Recent results
Match archive →Results appear on this tape as soon as a match finishes and its record is ingested.
Movement report
No 30-day movement clears its minimum sample yet
Movement is the difference between a window and the block before it, so it needs two comparable blocks and enough observations in each for the metric to consider the value measurable. No board in the corpus currently satisfies both conditions. Rankings below are unaffected — they describe the window itself, not a change in it.
What changed
Last 30 days against the 30 days before themA change needs two comparable blocks and enough observations in each. Coverage by game, below, shows where the corpus is currently too thin for that.
Movement coverage by game
- Counter-Strike 20 of 0 players clear the 75-duels minimum for Duel+ in 30 days — no movement is reportable.
- VALORANT0 of 0 players clear the 75-duels minimum for Duel+ in 30 days — no movement is reportable.
- Fortnite0 of 0 players clear the 12-matches minimum for PPA in 30 days — no movement is reportable.
- Rainbow Six Siege0 of 0 players clear the 20-opening duels minimum for Entry+ in 30 days — no movement is reportable.
- League of Legends0 of 0 players clear the 8-games minimum for GD@14+ in 30 days — no movement is reportable.
Values are in each metric’s own units and are not comparable between games.
Research by game
Each title analysed in its own unitsUtility Value is not computable on the current corpus. No CS2 player has 100 rounds of full telemetry in the season window.
Duel outcomes above expected
No VALORANT player clears the 75-duel minimum.
wins per 100 duels · ≥75 duels
Agent-level work, not yet computable
- Ability Value Added needs 100 rounds; no player reaches it in the corpus.
- Entry Value Above Expected needs 25 opening duels; the deepest VALORANT sample is short of it.
Placement Probability Added is not computable on the current corpus. No player has 12 matches in the window.
Telemetry composition
Operator, floor and gadget analysis reads from round telemetry. With 0 of 0 matches at that level, the only R6 model that resolves is the entry coefficient set, which runs on opening duels.
Objective probability added
No player clears the 8-game minimum for this metric.
Win-probability swing from dragons, heralds, barons and towers taken while the player was alive and nearby · ≥8 games
Round Win Probability Added
How much each player's actions moved their team's chance of winning rounds. A 1v3 clutch kill swings far more probability than a cleanup frag in a 5v2.
Round Win Probability Added needs 100 rounds in the window. Every player in the corpus is currently below it, so the board is empty rather than ranked on unusable samples.
Events
Events appear here once a tournament and its matches are ingested.
How these numbers are made
5 games0 matches0 full telemetry
- Source connectors
- 8 of 22 enabled · 14 awaiting approval
- Every connector is an official API, an authorized export, or a user upload, and each one carries a legal-review state. The ones awaiting approval are why the reference layer holds no results yet: rather than scrape around a pending credential, the platform runs the analysis on a Preview Lab corpus and labels it.
- Two layers, never blended
- Official records carry their source and confidence. Advanced values are model-derived from replays and carry a sample size, a percentile peer group and a model version. A record is one or the other, and every surface says which.
How data is labelled →Methodology →Metric glossary →Coverage by game →