Death Location Heatmaps Across Ranks and Maps
Where Do VALORANT Players Die Most, and Does It Change by Rank?
This report maps 40.33 million death positions from 260,846 European competitive VALORANT matches, covering Iron through Radiant between February and July 2026. Death locations are not evenly distributed, and where a player dies turns out to be a strong signal of their rank.
- A positional model given only where the ten players in a match died predicts that lobby's rank with a Spearman correlation of 0.78, tested on a held-out set of 52,519 matches.
- That model places a lobby within one rank division of its true rank 74.3% of the time, with a mean absolute error of 2.98 divisions, using no combat statistics at all.
- Ascent, Split, Pearl, Haven and Fracture show the clearest relationship between death position and rank. Abyss, Corrode, Summit and Sunset score lower purely because they are newer maps with only 1,000 to 2,000 matches each in the evaluation set.
- Heatmaps are published for all 12 maps across 8 rank brackets, giving 96 separate per-map, per-rank death distributions.
Positions come from the victimLocation field of official VALORANT match detail records, transformed into minimap space. Rank divisions are grouped in threes (Iron 1-3, Bronze 1-3, and so on) with Immortal and Radiant combined into a single top bracket.
Every death in a competitive match leaves a coordinate behind. Aggregated across enough matches, those coordinates trace out where players actually stand, peek, and die on each map. That pattern shifts noticeably as rank goes up. This report extracts death positions from 260,846 EU competitive matches (Iron through Radiant, February to July 2026) and renders them as per-map, per-rank heatmaps: 40.33 million individual death events in total.
Positions come from the match details's victimLocation field, transformed from ingame coordinates into minimap space. Divisions are grouped in threes to keep sample sizes usable: Iron 1 - 3, Bronze 1 - 3, and so on, with Immortal and Radiant folded into a single top bracket.
Can the Death Positions Predict a Player's Rank?
To check whether death position alone carries real rank information, a held-out set of 52,519 matches was scored against a simple positional model with no access to combat stats, only where each of the ten players died. The result: a correlation of 0.78 between predicted and true lobby rank. Positioning alone is a strong teller for a player's true skill.
Explore the Heatmaps
Per-map signal strength
Signal strength varies a lot by map. Ascent, Split, Pearl, Haven, and Fracture all show a strong, clean relationship between where players die and their rank. Abyss, Corrode, Summit, and Sunset score much lower, but that's due to a lack of sample data, not a real gameplay difference. These are the newest maps in rotation, each with only 1k - 2k matches in the evaluation.
All maps and rank divisions
Abyss
Ascent
Bind
Breeze
Corrode
Fracture
Haven
Lotus
Pearl
Split
Summit
Sunset
