Know every fight before you take it.

The headline number is a max-combo estimate

Every damage figure assumes the full burst rotation lands cleanly: every spell hits, every passive procs (sweet spots, on-hit marks, stacks), the ult is up, and crit auto-attacks roll their expected value. Real fights deliver less — when an enemy reads as 1,600 mitigated damage, treat it as the ceiling, not the average.

The math: combo damage = sum of each spell's base + AP / AD / bonus-AD scaling, plus modifier passives (% target-HP scalings like Vladimir E, ult damage amps like Aatrox R, empowered-cast chains like Aatrox Q1/Q2/Q3). Auto-attacks use expected AD = AD × (1 + crit_chance × (crit_damage − 1)). Mitigation applies per damage type (armor for physical, MR for magic, true damage ignores both). Time-to-kill adds sustained autos after the burst, accounting for each spell's cooldown.

What's real vs. what's estimated

Riot's Live Client only exposes the active player's stats in real time. For everyone else — including your allies — the overlay estimates AD, AP, HP, armor, and MR from base stats + visible items.

• Your stats: real (AD, AP, HP, pen, crit, attack speed, your keystone rune).
• Everyone else's stats: estimated from base stats + the items they have purchased.
• Enemy current HP: not exposed by the game — we use estimated max HP for lethality flags. The "LETHAL" badge for an enemy targeting you uses your real current HP.
• Enemy runes: not exposed — we pick the most-common keystone per champion (Conqueror for bruisers, Lethal Tempo for ADCs, Electrocute for assassins, etc.).
• Ability ranks on enemies: estimated from a Q-max-first heuristic. If they're W-max, the numbers will be off.

What we deliberately don't model

• Shields, heals, sustain. Brief bursts read accurately; drawn-out trades read pessimistically because Sona R, Soraka W, Aatrox passive heal, and lifesteal are all invisible to the math.
• Ability haste / cooldown reduction. Cooldowns shown are pre-haste, max-rank values. A 40% haste build re-rotates much faster than the TTK number suggests.
• Conditional empowerments tied to short-lived buffs (Renekton Fury, Riven Q3 sweet spot, Yasuo Q3 tornado bonus, Sett W grit conversion). We pick a representative case — usually the highest-damage one — and don't try to track the state machine.
• Active item passives (Hextech Rocketbelt, Stridebreaker, Goredrinker, Eclipse procs). Items contribute their flat stats but not their conditional triggers.
• Some champions are auto-parsed, not hand-curated — about 65 of 172 champions. Complex kits (Aphelios's 5 weapons, Udyr's stances, multi-hit ults, stack-counting passives) will be underestimated. Hand-curated champions cover all the high-pickrate cases.
• Anti-cheat / spectator delay. The Live Client returns data on a tick; we poll once per second. Fast burst sequences may flash through faster than the overlay refreshes.

When the numbers are most reliable

• Burst champions vs. squishies — Zed, Veigar, Annie, Lux. Single-rotation kills, no sustain, math is tight.
• Lethality checks against you — your HP is real, so the LETHAL flag for "Aatrox can one-shot you" is honest within the modelled-damage limits.
• Lane trade comparisons at fixed item milestones — both sides use the same estimation method, so the relative trade ratio is meaningful even if the absolutes drift.

When to take numbers with a grain of salt

• Drawn-out skirmishes involving healers, shielders, or any champion with significant sustain.
• Late-game with high CDR — re-rotation cadence makes burst-number comparisons less useful than DPS.
• Auto-parsed champions with complex kits (visible in-app as _autoGenerated: true in champion data). They under-count significantly.
• Right after a patch — base stats and item values may be a day or two behind the live game until we refresh.