How Damage Works

The stacking model behind every fight: bonuses add within a category, multiply across categories — and everything else follows.

The Codex

The model: additive within, multiplicative across

[P] Paper model

Lilith has never published the literal damage equations, so everything here is a working model that reproduces observed in-game behaviour. Its one governing principle: percentage bonuses of the same type are SUMMED into a single category total, and the resulting category multipliers are then MULTIPLIED against each other.

A = A₀ × (1 + Σaᵢ) D = D₀ × (1 + Σdⱼ)
Effective attack and defence: base stat times one plus the summed category bonuses
[P] Paper model

Every attack% you own — academy tech, commander skills, talents, gear, titles, VIP, runes, civilisation and alliance buffs — lands in the same sum. A worked archer-nuke march (Qin Shi Huang + Yi Seong-Gye):

SourceAttack bonus
Academy attack tech+25%
Commander + talents+15%
Legendary archer gear set+46%
Blue attack rune+10%
VIP+5%
Alliance / title buff+9%
Category total+110%
With a nominal base of 100,000, effective attack is 100,000 × (1 + 1.10) = 210,000.
[P] Paper model
Why this one rule matters

Everything else in this chapter — diminishing returns, the value of shred, why skill damage is special — is a direct consequence of additive-within, multiplicative-across. Learn the rule and the rest becomes arithmetic.

The lanes of a strike

[P] Paper model

Damage rises with your effective attack measured against the defender’s effective defence, then picks up its cross-category multipliers. Skill damage occupies its own lane: it multiplies active-skill hits only, never normal attacks or counterattacks.

Dmg_skill = f(A, D) × DF × (1 + Σsₖ) Dmg_normal = f(A, D)
f(A,D) is the damage kernel (rises with A, falls with D); DF is the skill’s fixed Damage Factor
[P] Paper model
Skill Damage ≈ Damage Factor × Troop Attack × Troop Number
The community-documented baseline kernel — a simplification, but it captures the dominant terms
[C] Community

The Damage Factor is set by the skill’s level and is independent of the enemy’s troop count. Two practical consequences: skill-damage% is gold on a nuker with a big active (YSG’s +24% from gear multiplies his entire active lane), and worthless on a commander whose skills carry no damage factor at all — a pure buffer like Scipio Africanus gains nothing from it.

Marginal value: what the next +10% is really worth

[P] Paper model

Because bonuses add within a category, each new point is diluted by every point you already own. The real gain of adding +x points to a category whose current total is T points:

Δreal = x / (100 + T)
Theorem 2 — strictly decreasing in T
[P] Paper model
Current attack stack TReal value of +10% attack
0%+10.0%
50%+6.7%
100%+5.0%
150%+4.0%
200%+3.3%
At a +150% stack, another slab of attack buys ~4% real damage — the quantitative core of “depth over breadth”.
[P] Paper model
The empty-category rule

A point is worth the most in the category where you hold the least. When your attack stack is tall, the next investment should almost always be an empty lane: skill damage, health, march speed, or the enemy-facing shred lane below.

Price your options in the Next Buff Advisor

Debuffs are multipliers on their whole stack

[P] Paper model

A defence-reduction debuff (“shred”) subtracts from the DEFENDER’s summed category, so it multiplies your effective damage no matter how tall your own attack stack already is:

D′ = D₀ × (1 + Σdⱼ − σ) ⇔ D′ ≈ D × (1 − shred)
σ is the debuff magnitude as a decimal
[P] Paper model

A defender at D = 160,000 hit with a 20% shred drops to 128,000 — and every hit from every march in the fight benefits. This is why a maxed +150% attacker who would gain only ~4% from another +10% attack instead reaps the full multiplicative swing from a shred deputy: Philip II and Liu Che scale disproportionately into the late game precisely because they operate on the opponent’s denominator.

Debuffs overwrite, they don’t stack

Two identical defence-shreds (or two slows) in one fight overwrite each other rather than adding — one of the eight off-pairing mistakes. Chain DIFFERENT debuffs, not duplicates.

The rage economy: casts per fight

[C] Community

Active skills fire the instant the rage bar reaches the skill’s cost (1,000 for most commanders); surplus above the cap is wasted. Over a fight, total skill output is governed by one ratio:

N_casts ≈ R_generated / C_rage
~91 rage/turn typical → a cast roughly every 11 turns at a 1,000 cost
[C] Community

Rage comes from normal attacks (~100 per hit), counterattacks (uncapped per turn — being engaged by several enemies charges you faster), damage taken, and restoration effects. Rejuvenate returns 60 rage per cast, 120 with it on both paired commanders; Sun Tzu Prime’s hit-based restoration is effectively a built-in Horn of Fury. Lowering the cost or raising generation multiplies total skill output — the single biggest damage lever in the game, worth more than any one +10% attack bonus.

Simulate your casts in the Rage & Casts calculator

Wounded vs. dead: the hospital rule

[M] Confirmed

Severely wounded troops are routed by combat CONTEXT, not a ratio. The death fraction φ (dead = φ × severely wounded) by situation:

Combat contextDeath fraction φ
Defending your own city0 — all go to hospital (capacity permitting)
Field / resource point / sanctum / Level 1 pass0 — all go to hospital
Defending an allied city or structure0.5 — half die instantly
Attacking cities or alliance buildings1 — all die
Lost Temple / Level 3 passes1 — all die
Troops beyond hospital capacity die regardless of context; slightly wounded always self-heal.
[M] Confirmed

The same battle that costs you nothing defending your own city (30,000 severely wounded, all hospitalised) loses all 30,000 permanently when you attack an enemy city. That asymmetry is the entire argument for hospital capacity before KvK, and for never open-field-attacking a city or holding a Level 3 pass without healing reserves in place.

Price the healing bill in the T5 Sustainment calculator

Related