Avg/NP power differences between Garmin Connect and TR ride summary

Hey guys, I’ve done a search for this topic and there are some that are close, but not exactly what I’m asking about. I’ve noticed that on a few of my rides some differences in how the overall ride is tallied.



Total Joules are the same in both, so average power should be the same, but they are not. I can’t confirm normalized power calculation since I don’t know how it is done and don’t have a training peaks account to upload to that, but with Garmin’s average power being correct based on total kJ I am calling out TR’s NP calculation as suspect as well. TSS is then vastly different, since TR is using only moving time (which I’m fine with), but then if only using moving time, shouldn’t average and normalized power be higher with TR?

Also as another nitpick… reported Speed in TR is average speed, not avg moving speed, yet time is still reported as moving time. If calculating based on moving time only, avg power, speed, NP should reflect that as well.

I’ve had the same issue with my wahoo bolt, strava and TR. My average power from the wahoo is always almost 30+ watts higher than what strava calculates which then is sent to TR. I’ve tried the dropbox upload instead of strava but its still less and drops my avg speed as well. My NP is usually fairly close (within several watts) as well as kJ. My TSS doesn’t vary as much as that though so I usually ignore the rest as I don’t see 2-5 points difference much of a deal breaker. My FTP is the same on wahoo & TR but no setting on strava since I don’t pay for summit, which from what I read through on the strava forum doesn’t resolve that issue either. It is very annoying though when all the numbers are different.

I filed a support ticket about that, drives me crazy because my brain does quick math on the numbers and average speed uses elapsed time, but it’s not shown.

Usually my power numbers align between GC, TR, TP, WKO4, and GoldenCheetah. Had a recent ride where they did not, and GoldenCheetah agreed with something. Can’t remember exactly, it was a recent outside ride. Will look it up if you ask.

Seems like there’s only really a difference if there’s any real stopping, seems to throw a wrench in the NP and AP calculations. It’s happened a bit on a few road rides where I’ve had to sit at lights a bit more than normal. When I do a gravel ride, there’s almost no stopping and the numbers match up perfectly.

My guess is it’s all to do with AutoPause (or pausing when stopped). I believe TR and Garmin handle it differently.

Are you using auto pause? If so try disabling it for a ride and see if that makes the numbers match up

For rides with stops, and auto-pause enabled on Garmin Edge 520, I have two different average power numbers:

  • TrainerRoad, Xertonline, and GoldenCheetah agree on a lower average power
  • GarminConnect, WKO4, TrainingPeaks agree on a higher average power

no auto pause going on. I was originally thinking that Garmin was doing that automatically, but when I calculated AP, it was based on total time.

do you have any GarminConnect data screens activated? I recently found a difference in Garmin and another platform based on the fact that I had a data screen (DozenCycle) that by default was calculating an estimated power. for some reason TP was using that as the power signal vs my PM, which was being prioritized by garmin. Max power of 5kW on the garmin seems suspect.

Hey there @Bioteknik,

The way that we calculate your average power differs from Garmin in that we use elapsed time to get a more accurate (from a training perspective) picture of your effort.

We did it this way because if you are taking large stops and going hard in between, then we believe your average power should reflect the recovery that you took during that break.

Also, as a side note, your 5000-watt power spike should have been filtered out, but that is an issue that we are aware of and currently looking into :+1:.

I hope this helps clear things up a bit!


Completely agree. I think Garmin/TrainingPeaks get it wrong, and TrainerRoad/Xert/GoldenCheetah get it right.


Garmin does use the elapsed time, bu now I’m seeing that there is a different total for total kJ, which both work out as correct…

So then the question is why such a large difference in total kJ? The power spike at most accounted for 5 kJ difference in total work, not 88

And yes, off road riding does tend to get some silly power spikes from descending.

Seems the biggest difference is because of using Strava to upload to TR. The ride is recorded in Garmin, pushed to Strava, pushed to TR. The total kJ in TR is closer to the Strava file, which may be smoothed. I know that the running pace data is heavily smoothed when pushed to Strava. My running intervals look perfect in connect, and look like a bunch of random paces in Strava.

Why not skip Strava and have GC push directly to TR?

My TR has both GC and TR sync.

That’s a question for TR, I have both Strava and GC sync’d, not sure if I can set a preference or not.

1 Like

Without having both workout files it is impossible to say, however, it is likely that we filtered out some of the power spikes that Garmin did not, thus lowering your total kJ. For a MTB ride with a lot of power spikes, this could easily equate to a loss of 88 kJ.

Is there a known issue with the way TrainerRoad pulls the data from a multisport file? I recently competed in Ironman UK and my garmin connect shows NP=197W whereas TP and TR show NP=184W.

When TR initially pulled the data in it also showed my swim and run time so I trimmed the fit tile (using fitfiletools). When checking the TP data is seems like it is also including the swim time in the calculation, likely being what is leading to the lower value. FWIW my garmin during the event ties in with the higher value.

Like I mentioned above, we base your calculation on your elapsed riding time rather than the total moving time that Garmin uses for their calculation. This will lead to slightly different AP/NP readings.

That being said, I would think if you properly trimmed out the non-riding portions of the event (and you didn’t take any breaks on the ride :wink: ) , then both Garmin and TrainerRoad should calculate similar values. After the trimming, what value did TrainerRoad display for AP/NP?

88 kj is a huge amount to trim from the file when we’re talking about at most 3 to 5 seconds of spikes. Im thinking about stopping the sync between strava and tr… but not sure if my tr rides will get pushed via gc to strava. All this does is makes the performance chart inaccurate for the stress incurred for a ride. It’s already hard enough to gauge overall stress of an off road ride when judging only by power at the pedals. Strava definitely compresses the visual charts of rides and runs compared to garmin connect and usually has lower kj than the raw file, while gc matches exactly with the device.

If you “break” the connection between TrainerRoad and Strava, the workouts should still sync through Garmin Connect. You are recording your ride on a Garmin head unit, so first, the ride will upload to Garmin Connect. From there, Garmin Connect will send the workout to both TrainerRoad and Strava.

And I agree that 88 kj is quite a big difference for just power spikes, but without seeing the workout files I can’t know for sure what is causing the discrepency. I would recommend reaching out to our Support Team at support@trainerroad.com so that we can dig deeper into this issue.

Be sure to send links to both workout files so that they can compare the two and reach a definitive conclusion :+1: