Documentation

Summoners War Mechanics

Practical SWCALC reference for damage, speed, and game-mode mechanics used by the optimizers.

Chapter 1: Damage Formula

Complete damage pipeline with terms 1 through 8.

Final Damage Equation

This is the full execution order used by the model.

Damage=(Multipliers⋅Crit⋅DMG%⋅DefenseFactor⋅Variance+Additional)⋅ReductionsDamage \allowbreak = \allowbreak \Bigl( \text{Multipliers} \cdot \allowbreak \text{Crit} \cdot \allowbreak \text{DMG\%} \cdot \allowbreak \text{DefenseFactor} \cdot \allowbreak \text{Variance} + \text{Additional} \Bigr) \allowbreak \cdot \allowbreak \text{Reductions}

Jump to term

Term 1

Base Hit Scaling (Multipliers)

Core skill scaling from ATK/DEF/HP/SPD coefficients per hit.

Multipliers=∑Stat⋅Multiplier\text{Multipliers} = \sum \text{Stat} \cdot \text{Multiplier}
Multipliers
The sum of each scaling stat multiplied by its coefficient.
Stat
The stat used by the hit (ATK, DEF, HP, SPD, etc.), including ATK/DEF/SPD buffs on the final stat: +50% / +70% / +30% of the overall stat.
Multiplier
Skill coefficient for that stat contribution.
Example

Multipliers are the values shown in swarfarm or in the skill selector in `/damage-optimizer`.

1 × [ ATK × 1.9 + MAX HP × 0.15 ]

Buffed stats: ATK/DEF/SPD scaling uses the overall stat, so ATK buff, DEF buff, and SPD buff multiply the total stat by 1.5, 1.7, and 1.3 respectively before the hit scales from it.

ATK/DEF Increasing Effect: the bonus is calculated as BaseBuff% × (1 + effect/100) and truncated. Example: +11% ATK up effect gives ATK buff 50 × 1.11 = 55.5% potency, which gets truncated to exactly 55%, not 56%. Similaryly DEF buff 70 × 1.11 = 77.7%, which gets truncated to 77%, not 78%.

SPD Increasing Effect: uses the same truncation rule. Example: +24% effect gives SPD buff 30 × 1.24 = 37.2%, which gets truncated to 37%.

Term 2

Critical Damage Term

Adds all crit-related sources into one multiplicative crit bucket.

Crit=\text{Crit} =(1+Skillups\bigl( 1 + \textbf{Skillups}+CDrune+ \text{CD}_{\text{rune}}+CDarti+ \text{CD}_{\text{arti}}+CDbonus+ \text{CD}_{\text{bonus}}−CDtaken)- \text{CD}_{\text{taken}} \bigr)
Skillups
"Damage +X%" skill-up bonuses; additive with all crit damage sources.
CDrune
Crit damage shown on the monster in-game.
CDarti
Artifact lines that include crit damage; additive in the same term.
CDbonus
External bonuses such as Euldong, tower, lead.
− CDtaken
Enemy artifact "Crit DMG Taken −N%" applied as subtraction.
Artifact crit damage line behavior
  • CD High/Low HP +N% always active, applies to all hits.
  • S1 / S2 / S3 Crit DMG +N% only applies when that specific skill is used.
  • First Attack Crit DMG +N% applies to the monster's first attack against each enemy on any given turn. This means Teshar S3's AoE damage will benefit from it (the primary target will not). Similarly, Astar S1's final hit on the other units will benefit from S1 CD.
  • Own 1-target CD +N% applies on the monster's turn for all hits. Even if the monster happens to trigger Co-op.
Term 3

Damage Bonus Term (DMG%)

Elemental damage dealt, co-op modifiers, and passives/skill effects.

DMG%=1+ArtifactOnElement+ArtifactCoOp+Other\text{DMG\%} = 1 + \text{Artifact}_{\text{OnElement}} + \text{Artifact}_{\text{CoOp}} + \text{Other}
ArtifactOnElement
Damage dealt on element lines.
ArtifactCoOp
Co-op attack artifact modifiers. Also applies for the monster triggering the co-op attack.
Other
Skill/passive-based damage increase effects.
Examples of other bonuses

Other includes passives (Lucas, Sonia, Leah) and skill effects (Brandia S3, Argen S3).

Term 4

Defense Factor

Enemy defense, ignore-defense ratio, and defense break combine into one reduction factor.

DefenseFactor=10001142+3.572⋅DEF\text{DefenseFactor} = \dfrac{1000}{1142 + 3.572 \cdot \text{DEF}}DEF=Defense⋅(1−Ignore%)⋅Dbreak\text{DEF} = \text{Defense} \cdot \left(1 - \text{Ignore\%}\right) \cdot \text{Dbreak}
Defense
Target defense stat before reductions.
Ignore%
Ignore-defense ratio from skills and effects.
Dbreak
0.3 with Defense Break active, 1 otherwise.
Deborah and Herteit modeling details

Deborah Deborah is modeled through the effective DEF break:

Dbreak = 1 - EffectiveDefBreak

Effective DEF break is 91% (normal). Boss Deborah reaches 80.5%, which the game rounds down to exactly 80%.

Herteit Herteit S3 Absorb is treated as stat transfer:

stolen = rate × knowledge × BaseStat

Rate is 0.25 (normal) or 0.10 (boss).

Defense Factor and Effective HP
Defense factor Effective HP
X-axis: enemy DEF from 0 to 4000.
0.200 0.400 0.600 0.800 0 1000 2000 3000 4000 30,859 61,718 92,577 123,436 Final multiplier Enemy DEF Effective HP
Defense: 1000
Effective DEF: 1000
Defense factor: 0.212
Effective HP: 47,139
Term 5

Ignore Defense

Term 5 in damage: ignore defense reduces effective DEF input but does not remove the DEF=0 floor.

DefenseFactor=10001142+3.572⋅DEF\text{DefenseFactor} = \dfrac{1000}{1142 + 3.572 \cdot \text{DEF}}DEF=Defense⋅(1−Ignore%)⋅Dbreak\text{DEF} = \text{Defense} \cdot \left(1 - \text{Ignore\%}\right) \cdot \text{Dbreak}
DefenseFactor∣DEF=0\text{DefenseFactor}\big|_{\text{DEF}=0}=10001142+3.572⋅0= \dfrac{1000}{1142 + 3.572 \cdot 0}=10001142≈0.876= \dfrac{1000}{1142} \approx 0.876
DefenseFactor|DEF=0
At DEF = 0, the factor is still 1000/1142 ≈ 0.876, not 1.
Ignore defense
Reduces the DEF input; does not remove the formula floor.
Term 6

Random Variance

Triangular variance: ±2.8% for attack/stat scaling and ±2.35% for direct Max HP scaling.

AttackVariance∈[ 0.972, 1.028 ],MaxHPVariance∈[ 0.9765, 1.0235 ]\text{AttackVariance} \in [\,0.972,\ 1.028\,],\quad \text{MaxHPVariance} \in [\,0.9765,\ 1.0235\,]
Variance
Independent triangular factors: [0.972, 1.028] for attack/stat scaling and [0.9765, 1.0235] for direct Max HP scaling.
Scope
Applies to damage before fixed/additional bucket is added.
Practical sampling model used

Both scaling buckets use independent triangular samples:


function sampleRandomVariation () {
  const triangular = spread => 1 + (Math.random() + Math.random() - 1) * spread
  return {
    attackScaling: triangular(0.028),
    maxHpScaling: triangular(0.0235)
  }
}
          
Julie precision dataset

Julie precision dataset

Julie S3 damage against the level 1 target used for the main variance calibration.

Optimizer expected
7,158
Observed mean
7,159.609
+0.022%
Observed range
6,964–7,353
−2.710% / +2.724%
Hits
5,533
Optimizer expected Observed mean Hits per 20-damage bin
0150300450550 6,9607,0407,1207,2007,2807,360 6960–6979: 29 hits6980–6999: 95 hits7000–7019: 141 hits7020–7039: 174 hits7040–7059: 207 hits7060–7079: 312 hits7080–7099: 365 hits7100–7119: 440 hits7120–7139: 478 hits7140–7159: 550 hits7160–7179: 547 hits7180–7199: 457 hits7200–7219: 397 hits7220–7239: 375 hits7240–7259: 272 hits7260–7279: 231 hits7280–7299: 182 hits7300–7319: 153 hits7320–7339: 86 hits7340–7359: 42 hits

Julie S3 test: 5,533 hits. Expected damage is 7,158; observed mean is 7,159.609.

Ramira-target dataset

Ramira-target dataset

Julie S3 damage against the Ramira target.

Stated setup prediction
8,827
DEF 152
Observed mean
8,810.969
−0.182%
Observed range
8,578–9,047
−2.644% / +2.679%
Hits
959
Optimizer expected Observed mean Hits per 20-damage bin
025507592 8,5758,6508,7258,8008,8758,9509,0259,050 8575–8599: 8 hits8600–8624: 13 hits8625–8649: 28 hits8650–8674: 40 hits8675–8699: 45 hits8700–8724: 60 hits8725–8749: 75 hits8750–8774: 85 hits8775–8799: 90 hits8800–8824: 86 hits8825–8849: 92 hits8850–8874: 69 hits8875–8899: 83 hits8900–8924: 54 hits8925–8949: 42 hits8950–8974: 40 hits8975–8999: 24 hits9000–9024: 18 hits9025–9049: 7 hits

Ramira-target test: 959 hits. Observed mean is 8,810.969. Current optimizer assumptions predict 8,827 at effective target DEF 152; DEF 153 predicts 8,809 and matches the data.

Luna, Shamann, and Giants Max HP dataset

Result: direct enemy Max HP damage is now modeled at ±2.35%. Damage Optimizer exports match the observed means within 0.36% while keeping attack/stat scaling at the known ±2.8% variance.

DEF break does not appear to change variance. Shamann with and without DEF break have close relative standard deviations, and a permutation Levene check does not show a meaningful split.

Luna and Shamann look same-ish with this sample size. Giants has more attack-scaling contribution, so Hall of Water remains the cleaner Max HP calibration.

Shamann with DEF break

2A Shamann S2 against the Hall of Water boss with DEF break.

Optimizer expected
1,071,348
Observed mean
1,070,264.592
-0.101%
Observed range
1,045,448–1,093,259
−2.418% / +2.045%
Hits
206
Max HP model ±2.35%
Optimizer expected Observed mean Hits per 20-damage bin
071421 1,045,0001,055,0001,065,0001,075,0001,085,0001,095,000 1045000–1047499: 3 hits1047500–1049999: 4 hits1050000–1052499: 7 hits1052500–1054999: 4 hits1055000–1057499: 11 hits1057500–1059999: 12 hits1060000–1062499: 7 hits1062500–1064999: 15 hits1065000–1067499: 16 hits1067500–1069999: 21 hits1070000–1072499: 18 hits1072500–1074999: 16 hits1075000–1077499: 14 hits1077500–1079999: 16 hits1080000–1082499: 12 hits1082500–1084999: 13 hits1085000–1087499: 5 hits1087500–1089999: 9 hits1090000–1092499: 0 hits1092500–1094999: 3 hits

Shamann with DEF break: 206 hits. Exported model predicts 1,071,348; observed mean is 1,070,264.592.

Shamann without DEF break

2A Shamann S2 against the Hall of Water boss without DEF break.

Optimizer expected
557,951
Observed mean
558,431.415
+0.086%
Observed range
546,283–568,691
−2.091% / +1.925%
Hits
65
Max HP model ±2.35%
Optimizer expected Observed mean Hits per 20-damage bin
0247 546,000552,000558,000564,000570,000 546000–547199: 1 hits547200–548399: 2 hits548400–549599: 2 hits549600–550799: 3 hits550800–551999: 1 hits552000–553199: 1 hits553200–554399: 5 hits554400–555599: 4 hits555600–556799: 3 hits556800–557999: 7 hits558000–559199: 6 hits559200–560399: 6 hits560400–561599: 4 hits561600–562799: 7 hits562800–563999: 3 hits564000–565199: 1 hits565200–566399: 4 hits566400–567599: 2 hits567600–568799: 3 hits568800–569999: 0 hits

Shamann without DEF break: 65 hits. Exported model predicts 557,951; observed mean is 558,431.415.

Luna with DEF break

Luna S3 against the Hall of Water boss with DEF break.

Optimizer expected
729,969
Observed mean
728,557.993
-0.193%
Observed range
712,543–745,095
−2.387% / +2.072%
Hits
271
Max HP model ±2.35%
Optimizer expected Observed mean Hits per 20-damage bin
0102031 712,000719,500727,000734,500742,000746,500 712000–713499: 2 hits713500–714999: 1 hits715000–716499: 3 hits716500–717999: 9 hits718000–719499: 10 hits719500–720999: 12 hits721000–722499: 15 hits722500–723999: 19 hits724000–725499: 16 hits725500–726999: 17 hits727000–728499: 31 hits728500–729999: 21 hits730000–731499: 23 hits731500–732999: 26 hits733000–734499: 17 hits734500–735999: 11 hits736000–737499: 12 hits737500–738999: 8 hits739000–740499: 9 hits740500–741999: 4 hits742000–743499: 2 hits743500–744999: 2 hits745000–746499: 1 hits

Luna with DEF break: 271 hits. Exported model predicts 729,969; observed mean is 728,557.993.

Luna Giants

Luna S3 against the Giants Abyss Hard boss without Deborah.

Optimizer expected
129,709
Observed mean
129,772.940
+0.049%
Observed range
126,977–132,545
−2.106% / +2.186%
Hits
50
Max HP model ±2.35%
Optimizer expected Observed mean Hits per 20-damage bin
0246 126,900128,100129,300130,500131,700132,900 126900–127199: 1 hits127200–127499: 0 hits127500–127799: 0 hits127800–128099: 4 hits128100–128399: 3 hits128400–128699: 2 hits128700–128999: 6 hits129000–129299: 6 hits129300–129599: 2 hits129600–129899: 4 hits129900–130199: 5 hits130200–130499: 3 hits130500–130799: 2 hits130800–131099: 3 hits131100–131399: 4 hits131400–131699: 1 hits131700–131999: 1 hits132000–132299: 1 hits132300–132599: 2 hits132600–132899: 0 hits

Luna Giants test: 50 hits. Exported model predicts 129,709; observed mean is 129,772.940.

Term 7

Additional Damage Bucket

Flat add-on damage from fixed hits and Additional Damage by +N% of stats.

Additional=∑Stat⋅Fixed⋅(1+DMG%)+Stat⋅AddStat\text{Additional} = \sum{ \text{Stat} \cdot \text{Fixed} \cdot \left( 1 + \text{DMG\%} \right) + \text{Stat} \cdot \text{Add}_{\text{Stat}}}
Fixed
Skill fixed damage bypasses crit/defense interactions, but still scales with DMG% bonuses.
ATK · Addatk
Additional damage by +N% of final ATK. Not affected by DMG% bonuses.
DEF · Adddef
Additional damage by +N% of final DEF. Not affected by DMG% bonuses.
HP · Addhp
Additional damage by +N% of final HP. Not affected by DMG% bonuses.
SPD · Addspd
Additional damage by +N% of final SPD. Not affected by DMG% bonuses.
Behavior and fixed-hit example

Additional damage by X% cannot crit, does not benefit from DMG dealt on Element, and bypasses DefenseFactor. Skill-based fixed damage also cannot crit and bypasses DefenseFactor, but unlike artifact additional damage it is still multiplied by (1 + DMG%).

Example, Coco S3: 5 hits dealing Fixed damage. Each fixed hit still scales with artifact DMG% bonuses, while artifact Additional Damage by +N% stays as a separate flat add after that.

5 × [ ATK × 1.5 Fixed × (1 + DMG%) ]
Term 8

Reductions

Late-stage reduction stack applied after the additional damage bucket.

Reductions=∑Additive⋅Multiplicative1⋅Multiplicative2...\text{Reductions} = \sum{\text{Additive}} \cdot \text{Multiplicative}_{1} \cdot \text{Multiplicative}_{2} ...
Additive
Artifact + Passive - Mirinae/Branding.
Multiplicative
Other damage decrease effects such as Camilla S3 or Reflect Damage. Apply each DamageDecrease effect one by one.
Additive
Artifact, Passive, Branding, Mirinae S3 all stack additively.
Passives in this bucket DO NOT stack, only the highest will apply.Passives that do not stack: Darion, Dias, Byungchul, Grego. Usually if a passive says "reduce damage by %" it falls in this bucket
Multiplicative

Reflect Damage buff Reflect Damage buff: Acts as a separate DamageDecrease effect and reduces incoming damage by 30%. Passives that reflect damage also decrease the damage taken and fall in this bucket. Reflected damage is calculated from the Raw damage before any damage decrease but after defense factor. Examples of effects: Camilla, Dark golem, Reflect damage buff These lists are not exhaustive, and each effect not listed here, should be tested on a case by case basis.

Chapter 2: Speed Calculation

Tick simulation, speed buckets, and ordering rules.

Speed Calculation

Tick-based turn simulation with combat speed recalculated from active effects.

ΔATBper tick=CombatSpeed⋅TickSize100\Delta \text{ATB}_{\text{per tick}} = \text{CombatSpeed} \cdot \dfrac{\text{TickSize}}{100}
CombatSpeed=⌈BaseSpd⋅(1+∑Spd%)+∑FlatSpd⌉⋅Slow⋅SpdBuff\text{CombatSpeed} = \left\lceil \text{BaseSpd} \cdot \left( 1 + \sum{\text{Spd\%}} \right) + \sum{\text{FlatSpd}} \right\rceil \cdot \text{Slow} \cdot \text{SpdBuff}

Jump to term

∑Spd%=Totem+Lead+Other Spd% Effects\sum{\text{Spd\%}} = \text{Totem} + \text{Lead} + \text{Other Spd\% Effects}∑FlatSpd=RuneSpd+OtherFlatBbonuses\sum{\text{FlatSpd}} = \text{RuneSpd} + \text{OtherFlatBbonuses}
SpdBuff%=⌊30⋅(1+SpeedUpEffect100)⌋\text{SpdBuff\%} = \left\lfloor 30 \cdot \left(1 + \dfrac{\text{SpeedUpEffect}}{100}\right) \right\rfloor
Slow={0.7,if Slow debuff is active1,otherwise\text{Slow} = \begin{cases} 0.7, & \text{if Slow debuff is active} \\ 1, & \text{otherwise} \end{cases}SpdBuff={1+SpdBuff%100,if Speed Buff is active1,otherwise\text{SpdBuff} = \begin{cases} 1 + \dfrac{\text{SpdBuff\%}}{100}, & \text{if Speed Buff is active} \\ 1, & \text{otherwise} \end{cases}
ΔATBper tick
Attack bar gain each tick from current combat speed and tick size.
CombatSpeed
Ceiled speed from base speed scaling, flat speed, slow, and speed buff.
SpdBuff%
floor(30 × (1 + SpeedUpEffect/100)).
Slow
0.7 if slow debuff is active, 1 otherwise.
SpdBuff
1 + SpdBuff%/100 if speed buff is active, 1 otherwise.
Implementation rules and edge cases

Swift correction: the game display uses ceil() on rune speed and ignores speed lead/tower/swift interaction in that view, so SWCALC corrects the rounding error when calculating true combat speed.


const frac = (BaseSpd * 0.25) % 1
const swiftCorrection = (isSwift && frac > 0) ? (1 - frac) : 0
const speed = speed - swiftCorrection
          

Speed-up effect truncation: potency is truncated down. Example: +22% speed-up effect gives floor(30 * 1.22) = floor(36.6) = 36%, not 36.6%.

Leo rule: if Leo exists, non-Leo monsters are capped to the lowest Leo combat speed.

Turn trigger: a unit can move when attack_bar >= 100. If multiple units are ready, highest attack_bar goes first.

Same speed order: when monsters have the same speed/priority, they move in the same order they were selected in setup.

Worked example

Unit has 105 base SPD, lead 33%, totem 15%, +150 speed on Swift, Speed Buff active, and Speed Up Effect of 13%.

Step 1: Base-with-lead+tower speed = 105 × (1 + 0.48) = 155.4.

Step 2: Swift correction: frac = (105 × 0.25) % 1 = 0.25, so correction = 1 - 0.25 = 0.75, and effective rune speed = 150 - 0.75 = 149.25.

Step 3: Add rune speed = 155.4 + 149.25 = 304.65.

Step 4: Round up before buff: ceil(304.65) = 305.

Step 5: Speed Buff potency = floor(30 × (1 + 0.13)) = floor(33.9) = 33%, so buff multiplier is 1.33.

Step 6: Apply buff: 305 × 1.33 = 405.65 combat speed.

Result: Final combat speed is 405.65.

Note: If the unit is not Swift, no correction is applied, so pre-round speed is 155.4 + 150 = 305.4 and ceil(305.4) = 306, exactly 1 higher.

Does Agrius Ignore Defense in Dragons?

Agrius has 104 base SPD, tower 15%, +145 speed on runes, Speed Buff active, Speed Up Effect of 24%, no lead.

Step 1: Base-with-tower speed = 104 × 1.15 = 119.6.

Step 2: Add rune speed = 119.6 + 145 = 264.6.

Step 3: Round up before buff: ceil(264.6) = 265.

Step 4: Speed Buff potency = floor(30 × (1 + 0.24)) = floor(37.2) = 37%, so buff multiplier is 1.37.

Step 5: Apply buff: 265 × 1.37 = 363.05 combat speed.

Step 6: Enemy speed is 163, so gap = 363.05 - 163 = 200.05.
Result: Agrius is more than +200 faster, so he gets full ignore defense.

Agrius Ignore-Defense Check

Computes combat speed and checks whether the speed gap meets the ignore-defense threshold.

Base SPD: 104 Enemy SPD: 163 Threshold: 200
Spd Buff
37%
× 1.37
Pre-Buff Speed
265
Raw: 264.60
Combat Speed
363.05
Pre-round ceil: 265
Speed Gap
Ignores
200.05
needs ≥ 200
▸ Show math (exact steps used for the numbers above)

Step 1: Base-with-tower speed = 104 × 1.15 = 119.6.

Step 2: Add rune speed = 119.6 + 145 = 264.6.

Step 3: Round up before buff: ceil(264.6) = 265.

Step 4: Speed Buff potency = floor(30 × (1 + 0.24)) = floor(37.2) = 37%, so buff multiplier is 1.37.

Step 5: Apply buff: 265 × 1.37 = 363.05 combat speed.

Step 6: Enemy speed is 163, so gap = 363.05 - 163 = 200.05.

Result: Agrius is more than +200 faster, so he gets full ignore defense.

Pre-round raw: 264.60
SpeedGap check: 200.05 ≥ 200

Speed % Increases

Percent speed effects stack additively in one bucket before base speed multiplication.

BaseSpd⋅(1+∑Spd%)\text{BaseSpd} \cdot \left( 1 + \sum{\text{Spd\%}} \right)
ΣSpd%
Totem + lead + other percent-speed effects stack additively.
Example
+24% lead and +15% other source behave as +39% total on base speed.

Flat Speed Bonuses

Flat speed is applied after base-speed percentage scaling.

SpeedBeforeRound=(BaseSpd⋅(1+∑Spd%))+∑FlatSpd\text{SpeedBeforeRound} = \left( \text{BaseSpd} \cdot \left( 1 + \sum{\text{Spd\%}} \right) \right) + \sum{\text{FlatSpd}}
ΣFlatSpd
Rune speed + other flat speed effects from skills/passives.
Order
Applied after base-speed percent calculation, before slow/buff multipliers and ceil rounding.

Temple of Chaos

Exact boss HP, ATK, and final score multiplier for levels 1 through 100.

Constant defense
1,000
Constant resistance
50%

Higher level means more HP and ATK.

Score multiplier rises every level. ×1.000 means 1 score per 1 damage; ×2.780 means 2.78 score per 1 damage.

Score gained = floor(damage from each hit × score multiplier)

Each hit is rounded down separately. Overkill damage still counts. Co-op attacks and damage that scales with enemy Max HP receive a 90% damage penalty. Score from secondary co-op attacks is calculated after this reduction, with no further penalty. Monsters that trigger the co-op attack receive no damage penalty.

Temple of Chaos boss stats for levels 1 through 10
Levels 1–10 HP + 5,363 ATK + 203 Score × + 0.015
1 41,250 1,450 ×0.230
2 46,620 1,653 ×0.245
3 51,975 1,856 ×0.260
4 57,345 2,059 ×0.275
5 62,700 2,262 ×0.290
6 68,070 2,465 ×0.305
7 73,425 2,668 ×0.320
8 78,795 2,871 ×0.335
9 84,150 3,074 ×0.350
10 89,520 3,277 ×0.365
Temple of Chaos boss stats for levels 11 through 15
Levels 11–15 HP + 6,188 ATK + 210 Score × + 0.020
11 103,125 3,553 ×0.385
12 109,320 3,763 ×0.405
13 115,500 3,973 ×0.425
14 121,695 4,183 ×0.445
15 127,875 4,394 ×0.465
Temple of Chaos boss stats for levels 16 through 20
Levels 16–20 HP + 6,188 ATK + 210 Score × + 0.023
16 134,070 4,604 ×0.488
17 140,250 4,814 ×0.511
18 146,445 5,024 ×0.534
19 152,625 5,235 ×0.557
20 158,820 5,445 ×0.580
Temple of Chaos boss stats for levels 21 through 40
Levels 21–40 HP + 7,013 ATK + 210 Score × + 0.025
21 181,500 5,655 ×0.605
22 188,520 5,865 ×0.630
23 195,525 6,076 ×0.655
24 202,545 6,286 ×0.680
25 209,550 6,496 ×0.705
26 216,570 6,706 ×0.730
27 223,575 6,917 ×0.755
28 230,595 7,127 ×0.780
29 237,600 7,337 ×0.805
30 244,620 7,547 ×0.830
31 251,625 7,758 ×0.855
32 258,645 7,968 ×0.880
33 265,650 8,178 ×0.905
34 272,670 8,388 ×0.930
35 279,675 8,599 ×0.955
36 286,695 8,809 ×0.980
37 293,700 9,019 ×1.005
38 300,720 9,229 ×1.030
39 307,725 9,440 ×1.055
40 314,745 9,650 ×1.080
Temple of Chaos boss stats for levels 41 through 100
Levels 41–100 HP + 7,013 ATK + 210 Score × + 0.020
41 321,750 9,860 ×1.100
42 328,770 10,070 ×1.120
43 335,775 10,281 ×1.140
44 342,795 10,491 ×1.160
45 349,800 10,701 ×1.180
46 356,820 10,911 ×1.200
47 363,825 11,122 ×1.220
48 370,845 11,332 ×1.240
49 377,850 11,542 ×1.260
50 384,870 11,752 ×1.280
51 391,875 11,963 ×1.300
52 398,895 12,173 ×1.320
53 405,900 12,383 ×1.340
54 412,920 12,593 ×1.360
55 419,925 12,804 ×1.380
56 426,945 13,014 ×1.400
57 433,950 13,224 ×1.420
58 440,970 13,434 ×1.440
59 447,975 13,645 ×1.460
60 454,995 13,855 ×1.480
61 462,000 14,065 ×1.500
62 469,020 14,275 ×1.520
63 476,025 14,486 ×1.540
64 483,045 14,696 ×1.560
65 490,050 14,906 ×1.580
66 497,070 15,116 ×1.600
67 504,075 15,327 ×1.620
68 511,095 15,537 ×1.640
69 518,100 15,747 ×1.660
70 525,120 15,957 ×1.680
71 532,125 16,168 ×1.700
72 539,145 16,378 ×1.720
73 546,150 16,588 ×1.740
74 553,170 16,798 ×1.760
75 560,175 17,009 ×1.780
76 567,195 17,219 ×1.800
77 574,200 17,429 ×1.820
78 581,220 17,639 ×1.840
79 588,225 17,850 ×1.860
80 595,245 18,060 ×1.880
81 602,250 18,270 ×1.900
82 609,270 18,480 ×1.920
83 616,275 18,691 ×1.940
84 623,295 18,901 ×1.960
85 630,300 19,111 ×1.980
86 637,320 19,321 ×2.000
87 644,325 19,532 ×2.020
88 651,345 19,742 ×2.040
89 658,350 19,952 ×2.060
90 665,370 20,162 ×2.080
91 672,375 20,373 ×2.100
92 679,395 20,583 ×2.120
93 686,400 20,793 ×2.140
94 693,420 21,003 ×2.160
95 700,425 21,214 ×2.180
96 707,445 21,424 ×2.200
97 714,450 21,634 ×2.220
98 721,470 21,844 ×2.240
99 728,475 22,055 ×2.260
100 735,495 22,265 ×2.280

Unit-Specific Info

Special-case behavior for selected units.

Deborah

Deborah Deborah is modeled through the effective DEF break:

Dbreak = 1 - EffectiveDefBreak

Effective DEF break is 91% (normal). Boss Deborah reaches 80.5%, which the game rounds down to exactly 80%.

Herteit

Herteit Herteit S3 Absorb is treated as stat transfer:

stolen = rate × knowledge × BaseStat

Rate is 0.25 (normal) or 0.10 (boss).

Chilling

Chilling Chilling: Chilling gets exactly +20 flat speed for each buff on himself. If 2 monsters have the same highest attack bar, the tie-break turn order only considers pre-combat speed, which excludes passives. That means Chilling will always go last in a speed tie-break scenario.

Tested with Rigna and Tablo on various speeds, speed lead, and swift/non-swift configurations.

Sonia

Sonia Sonia: Sonia gets full bonus at +50 speed faster than the enemy.

Leah

Leah Leah: Leah gets full bonus at +150 speed faster than the enemy.

Agrius

Agrius Agrius: Agrius gets full ignore defense at +200 speed faster than the enemy.

Birgitta / Rigna / Tirsa

Full ignore defense at +126 speed faster than the enemy.

Rigna: Rigna will auto-crit in case she is faster than the enemy by any amount (tested at 0.02 speed faster). If 2 monsters start at 0 speed in the same tick and Rigna outspeeds but does not auto-crit, the 2 monsters have exactly the same speed.

Brita

Arena/RTA (towers only): +730 ATK, +884 DEF, +129 SPD. Siege/Guild War (towers + guild flags): ~+583 ATK, ~+737 DEF, +129 SPD.

Cordelia

Gets 3.5x combat speed added to her attack.

Mirinae

Skill 3 works as 25% negative damage resistance (the recent buff said "25% > 30%", but the actual tested damage numbers did not change). It stacks additively with -DMG% taken artifacts, but neither effect applies to bomb or fixed damage.

Lucifer

Each attack adds a damage bonus for future attacks. Whenever an attack deals more than 5000 damage increases the damage of further attacks by damage / 5000 (rounded down) and increases the attack bar of all allies by the same amount. The damage bonus is cumulative. A multi-hit attack uses one unchanged bonus for every hit; its bonus is added only after the whole attack ends.

added bonus % = attack damage dealt ÷ 5,000

  • 5,000 damage: future attacks get +1% damage, and all allies gain 1% Attack Bar.
  • 12,500 damage: future attacks get +2% damage, and all allies gain 2% Attack Bar.
  • Multi-hit: For example Julie S3 - all 6 hits use the same damage bonus, so hit 6 gets no more damage than the first hit.
Pholus Needs Testing

Needs testing.

Chow Needs Testing

Needs testing.

Deva Needs Testing

Needs testing.

Notes

Practical model scope and validation status.

Predicted damage is usually within ~0.1%, and only rarely up to ~0.3% difference.
These are model formulas for prediction and tuning, not reverse-engineered game source code.