Hero Grades

Most hero tier lists are somebody's opinion. Ours is a measurement.

Every hero on this site is graded from the game's own code, and the two combat systems are graded two different ways.

For Arena and Patrol, we run the game's actual battle engine — the same code the game itself uses to resolve a fight — simulated over millions of battles. Not a model of the engine. The engine, pulled from the actual game files, resolving actual battles.

For world-map deployment, the game resolves the fight on its own servers, where nobody outside the studio can watch it. So we grade from the game's own configuration tables instead: the real buff values the game feeds its own math, combined the same way the game's own code combines them. Not a community poll. Not a feel.

What zero is. Zero is a hero that does nothing. Not the worst hero in the game — a placeholder with its class's base stats, a basic attack, and no skills at all. We measured what that placeholder achieves and called it zero. No real hero sits there, which is the point: the bottom of the scale is a fixed reference, not whoever happens to be worst this patch.

What 100 is. A hundred is a target set a little beyond the best performance we have ever measured. It is deliberately just out of reach, so the top of the scale stays open. That is why the strongest heroes land in the 80s and 90s and nobody is at 100.

We do not spread the grades out. A grading scale that always produces a tidy spread is a ranking wearing a costume. Ours does not. If the roster has two standouts and then a long drop, the numbers show the drop, and whole letter bands come out empty. We would rather print an awkward table than a flattering one.

Nobody's opinion is in them. No voting, no vibes, no “this hero feels strong in my alliance.” A hero's grade is whatever the numbers say it is. When the numbers disagree with conventional wisdom, we publish the numbers.

Every hero is tested identically. Heroes are measured at one fixed, identical build, so nothing in a grade comes from one hero having been invested in more heavily than another. You are seeing the hero, not somebody's account.

The Arena/Patrol scale does not move. That grade is an absolute number, not a ranking against the current roster. Zero on the scale is a hero that contributes nothing at all. No real hero sits there, and that is the point: it is a fixed reference, not the worst hero we happened to find. When a new hero launches, it does not quietly shuffle everybody else up or down the list. The hero you built last season still scores exactly what it scored last season, unless the developers actually changed it.

That is rarer than it sounds. Most tier lists are relative by construction, which is why they churn every patch even when nothing about your heroes changed.

The Deploy scale is different, and we will not pretend otherwise. World-map battles are resolved on the game's servers. We cannot watch them, so there is no fixed set of battles to measure a hero against, and no honest way to make that grade absolute. The Deploy grade ranks a hero against the rest of the current roster instead.

That has a consequence worth stating plainly. When the developers release a hero who is stronger than everybody else, or rebalance the hero who was, every Deploy grade is restated against the new top. The heroes who were the best at deployment stop being the best, and their grades say so. That is not churn. It is the meta actually moving, and a Deploy grade that ignored it would be worse than useless.

So a Deploy grade carries a date and a game version. An Arena/Patrol grade does not need one.

The game runs two completely different combat systems, and a hero can be excellent at one and unremarkable at the other. Collapsing that into a single letter throws away the part you actually needed. So every hero carries:

• Arena / Patrol Grade — hero-versus-hero combat
• Deploy Grade — the world map, where the hero leads your troops

Under the Deploy grade sit two role scores, because leading troops is two different jobs:

• Troop Damage — how much harder the hero makes your troops hit
• Troop Protection — how much longer the hero keeps them alive

A hero can sit near the top of one and near the bottom of the other, and several do. The Deploy grade weighs the two equally. We do not assume that hitting harder matters more than lasting longer, or the reverse. That is a judgement about how you play, and nothing in the game files states it.

Two heroes, Brooke and Andre, carry a talent that is always on. Brooke's Judgement Day and Andre's Natural Leader each strengthen your deployed squad, and each one works whether that hero is deployed or not. Owning the hero is enough. Both talents apply to world-map deployments only — neither does anything in Arena or Patrol.

Neither talent counts toward that hero's Deploy grade. That is on purpose.

A Deploy grade answers one question: you have a slot to fill, so who goes in it? An always-on talent cannot help you answer that. If you own Brooke, every deployment you send already has her talent on it — the ones she is in and the ones she is not. It does not stack, and it does not care which hero you picked. Folding it into her own grade would make her look like a better pick for the slot than she is, and would credit her for something you were getting anyway.

So we keep it separate: a talent is a reason to own a hero, not a reason to field them. Both talents are strong, and counting them would have moved both heroes by a full letter — which is exactly why the grade has to mean one thing and not two.

Game updates can change hero stats, skills and the battle engine itself, often without saying so in the patch notes. Every grade is re-measured after a version bump and stamped with the game version it was measured at. If a grade on this site is stale, the version stamp will tell you.

We would rather be trusted than impressive.

Some hero abilities resolve on the game's servers, where no amount of analysis of the game client can see them. Where that is the case we say so on the hero, rather than guessing and presenting the guess as a measurement. A handful of heroes carry real uncertainty for that reason, and those heroes are flagged.

A few hero skills only fire in one kind of deployment — a rally, or a garrison, or only on defense. Those are real, and they are listed on the hero, but they are held out of the Deploy grade rather than averaged into it. A skill that works one time in five should not quietly raise a hero's score the other four.

We also do not publish a “best position” for a hero. The best position depends entirely on what you are up against, and any site that gives you one number for it is making it up.

Full grades, role scores and the per-hero breakdowns are available to subscribers only.

The method behind them — how the battles are chosen, how contribution is measured, and how it all maps onto the final score — is proprietary and stays that way. It is built directly on the game's own combat code, and that is the reason these grades are worth reading.

As is the case with some of our other site content, there are real costs associated with producing these grades (time, effort, and money), and subscriptions help cover them. Our battle simulation server uses 192 CPU cores and 384 GB of RAM. That is not a cheap server to run but it is what allows us to run millions of battles and produce the grades you see here with as accurate information as possible.