Reference run

A reference run is the canonical pass for a track session or a public event. It defines the track itself — where it starts, where it ends, and which intermediate splits matter. Every other run in the same session or event is aligned against the reference and compared to it.

Why one run is special

Racecraft is built around relative analysis. To say "this run was 0.4 seconds faster than that one at the third corner," there has to be a shared definition of "the third corner." The reference run is that definition.

• the reference's start split defines when the timer starts
• the reference's finish split defines when the timer stops
• the reference's intermediate splits define the section boundaries for the leaderboard and the per-split deltas

If the reference's splits are sloppy, every comparison is sloppy.

Reference run vs target run

A target run is any run that is not the reference. Target runs do not themselves define the track — instead, their splits are predicted by aligning their video against the reference run's video.

Both the reference and the target runs are otherwise the same kind of object. A run that is currently the reference can stop being one (the field gets repointed), and a run that's a target can be promoted to reference. The underlying run document is unchanged.

Choosing the reference

The strongest reference is:

• complete — the full track from intended start to intended finish
• clean — no crashes, no walking, no major mistakes
• clear — stable mount, no fogged lens, no obstruction
• representative — the line you actually want target runs measured against

The reference does not need to be the fastest run. The reference defines geometry; the leaderboard captures speed.

For organizer-specific guidance, see The reference run for a public event. For solo and team use, see Choosing a reference run.

Changing the reference

Re-pointing the reference re-predicts splits on every other run in the session and recomputes the leaderboard against the new template. Confirmed splits on target runs are preserved; unconfirmed predictions are recomputed.

This means changing the reference is reversible-ish: you can re-point back, the runs are still there, the math just gets re-applied each time.

Visibility

In a team session, designating a run as the reference makes it visible to every team member, even if the underlying run was private up to that point. This is what keeps the team's reference canonical — one shared template for everyone.

In a public event, the reference run is shown on the public page so spectators can see the track even before any other rider has uploaded.