Why systems beat cards
A single card has one number. A run has six multipliers stacked on the same handful of cards. Whoever stacks 3–5 of them by year 15 wins; single-card strength only matters as "can this card carry a multiplier".
real power ≈ base × ①levels × ②units × ③adjacent buildings × ④policies × ⑤perk triggers × ⑥enemy structureThe six zones
Level growth
cardFields MHPBoost / MAttackStrengthBoost / MAttackDelayBoost applied per level; unit count also x level.
Check: Is the carry level 3 by year 5? Is there a way to buy levels (Shrine/Concabulator)?
Unit count
DPS = per-unit DPS x unit count. Farm adds +1 adjacent unit per year; Populate adds more per placement (value unmeasured).
Check: Is unit count rising every year, or frozen at the card face value?
Adjacent support buildings
Farm unitsPerEffect = 1; Blacksmith strengthPercentage = 1.02; Beacon speedBoost = 1.02; Quarry damageBoost = 1.02; Trebuchet +1 projectile per building.
Check: Count how many support buildings actually touch your carry. Zero means this zone is 0% of your power.
Policies
Repeatable / RepeatableTimes decide stacking (Empower, Resilience, Exploration up to 20). Multiplicative beats additive.
Check: How many of your policies multiply, rather than add a flat percentage?
Perk triggers
The perk board is a 5x3 grid; rainbow perks retrigger adjacent/vertical/left perks. Max level is 3, and rainbow perks can boost each other.
Check: Is any rainbow perk placed next to the perks you actually want doubled?
Enemy structure
Enemy waves are fixed formations that scale with year. The same unit is bad vs Greed (back-line dive) and core vs Blood.
Check: Did you change positioning for this king, or are you replaying one comp against everything?
Three chains you can recompute
Chain A: imp 48.75 → 707.9 (reproducing the current site)
- L1: 5 attack / 1.4s = 3.571 per unit; crit expectation 1 + 0.05 x (2 − 1) = 1.05; 3.571 x 13 units x 1.05 = 48.75 DPS.
- L3: units 13 x 3 = 39; attack 5 x 2^2 = 20; interval 1.4 / 1.1^2 = 1.1570s; 20 / 1.1570 = 17.286 per unit.
- Total: 17.286 x 39 x 1.05 = 707.8 (the page shows 707.9 after rounding).
Takeaway: The arithmetic is right; the problem is that the input stops at level 3. Everything above is a multiplier the site does not model.
Chain B: gold → Dispenser attack speed (greed engine)
- Vault currencyPerLevel = 3; level-99 vault produces 3 x 99 = 297 gold/year (base income is 9).
- Six vaults ≈ 1,782 gold/year; Dispenser boostPerCoin = 1.005.
- 1.005^1782 = e^(1782 x 0.0049875) ≈ e^8.89 ≈ 7,240x attack speed.
Takeaway: This is an estimate: the game clamps attack speed (one community post reports a 99,999/s cap), so the real number is lower. The point stands: gold stock is a multiplier, and a per-gold ROI page counts it backwards.
Chain C: Catharsis returns differ 11x
- Catharsis sets every troop to the damage of your highest-damage troop; value = unit count x (top damage − own per-unit damage).
- 39 imps (17.3 per unit) aligned to a 100-DPS wizard: 39 x (100 − 17.3) ≈ 3,225 extra DPS.
- 3 paladins (2.1 per unit) aligned to the same wizard: 3 x 97.9 ≈ 294 extra DPS.
Takeaway: Same policy, 11x difference — decided only by unit count. This is why a single-card tier list cannot rank policies or builds.
What a tier list cannot answer
- It does not answer "how do I win this run": winning is 3–5 zones stacked together, while the ranking models one and a half.
- It does not answer "when to take it": Ballista/Defender are different cards at L1 and L3, and the ranking only has an L3 slice.
- It does not answer "strong against whom": the same unit can be good vs Blood and bad vs Greed.
- It does not answer "is there another route": mechanic units (reflect, first shot, limitless levels) have no sustained DPS to compare.