Garmin and Strava power discrepancy

Why there is a such a big power diff between Strava and Garmin?
I had a race yesterday and Garmin says: 309NP / 278avg, and Strava 293NP / 268avg.

I didn’t have any pause or anything…

Your FTP is definitely set the same in both platforms?

They use different calculations for NP. Garmin uses the Coggan method. Strava did not want to pay for the trademark, so they use their own calculation.

And I also note a great difference in how the average power differs with their auto pause. But I have no explanation for that


I think is 5 watts different, but not sure why that would make any impact to avg or NP…

Is one including zero’s and the other not ? so when you freewheel that is included in the average ?

isn’t that a difference between avg and NP, including zeros and not.

I don’t think so, otherwise you wouldn’t see a difference between AP and NP on TR workouts, NP is more an average of what wht you would have averaged if you had ridden at more steady state (which is affected by low and high eadings)

Any why it’s a setting within Garmin that doesn’t affect NP

Strava seems to do it’s own thing. I get a discrepancy on calories in strava coming from TrainerRoad (and onto MFP). Even though “work” is the same, it seems to not take kj = calories and does something different. Seems to be only TrainerRoad affected, it’ll take the Garmin figure straight through.


Consistently I see numbers on my Wahoo head unit which are taken directly from my HRM / PM and then when the ride is upload to Strava they always drop a little (peak power / average speed etc).

They have a metric they call weighted average power which I think differs to normalised power in terms of how its calculated; but I’ve no idea / insight into what the difference is.

Guess it’s just one of those things arising from a difference in how the data set is analysed and presented.

install Sauce browser plug-in and you’ll get NP, IF, TSS:


as Strava uses a slightly different calculation to avoid using the trademarked terms.

Similar to how TrainerRoad doesn’t use CTL and instead uses 6-week average load.


I had one issue before with different avg numbers and it had to do with data dropouts. My stages l/r pm tracks not only the single data points. But keeps track of accumulated numbers.

Maybe you have the same problem?

They don’t use NP but “weighted average power” which is a copy of xPower, 25 seconds rolling average vs 30 seconds. Strava Ride Statistics – Science4Performance

There should be no difference in average power assuming it’s the same dataset.

1 Like

Todays example…

How’d the race go? What place?

This was a trainer ride