REAL REPLAY PROTOTYPE — decoded from one uploaded archive. No player names, BattleTags, account IDs, chat, or replay bytes are included. This is not a coaching report.
NexusCoach — actual replay on the map
Braxis Holdout · one real match · sparse tracker observations, no interpolation
real decoded replayanonymous participantsraw loops onlyprototype
Step 1 · participant
Identity is not inferred. One anonymous candidate appears in every sampled match; you may confirm it as you, or inspect any participant without a personal claim.
No player selected. The map shows every recorded observation for all participants; nothing is attributed to you.
Participants (hero only)
Blue team first. Tracker ID is the replay's internal player number, not an account ID.
Match facts
What this view does and does not show
Observed: decoded tracker positions and PlayerDeath events. Not observed: continuous movement, intent, camera, chat, or names. Sampling: tracker positions are emitted for units that take damage. Gaps are missing observations, not proof of inactivity. No lines by default: straight segments between sparse samples would imply movement that was not recorded.
Map + transform
Background is the in-game Braxis Holdout minimap (Heroes Profile community asset), cropped to game units x 10.5–240.5, y 38.5–176.5. Prototype use only. u = (x − 10.5) / 230 · v = (176.5 − y) / 138
Positions use tracker item coordinates unscaled; they match structure and death coordinates in this build. The README's ×4 convention did not apply here.
Map — recorded observations
Braxis Holdout · in-game minimap · prototype
blue teamred teamdeath (PlayerDeath)last recorded position (held)near playhead
loop 0
Connected mode: dashed straight segments join consecutive recorded samples. These are not movement paths and may cross terrain. Off by default.
Playback cadence: the spatial stream is 104 hero-observation snapshots at 240-loop spacing (from 105 tracker position events), not continuous movement. Rings hold each hero's last recorded position and jump at the next snapshot; opacity fades with age. Playback speeds use 16 loops/s only as a viewing convention, not a validated client clock.
Observation timeline (raw loops)
Select a participant to see that participant's observations and deaths.
Marks are recorded observations, not a continuous track. Drag or click the lane to seek. No seconds conversion.
Limits
Sparse sampling. 695 hero observations across 10 participants over 26,336 loops, delivered in 104 hero-observation snapshots at 240-loop spacing. A participant appears only in some snapshots. Many loops have no observation for a given hero. A missing sample cannot establish position, rotation, or lack of action.
Playback is coarse. At 1x/2x the rings hold each hero's last recorded position and jump at the next snapshot; they do not interpolate. This is enough for rotation-level review, not continuous game replay. The 16 loops/s rate is a viewing convention, not a validated client clock.
Coordinate evidence. The map alignment was checked against decoded structure and death coordinates. It was not checked frame-by-frame against the game client, and the minimap background is not the competitive map projection.
Identity. The candidate is a statistical intersection of the sampled archive, not a logged-in account. Confirmation is a product state, not verification.