Banner, scored from effects (2026-10-02)

T-527, for T-523 (Jordan, 1 Oct (Pacific), via the design session). Made by node tools/banner.js. A report, not the site: nothing here changes a published score or page.

In short

The method works where we have effects, and that's about half the characters. Of 451 non-Legends characters:

• 184 have leader effects written as dice (127 offensive, 44 defensive, 13 both);
• 129 have an entry with no dice effects (the rule exists, but it does something we can't roll);
• 138 have no entry at all.

291 lead or support a squad we could score. 11th edition's Support characters, the Apothecary Biologis among them, join squads as Leaders do; the first run missed all 48 of them, and the public list's comparison caught it.

• Most leaders beat bodies: 203 of 291 add more than spending their points on more models of their best squad would. The best do far more. Junith Eruita (with Retributors) and Chief Librarian Tigurius (with Desolators) more than triple what bodies would add. Kravek Morne lifts Obliterators' damage 2.3-fold, Typhus Poxwalkers' 4.5-fold. The worst are command squads, Troupe Masters and Shadowseers. Their worth is in rules we don't model: orders, the Shadowseer's protection, the Troupe Master's re-rolls on a choice.
• Agreement with what players take is still weak, but better than today's measure. Among characters, ρ with list share is 0.060, against 0.052 for today's Utility support part (0.143 for Value). Fed into the tier test's V2, the new Banner lifts V2's ρ from 0.113 to 0.127 and moves Mortarion from C to B. The Apothecary Biologis, a public list's best support character, scores low here (0.62, 28th percentile): its Lethal Hits on Eradicators adds less than 70 pts of more Eradicators would.
• The gap list is the real output. 85 characters taken in winning lists lead squads but have no dice effects yet. Logan Grimnar (10 of 11 Space Wolves lists), Lucius (17 of 22), the Coldstar Commander (17 of 24), the Dawneagle Shield-Captain (19 of 29) and the Swarmlord (13 of 23) head it. That's the next hand-writing batch, as T-343 did for the first. Until they're written, the method can't see why players take them.
• Choices made:
  • Default wargear for squad and character.
  • "Bodies" means the squad's next size up, or its own rate per point where it has only one size.
  • An aura is worth half the median leader's gain over a bare body, across 2 units in range (a proxy).
  • A transport counts a quarter of a squad's damage (a proxy).
  • Army-wide effects and CP generation aren't modelled.

The method

For each character and each squad it can lead or support (BSData's Leading and Supporting links; 11th edition's Support characters, like the Apothecary Biologis, join a squad as Leaders do):

• Damage: the joined unit's wounds removed per turn into the five standard targets (GEQ, MEQ, TEQ, a light and a heavy vehicle; the better of shooting and melee; default wargear), the character's own attacks included, with its effects from leaderfx.json (our own words) switched on, against the squad alone.
• Durability: the hits it takes to destroy the joined unit from the five standard attackers, with the character's defensive effects on the squad (Feel No Pain, an invulnerable save, −1 to hit or to wound, AP or damage worsened, cover, a better save) and the character's own wounds behind it, against the squad alone. A character with revives counts 15% more.
• Bodies: the same points spent on more models of the squad: its next larger size where the datasheet has one (the points actually paid for the models added), otherwise its own damage and durability per point at its smallest size.
• Multiplier vs bodies: the average of damage added ÷ what bodies would add, and durability added ÷ what bodies would add. Above 1, the character beats bodies.
• Banner score: the best squad's multiplier (the squad is recorded), percentiled over every character scored.

What's modelled and what's a proxy.

• Modelled: each character's leader effects on its squad, its own attacks and wounds, and bodies.
• Proxy, an aura: worth the median leader's multiplier above a bare body, shared over 2 units in range (stated, not measured: our data has no aura effects yet).
• Proxy, a transport: delivery counts 0.25 (a quarter of a squad's damage).
• Not modelled: army- and detachment-wide effects (×3 in the brief) and CP generation. Neither is in our data as dice effects; both wait for hand-written entries.
• Loadout: default wargear for the squad and the character (Offense uses each target's best squad; here the comparison is like for like).

Coverage

291 characters lead at least one squad we could score; 160 lead none (lone heroes, or squads missing from the data) and score on proxies only.

The next hand-writing batch (like T-343): characters taken in winning lists that lead squads but have no entry in leaderfx.json or one without dice effects, most taken first (85):

Agreement with what players take

Spearman's ρ with the share of winning lists, 426 characters some 5 or more lists could field: the new Banner score 0.060, today's Utility support part 0.052, today's Value 0.143.

The top 20

The bottom 20 (of those that lead a squad)

Beats bodies?

203 of 291 leaders beat spending their points on more models of their best squad (multiplier above 1). On damage alone 214, on durability alone 83.

Mortarion

Mortarion leads no squad, so his Banner is the aura proxy only: 1 aura, score 0.8, 33th percentile of characters. This is the weakest part of the method. His Lord of the Death Guard buffs reach every Death Guard unit near him and aren't in our data as effects, so neither the aura proxy nor anything else here sees them. Writing them as our own effects (an entry like leaderfx.json's, but for nearby units) is what would let this method measure him.

Every character