How Head-to-Head Data Works
What counts as a meeting
A meeting is a completed series between two teams at a tournament I track. Single games, show matches and scrims don't count. If a series is on the site with a result, it's in the head to head.
Teams that played an event under a different name (it happens at Valve events and with sponsor changes) are folded into their main team. They share one history and one rating, so a rename doesn't reset a rivalry.
Head to head on the match cards
Every card shows the series record between the two teams and coloured dots for the last five meetings, newest on the left. It's context, not gospel: I only weigh head to head when the rosters and meta are comparable. A win two years ago with different players doesn't mean much.
Rivalry pages
Once two teams have met 3 times or more, the pair gets its own page — for example PARIVISION vs Team Spirit. Each pair has one address, with the names in alphabetical order. On the page:
- the overall record and every meeting with its tournament, date and score;
- both teams' ELO over time on one chart;
- the model's chances for their next meeting.
Fewer than three meetings is too thin to read anything into, so those pairs don't get a page. The record still shows on the match cards and in the team comparison.
Team comparison
The compare tool works for any two Division 1 or Division 2 teams — even ones that have never met. Rating and division rank, 30-day change, last five series, the model's BO1/BO3/BO5 chances (with the division adjustment), every meeting, and both rosters. Your choice is saved in the URL, so a comparison can be shared.
Player vs player
These compare players in the same position — carry vs carry, mid vs mid — who played against each other, on opposite sides, in tracked pro games. A pair needs at least 3 such games.
The per-game numbers come from OpenDota: K/D/A, last hits and denies, GPM and XPM, hero damage, net worth and the heroes played. The page shows the averages across all meetings, each player's heroes, and every game.