Master thesis research

Skiing with data: pace and decisive course segments

A data-driven study of cross-country skiing performance using high-resolution GPX files and competition-specific insights.

Master thesis in Statistics (2025-2027) · Birk Møller Gundersen

Research vision

Data analytics has transformed many sports, but cross-country skiing still has major untapped potential in the detailed race data already being collected.

Main goal: Use data to understand what decides races.

Traditional analysis often relies on lap times and checkpoints. To understand what really creates differences, we need high-resolution analysis of speed, terrain, and pacing across the full course.

Core thesis topics

Three connected questions about performance in ski racing

Pacing Strategy

Analyze what type of pacing strategies correlate most strongly with top placements. Identify at what points in a race top skiers ski at higher pace compared to competitors.

Critical Segments

Discover where ski races are actually decided. Understand which terrain segments—steep climbs, transitions, corners, or downhills—influence final results the most.

Micro-Terrain Impact

Examine how micro-terrain features, specifically transitions between uphill and flat/downhill sections, impact fatigue development and speed decay throughout a race.

Data used

High-resolution race tracking from Norwegian national competitions

Source
Garmin GPX Files

Race data from athletes competing in Norwegian national cup competitions

Resolution
1 Hz (1 second)

Location data captured every second for high-resolution accuracy

Metrics Captured
Speed, Distance, Elevation

Calculate precise pacing patterns and terrain analysis across the course

Coverage
Multiple Courses

Data from different venues throughout the season, various distances and techniques

Important note on accuracy: GPS and altitude data from watches introduce a natural error margin. The analysis accounts for this measurement variation.

Processing pipeline

From raw GPX to usable insight in four stages

1

Upload GPX Data

When you upload a race GPX file, our system normalizes the track to fixed 10-meter intervals and detects the true race start.

2

Analyze Performance

View high-resolution speed profiles, elevation grade maps, and interactive course maps with heat-coded speed segments.

3

Identify Patterns

Compare your pacing against competitors. See exactly where you lose time, detect pace changes, and find critical segments.

4

Extract Insights

Generate professional PDF reports with detailed lap analysis, sector breakdowns, and performance metrics for coaching review.

Technical specifications

The system normalizes and analyzes GPX data with a consistent method across the entire course.

GPX Normalization

Resampling to fixed 10-meter intervals with Gaussian smoothing for stable speed analysis across the course.

Performance Metrics

Speed, grade, and pacing are calculated on the same distance reference for direct comparison.

Start Detection

Automated detection of the real race start using a speed threshold across a continuous 20-meter window.

Lap Detection

The system identifies repeated course sections automatically and enables detailed comparison of each lap.

Thesis details

Type: Long master thesis

Student: Birk Møller Gundersen

Timeline: 2025-2027

The thesis builds on established research in ski physiology and endurance performance while studying pacing, terrain management, and the link between course segments and final results through a clear data-driven approach.

Contribute to the research

Upload race data and help improve our understanding of what creates differences in elite ski racing.