Professional matches broken down into data.
Discover advanced stats from the professional Dota 2 scene, compare players and heroes and follow metrics that ordinary results sites don't show. All models run on live data and show how head-to-head matchups may develop.
Every match leaves data behind. We turn it into insight.
Compare the start of a match before the first creep.
Pick a player or hero and see historical last hits by minute 5. For every lane you see the core, the support, denies and any roam.
How we read it
What we compare. Farming efficiency of players under the same game conditions: last hits and denies by minute 5, the lane and roaming supports.
Coverage. Since the last major patch, continuously updated with new matches.
Reliability. The model requires complete lane data, unprocessed matches are excluded from the calculation. Lanes with a roamer (P Roam / E Roam) are excluded from the head-to-head calculation.
Who gets more kills?
Compare average kill stats of players and heroes in specific matchups or across the whole professional scene.
How we read it
What we compare. Who drives the kills in a matchup: kills by the player or hero, match duration and result.
Coverage. Since the last major patch, continuously updated with new matches.
Reliability. Kill models are sensitive to small samples, large swings can distort the average.
How many towers does each team take by 10:59?
Follow the towers that fall by 10:59. The table shows how many a team takes/loses as well as the latest head-to-head matches, which help you better understand how the early game develops.
How we read it
What we compare. The first tower, the tower count and map control in the first ten minutes of the game: towers taken and lost by minute 11, form over the last 10 games and head-to-head matches.
Coverage. Since the last major patch, continuously updated with new matches. We deliberately keep the window from the patch so the model reflects the current game, not the game before the changes.
Reliability. The model is a statistical estimate based on averages, not a prediction of a specific match.
From hero to player, from player to numbers.
The models above answer a specific question. The Explorer is for exploring - you start with a hero, move on to the players who played them and end up at their match stats.