Tyranid lists, and that tension is the dial search's question, not this report's.
8. The ledger's own capabilities (the cross-check)
The same fit on the 586 datasheets the ledgers carry, with the ledger's numbers for capabilities: Kill into four groups of its eight target classes at β 6 (horde; infantry: elite, heavy, cavalry and beasts; armour: monster, heavy and light vehicle; character), Soak against shooting and in melee, OC (all absolute, the best over the unit's minimum-size forms, its loadouts and detachments), and Move. Penalty λ = 30. Left-out R² 87.4% (the rankings' numbers on the same units 88.3%; the ledger's capabilities alone 84.1%). Perk prices rank-correlate 0.66 with the main fit's (column "Ledger caps" in section 2); residuals 0.85. The ledger's Kill and Soak price the points about as well as the rankings' table: neither is closer to how the designers think.
9. The biases, named
- Whose game. Points are set for every kind of play (casual, narrative, crusade and matched), with model ranges in mind, not only for tournaments. A perk that matters most at the top tables (Lone Operative, scoring bodies, Fights First in a melee meta) can be cheap in points and dear in letters.
- When. Points answer the previous balance pass: a unit priced up after it dominated shows as over-priced here and may still be strong, and one priced down after a bad season as under-priced. Points move every few months; this list is at BSData 374f505 and goes stale with the next dataslate (re-run it after a deliberate re-freeze of the ledger inputs).
- Steps. Points come in steps of 5 on fixed squad sizes: a 5% price is one step on a 100-point unit, so small prices sit at the edge of what points can say.
- Whole datasheets. The designers price datasheets, not perks; the regression splits a datasheet's price among its perks by how they co-occur. Perks that travel together (Monster with Toughness, wounds and durability; Leader with Character; Transport with capacity; Psyker with psychic weapons) can only be priced jointly (section 4).
- Our engine in the middle. "At equal capabilities" means equal by our engine. A perk the engine undercounts (an aura, a leader's bonus, Stealth: the rankings' table has no cover) shows as a premium, one it overcounts as a discount. The residuals carry the same mix (section 6).
- Shrinkage. Ridge pulls every price toward zero, rare perks most: a small price with a wide range means "the points can't say", not "worthless".
- Army context. Within-army deviations price a perk against its own army's units; a perk nearly every unit of an army has (Synapse on Tyranids, an army rule) is priced only where some lack it, and an army's overall price level (its points against the pattern) is set aside.
- Legends and Crucible. 243 Legends datasheets are in, with a flag (+5% against the pattern: their points are maintained less carefully); without them the prices barely move (section 1). The 56 Crucible datasheets are left out.
Tool: tools/pricelist.js (about 39 s). Numbers: reports/tiers/2026-10-05-pricelist.json.
