Strava, Roller Skiing, Rolling Resistance and Power Data – What Large-Scale Segments Reveal About Performance

In our recent posts, we’ve talked about the power output we measure for cross-country skiers. Since we are the first to measure real power in skiing, there isn’t much historical data to compare with — and it’s still challenging to put our numbers into context. To better illustrate and understand what performance levels we can expect, we spend time calculating skiing performance — just as we did for Lysebotn Opp (Musgrave Smashes the Lysebotn Record – SKISENS).

Using Strava segments to benchmark roller ski performance

To avoid limiting ourselves to a single race, we took a broader approach. A huge amount of data exists in Strava segments recorded by active roller skiers around the country. However, this data comes with several uncertainties that we cannot control. Beyond obvious sources of error such as GPS inaccuracy, we also rarely know which wheels were used or whether the skier actually went all-out or simply skied through the segment.

Based on the principle that a large dataset always contains enough valuable outliers to reveal a trend above the noise, we downloaded distance, time, and elevation data for more than 100 frequently skied Strava segments across Sweden. For those curious, we focused on Östersund, Falun, Ulricehamn and Gothenburg — all hotspots for roller skiing. We set a minimum criterion: each segment must have been skied at least 100 times by at least 50 different athletes. The results are illustrated below, where our empirically developed power profiles are compared against calculated average power from Strava data.

Skisens power profiles compared with watt/kg calculated from Strava segments — women on the left, men on the right.

How Strava performance aligns with real power profiles

The figures above clearly show that the Strava-based data follows the same trend as our measured power profiles — even though some natural spread, caused by uncertainties in the input data, is expected. For reference, we assumed a standardized rolling resistance of µ = 0.018, corresponding to fast “2” wheels on dry asphalt.

If an athlete used faster wheels — for example racing wheels — we would overestimate their power, placing the data point above the profile. If they used slower wheels (3s or 4s), the point would fall below the profile. The magnitude of this error depends on the steepness of the segment. For shorter segments, we focused primarily on uphill segments where wheel choice has less impact.

Why female data points are fewer — and often lower

Another clear reason why many points fall below the line is simply that some segments may not have been attacked by strong skiers. This is particularly noticeable among women. There are significantly fewer Strava recordings from female roller skiers, which is reflected in the smaller number of valid data points meeting our criteria.

What this means for roller skiing, power measurement and performance tracking

Despite uncertainties in Strava data, the trend strongly highlights that our power profiles reflect the true power–duration relationship for a large population of roller skiers. The spread in the data also underscores how important it is to measure real power — or have precise knowledge of external conditions — if you want to track performance reliably over time.

As roller skiing continues to grow — on Strava and in training — accurate power measurement, rolling resistance awareness, and watt-based tracking will become crucial tools for athletes aiming to improve speed, efficiency, and technique.