TFT combat mechanics: mana, damage and targeting

You never touch the fight. That makes combat maths feel academic — but every item you build and every unit you position is a bet placed on these rules, and the players who understand them are making informed bets while everyone else guesses.

The combat loop

When the timer hits zero, every unit does the same four things on repeat until one side is gone:

1. Find a target. The nearest reachable enemy, by default.
2. Walk to it if it's out of range.
3. Auto-attack at its attack speed, generating mana each time.
4. Cast the instant the mana bar fills, then start refilling.

Everything interesting follows from step 3 and 4 being coupled. A unit that is attacking is charging its ability. A unit that is walking, stunned, or has nothing in range is doing neither. Which means: uptime is the real stat, and most of what tanks and positioning do is buy uptime for the unit that matters.

Mana

Units generate mana mainly by auto-attacking, plus a smaller amount from taking damage. When the bar fills they cast, and many then enter a brief mana lock where they cannot gain more — which is why casting twice in quick succession is harder than the numbers suggest.

Two consequences worth internalising.

Attack speed is an ability stat. On a caster whose damage comes from its spell, attack speed is not "some extra auto-attack damage" — it is how fast the spell comes out. This is why attack-speed items appear on units that barely auto-attack for damage at all.

Starting mana is worth more than it looks. A unit that starts a fight at 0 mana has to earn its whole bar before contributing its ability once. A unit that starts half-full casts in the opening seconds, which in a fight lasting fifteen seconds can be the difference between casting twice and casting four times. Mana items are not a stat boost, they change how many times the unit does its job.

The spread is enormous

Mana costs vary wildly between units, and the gap between starting mana and cast cost is what actually decides how soon a unit casts. From the current set:

The top rows cast almost immediately; the bottom rows need most of a fight to get there. A mana item on the first group is nearly wasted — they were already casting. On the second group it can double their output. The per-unit numbers are on every champion page.

Targeting

The default is nearest reachable enemy, recalculated when the current target dies. That's it, and almost all positioning strategy is exploiting that one rule.

• Melee units walk toward the closest enemy hex. Your middle frontline hex is closest to the most of their board, so that's where their melee goes — put your best tank there deliberately.
• Ranged units attack the nearest thing in range, which is usually your frontline, which is the point of having one.
• Abilities frequently break the rule. Jumps to the furthest target, hooks on the closest, snipes on the lowest health — each is a specific exception and each has a specific positional answer, covered in the positioning chapter.

The practical takeaway: when you scout a board, you're not asking "are they strong". You're asking "what will their units walk toward, and is that what I want?"

Armour and magic resist

Two separate walls. Armour reduces physical damage only; magic resist reduces magic damage only; true damage ignores both. Damage taken is multiplied by 100 / (100 + resistance).

Read the third column, not the second. The percentage reduction shows diminishing returns — 100 to 200 armour only takes you from 50% to 67% — but effective health is perfectly linear. Every 100 armour adds exactly one extra bar of health against physical damage, forever. Resistance stacking does not fall off; it only looks like it does.

Health and resistances multiply, which is why they're better together than either alone. 2,000 health with 100 armour survives more physical damage than 3,000 with none. The failure case is specificity: 300 armour does nothing against a magic board, and that is the real reason to split defensive items rather than stack one type.

Crit and attack speed

Base crit chance is 25% and a crit deals 140% damage. So an average auto-attack is about 1.1× its listed damage before any items.

Crit chance above 100% is not wasted — it converts to crit damage at 50% of the excess. This is why stacking crit past the cap remains a real build on some carries rather than a rounding error.

The important exception: abilities can't crit by default. An item or effect has to grant it. That single rule explains why one specific item is build-defining on ability carries and close to dead on everything else, and why crit components on a pure caster are a trap unless you can get that effect.

What this means for your items

The whole chapter, compressed into decisions you'll actually make:

That's the engine. Next, the one part of this guide that expires: the Set 18 mechanic.