Downwind turns your GPS trace into rides, links, and dead water, and tracks whether you're riding more of the ocean. From the watch you already wear. On the runs you already repeat.
You come off the water and rate the run by feel: it felt connected, the ocean was good, you think you linked more than last time. Or you open the speed trace and count the spikes by hand. Downwind replaces the guess with a number and a picture.
Both columns describe the same run. Only one of them tells you what to work on next time.
No pace-per-mile concepts imported from running. Downwind segments every 1 Hz GPS recording deterministically into the units paddlers already think in.
Every bump you catch, found in the trace: when it started, how long you carried it, how fast it peaked. Counted, not remembered.
Rides connected into chains. Linking is the skill of downwind, and Downwind measures it directly: chain length, chain frequency, and how both are trending.
The stalls between rides, timed and located. Where you lost the ocean matters as much as where you rode it. That's where the next percent comes from.
The core ambiguity of downwind progression: was that you improving, or was the ocean generous today?
Downwind builds a baseline from the runs you repeat. Every session on the same route is measured against your own history there, so the trend that emerges is skill, not swell.
Most fitness apps will tell you a story whether or not the data supports one. Downwind's coaching is mechanically forbidden from doing that.
On GPS alone it describes what happened and how it's trending. Claims about why only unlock when wrist-sensor evidence exists to back them. And when the data is ambiguous, uncertainty resolves to silence, not a guess.
The result, for a sport riddled with luck and feel: an improvement it reports is real. Each session ends with one observable thing to try next run, framed as an experiment, not a diagnosis.
You spent 55.7% of this run on bumps, up 4.2 points on your Maliko baseline. Your two longest stalls came in the last third; earlier sessions stalled evenly throughout.
You were fading late in the run and your catch got lazy.
requires wrist-sensor evidence · claim not madeOn the last third, try committing to the first bump after each stall instead of waiting for a bigger one. Watch whether your late-run dead-water total drops below 40 s.
Garmin FIT files are a first-class input, and Apple Watch recordings work the same way. Any 1 Hz GPS trace is enough for the full core analysis.
You record every session. You re-run the same route. You're past the survival stage, plateaued, and there's no downwind coach within a thousand miles. This is for you.
A frigatebird reads the energy on the ocean's surface. Downwind reads it in your trace.
After every run you'll know whether you rode more of the ocean than last time, and what to try next. Downwind is in TestFlight beta now.