About

What this is

hikeflystats takes the GPS tracks a hike & fly race publishes and works out what each athlete was doing minute by minute (walking, flying, riding a lift, or packing up), then puts those numbers and charts next to the official ranking.

It is a personal side project by Simon Steiner, who raced the first event it covers, and it is not affiliated with the organisers or with burnair. The official results are burnair's; everything derived from the tracks is unofficial.

Covered so far: Trail & Fly 2026 (Lenk / Adelboden, Switzerland, 73 athletes).

Watch the days in 3D

The charts here are flat. Ogoy replays the same IGC tracklogs in 3D, so a day can be watched rather than read: fly along with a pilot, split the screen between the 3D view and a 2D map for orientation, and put several tracks in one scene to compare a group's flights from any perspective. A scene is simply the set of tracks watched together.

Trail & Fly 2026, one scene per category and day:

• Trail Fly Lenk - PRO - Day 1
• Trail Fly Lenk - PRO - Day 2
• Trail Fly Lenk - FUN - Day 1
• Trail Fly Lenk - FUN - Day 2

Where the data comes from

• Tracks, official ranks, goal times, penalties and points: burnair, which scores these races and publishes one IGC track per athlete and day.
• Turnpoint cylinders and goal deadlines: the organiser's XCTrack task files.
• Ground elevation: the swissALTI3D terrain model, queried through the federal geo.admin.ch profile API. © swisstopo.
• Maps: OpenStreetMap.

How the activity split is calculated

An IGC track holds one position and one GPS altitude per second and nothing else: no cadence, no heart rate. Speed is therefore derived from a centred, multi-second distance baseline that stays time-aware across GPS gaps, and every second of the track is labelled as one of four activities:

• Flying: sustained high ground speed, no footwork.
• Hiking: movement on foot, including steep climbs where GPS speed drops to a crawl.
• Transition: near-stationary packing, unpacking and other boundary activity.
• Lift: sustained ascent too fast to be on foot. A legitimate part of the Fun category, where gondolas were allowed, and a warning in Pro.

Height above the terrain model is part of the decision, not just speed: anything sitting more than 150 m above the ground for half a minute is airborne whatever its speed. That keeps a wing thermalling into a strong headwind, with ground speed below walking pace for minutes on end, from being shredded into hiking and lift fragments. Gondolas ride within about 25 m of the ground and GPS altitude can be off by 120 m on steep slopes, so the bar sits well clear of both.

Short flights survive the smoothing: a flying stretch of at least 30 s that covers 200 m and peaks above 25 km/h is kept as its own segment and marked as a hop instead of melting into the surrounding hike. A landing that turns straight back into a launch shows up as a touch & go, with a confidence level, because on an IGC track a ground contact is only a sustained dip below flying speed and a running touchdown looks much the same. Each track is trimmed to the race itself using the goal cylinder from the task file.

Results, ranks and points

Turnpoint counts, goal times, penalties, ranks and points are shown exactly as burnair published them and are never recomputed here. Entering the final cylinder does not by itself make an athlete a finisher; the official ranking decides that.

burnair scores every gender group against its own winner, so the first finisher of Pro Men, Pro Women, Fun Men and Fun Women each score 1,000 points and points from different groups are not comparable. The combined men-and-women tables therefore rank by time instead: goal time plus penalty, summed over the days for athletes who finished every day.

The turnpoint crossing times on an athlete page are the exception: those are detected from the GPS track here, not taken from burnair. burnair scores turnpoints strictly in order, so a crossing after a missed turnpoint does not count officially.

How accurate is it

Everything outside the official results is derived from consumer GPS. Ascent computed from GPS altitude is noisier than a barometric reading, so treat the metre totals as indicative rather than exact. The classifier thresholds were calibrated against a Garmin FIT track, which does record speed and cadence, for both race days, and validated against it; other tracks can still produce boundary cases, and outliers are worth a second look. Ground elevation is sampled every 5 s along the track; where terrain is unavailable the classifier falls back to speed and altitude alone.