Guide · OSCAR · race replay
How do you replay an offshore race after the finish?
Use the public tracker archive. OSCAR rebuilds the whole fleet from tracker positions, then compares every boat against its polar, the current field and a computed optimal route — so the debrief argues from evidence, not memory. It is free and needs no signup: open the Rolex Fastnet 2025 replay and scrub through 424 boat entries over 1,287 miles.
View as table
| Fact | Value |
|---|---|
| Course | Cowes → Fastnet Rock → Cherbourg |
| Distance | 1,287 nm |
| Boat entries | 424 |
| Elapsed racing in dataset | 162.5 hours |
What a replay actually gives you
A tracker replay is more than an animation. Because every boat's position sits on a common clock, you can ask questions that are unanswerable from the deck: who gained on which leg, which side of the course paid, and whether a loss came from boatspeed or from routing. OSCAR layers three references over the raw tracks — the boat's polar (was she sailing to her potential?), the current field (was the gain real, or tidal?), and an optimal-route calculation (what did the best available course look like?).
A debrief sequence that works
Replay the whole fleet at high speed first, to spot where the race broke apart. Then isolate your boat against two or three direct rivals and step through leg by leg. Finish with the polar-performance and sector views to separate speed deficits from course choices. Each race carries a GPS quality grade (Good / Fair / Poor), so you know how much weight fine-grained comparisons can bear.
Methodology
OSCAR ingests positions from public race trackers, reconstructs each boat's track on a common timeline, and splits the course into legs at timing gates. Performance references come from boat polars, model current fields and route calculations. The archive currently holds 3,413 races, 86,285 boat entries and 359,274 analysed legs (snapshot of 16 July 2026).
Limitations
Tracker positions are sampled at intervals of minutes, not seconds, so short-timescale manoeuvres are smoothed out. Position quality varies by race and is graded rather than hidden — treat comparisons inside Fair or Poor races as indicative. Polars are class-level where no better source exists, and currents come from model fields, not measurements.