Why does the speed chart for my activity in Strava show a sawtooth pattern?

If you’ve ever uploaded a ride or run from Cadence to Strava and noticed a strange “sawtooth” pattern in your speed graph, here’s what’s happening behind the scenes.

Cadence records both speed and distance once per second. The speed values come directly from GPS or connected speed sensors—they’re instantaneous, accurate, and smooth.

The distance values are recorded alongside them so total distance remains correct.

The issue isn’t with the data Cadence records—it’s with how Strava handles that data after upload.

Why it Happens

GPS location updates don’t always arrive exactly once per second—they’re sent asynchronously by the operating system. Cadence samples them every second, but the two clocks don’t always line up perfectly. When a GPS update arrives slightly early or late, that one second of recorded distance can look a little irregular.

The same is true for Bluetooth speed sensors. They count wheel revolutions, and one second might include three revolutions while the next includes four, creating a rhythmic distance pattern.

Normally that’s not a problem. The speed and total distance stays accurate, and most platforms display the smooth, correct speeds that Cadence records.

Strava Reprocessing

Strava automatically reprocesses uploaded files when it detects anything unusual in the distance data. Even a small timing mismatch can trigger it. When that happens, Strava discards the actual speed values from Cadence and replaces them with new, estimated speeds that it calculates from distance alone. And it’s not unique to Cadence—other apps and devices can run into this too.

Those invented values are what create the jagged “sawtooth” pattern in Strava’s speed chart. The same file looks normal in Cadence, Intervals.icu, Garmin Connect, and other apps because they use the original recorded speed data.

Unfortunately, Strava doesn’t publish what triggers its reprocessing or how the recalculation works, so it’s often unpredictable.

Post-Processing

Cadence could try to work around this by recalculating all distances from the GPS or Bluetooth sensors live speed before export, which would stop Strava from reprocessing. But that would also slightly alter your totals—distance, average speed, and more—which isn’t ideal.

Cadence keeps your original, accurate data intact. The sawtooth pattern you see in Strava isn’t a recording error—it’s simply the result of Strava replacing smooth, real speed data with its own estimates.