Kat’s verdictJudge every debuff against one opponent and one matching stat. The commonly tested ceiling is half of that enemy buff. Spend again only when the next point changes a fight you expect to take.

Use the opponent-relative ceiling

Specialist guides and player testing consistently report that a percentage debuff can remove no more than half of the opponent’s corresponding positive buff. Treat attack, defense and HP for every troop type as separate ledgers.

Debuff ceiling for one stat = enemy buff ÷ 2
Useful debuff = smaller of your available debuff or that ceiling
Unused potential against that opponent = available debuff − useful debuff

Worked example: an attacker has 2,400% ranged attack. The ceiling is 1,200%. If your build can supply −1,500%, only −1,200% can apply and 300 percentage points are unused against that attacker. Against a 3,200% ranged attack, the ceiling is 1,600%, so the full −1,500% can apply and 1,700% remains. Same defender. Different invoice.

A cap is a relationship, not a medal.

Read the report before the shopping list

  1. Open Troop Buff in the battle report. Specialist report guidance says the positive buff shown is before the opponent’s debuff is subtracted.
  2. Match the exact row. Ranged attack debuff does not pay the mounted HP bill, however confidently it is displayed.
  3. A report may show only the amount that could apply against that opponent. It does not prove your full theoretical total. A stronger controlled target may reveal more.
  4. Confirm that every source was active. Check whether each subordinate city was selected or linked and could participate in that battle. Buildings where subordinate cities cannot join, assistant restrictions and rally conditions can change the total.

Give the next upgrade a job

Run a small, boring test

Keep the march, target and conditions as constant as possible. Save the full report, including Troop Buff. Change one debuff source, repeat, and record the enemy buff, applied debuff and troop result. Three comparable reports are more useful than one dramatic screenshot. Test in a setting where the troop and resource cost is acceptable.

This calculation isolates one percentage stat. It does not predict the battle by itself. Troop tiers, layers, flat refines, march size, reinforcements, skills and which subordinate cities participate still matter.

Sources: Third-party Evony LiveOps example of general debuff attributes; Theria specialist buff and debuff guide; Server 806 battle-report guide and pre-debuff display note; Player discussion of report-capped debuff display.

Evidence reviewed 2026-09-30. The third-party Evony LiveOps guide lists general skills that reduce specific enemy attack, defense and HP attributes; it does not establish the 50% calculation rule as an official rule. The half-buff ceiling and report behaviour come from independent specialist guidance and repeated player testing. Game updates, activation conditions and the current in-game report remain the final check.

Build for opponents, not for the prettiest negative number. The forge is very brave with somebody else’s materials.