Enter the enemy draft and see which heroes beat it — with the mechanical reason, the sample size behind it, and the numbers for your own rank. Patch 7.41e, all 127 heroes.
A real result from the current data — not a mockup. These are the strongest counters to Lina right now, with the sample behind each.
Every one of these opens a page explaining the interaction that causes it.
A win rate tells you something is true, not what to do about it. Every matchup carries a short explanation of the mechanical cause — the ability interaction, damage type or timing window — written from the current patch's ability text rather than from a model's memory of an older one.
Standard error on a matchup is roughly the square root of one over the sample, so a big number on a thin sample is mostly noise. Figures are shrunk toward the expected result by sample size, and every row shows its match count and a confidence meter instead of hiding them.
A hero that counters the enemy draft on paper is useless if nobody at your rank can pilot it. Recommendations are weighted by how often heroes are actually picked at your bracket, so a hard counter has to earn its place over a hero you are more likely to play well.
Some heroes lose the laning stage badly and still win the match; others do the reverse. Lane outcomes are tracked separately from game outcomes, and the explanations say which phase an advantage belongs to — because those two call for opposite decisions.
The drafter takes up to five enemy picks, four allies and any bans, and ranks every remaining hero. Scoring runs entirely in your browser against preloaded data, so it re-ranks the instant you click rather than waiting on a request — which matters when the pick timer is at fifteen seconds.
Each of the 127 heroes has a page listing what beats it, what it beats, and which heroes it pairs with — every entry with an explanation and a sample size. Below that sits the popular item build for your bracket, ordered by the median minute each item is finished, so you can see the timings to aim for rather than just a list of items.
Builds are ranked by how often an item is bought, never by its win rate. Item win rates are misleading: the same item shows 82% bought at minute 30 and 55% at minute 48, because you reach expensive items sooner in games you are already winning. Purchase frequency has no such problem.
The same matchup differs between Herald and Immortal, and item timings shift by minutes. Everything is split across four brackets. Signing in with Steam detects your medal and selects the right one; you can override it at any time and the choice sticks.
Pick a position and roll — excluding anything already picked or banned. Positions come from what players actually play rather than from role tags, so asking for a position five gives you heroes genuinely played there.
It compares how often a hero beats another against what their general strength predicts, so a hero who is simply strong this patch does not read as a counter to everything. Thin samples are shrunk toward the expected result, and recommendations are weighted by whether players at your rank actually play the hero.
A percentage tells you that something is true, not what to do about it. Every matchup carries a short explanation of the mechanical reason — which ability interaction, damage type or timing window causes it — generated from the current patch's ability text rather than from memory.
Substantially. The same matchup can differ by several percentage points between Herald and Immortal, and item timings shift by minutes. Statistics are split across four rank brackets, and signing in with Steam selects yours automatically.
They often disagree. Some heroes lose the laning stage badly and still win the match, and some do the reverse. Lane outcomes are tracked separately from game outcomes, and explanations say which phase an advantage belongs to.
Free, no account required. Patch 7.41e, updated from legend–ancient match data.