TrainerRoad Outdoor Analysis at odds with PowerAgent, Cycling Analytics, Strava

Reviewing outdoor ride power data in TrainerRoad with PowerAgent, Strava and Cycling Analytics seems to highlight that TrainerRoad is at odds. I have seen this in all my outdoor rides but have drilled into today’s ride to illustrate the differences in detail.

Original data from PowerTap P1 pedals, via Joule head unit to PowerAgent, then Strava and on to TrainerRoad and Cycling Analytics

Unclear if/when time not pedalling is used to calculate NP, Avg Power- though PowerAgent does calculate for both scenarios (not included below).





Any ideas, or am I mising something? keen to make the most of TR features, but this may start to influence data as I start to complete more and more outdoor rides.


Stages PM into Garmin Edge 520. That gets synced to TrainingPeaks and Strava, then synced to TrainerRoad. As long as ftp is set correctly on Garmin, TP, and TR, all three are in agreement on normalized power, average power, TSS, etc.

More info starting with this post:


Strava does not calculate normalized power, they use their own average weighted power. That number is closer to median power.

Tr clearly shows your time at zero to low power situations. The power numbers from tr closely match the normalized power numbers from Garmin connect.

Cheers for the reply - info regarding Strava useful. There is a definite “error” in the TR evaluation of the power data during my 5 3 minute VO2 max intervals - it calculates an average power below my FTP, when the graphical presentation clearly shows all 3 minutes well above the FTP line.


If I’m reading the tiny, blurry print right, in the last screen shot, you have more time at higher power zones than in the TR shot before it. Also, somehow, you have 33 seconds of coasting in the TR shot. That makes me wonder if there isn’t something hinky in the data that affects TR.

In the TR shot, you have 2:22 of power and 0:33 of coasting; in the last shot, you have 3 minutes of power and no coasting. So, yeah, I’d expect differences given you’re comparing 2:22 of power (or 2:55, of which 0:33 is coasting) to 3:00 of power.

Maybe contact support and ask them to help understand where the 33 seconds of coasting comes from.

Meanwhile, here’s what I did for several months:

  • Record on Zwift, TR and Wahoo Bolt with Favero Assioma Duo.
  • Zwift FIT file and Bolt FIT file to Golden Cheetah
  • Zwift to Strava to TR
  • TR to TR

Ignoring Strava, which is just a data movement hub for me, and comparing kJ, NP, Avg/Max Power, Avg/Max HR, TSS, and a host of other statistics across the whole ride and across individual intervals, allowing a small margin for sampling differences:

  • All sources agree on GC
  • All sources agree on TR
  • GC and TR agree

Also, I don’t know about PowerAgent or any other power meter company’s analytics platform. The cycling community generally recognizes GC as a solid, reliable analytics platform, so I’m use it as a reliable reference standard. It’s also free and cross platform.

HI jmfm,

thanks for delving into my screen shots - it was the coasting error which caught my eye too - in those three minute intervals (ridden uphill) there was zero coasting. I will try GC I have used it in the past. Have taken your advice and contacted support.

Thanks again,