NP lower than Average Power (ZwiftPower)

this is of course mathematically impossible in a 1hour ish race, but I’ve seen it happening a few times recently.

Is this a known problem, or does anyone know why?

Did you wait for the Bolt to go from Green to Blue? When it’s Green it will sometimes show NP Lower.

it was a blue bolt. this is not my idea but it might be to do with internet dropouts affecting the ZP data, i think. apparently on strava the fit file is correct, but on zp there are dropouts

i think i remember i had one fondo some years ago where i was dealing with roadworks or some BS outside my house and it was like 2 minutes online, 2 offline etc the whole ride, so strava had a fit file with no dropouts even though i spent half the event offline. i can’t remember what my ZP data looked like though

Someone else did a bit of analysis on the fit file which was just sent over to us last night, and it looks like if you use the full fit file it spits out the NP on ZwiftPower… but if you use only the average for the race itself, it spits out the average power.

I haven’t fully confirmed that yet, but it looks right.

But I’m sure myself that isn’t what it did in the past, or there would be many instances of super low NP due to extended zone 2 efforts after a race. So maybe it’s something to do with that specific course, or just a bug that uses an extra bit of time (but not the entire file?).

so we’re down to two main theories at the moment, and a strong chance it’s neither.

  1. drop outs online (though surely this would affect both average and NP?)
  2. average and NP calcuating based on different durations.

Is it likely we’ll ever have someone look in to the code on this one or will we all be left just guessing?

not my style. I’m curious so I wanna know

@James_Zwift sorry in advance if you’re busy or not the right person to ask about this, but I have no idea who else at zhq knows anything about zwiftpower. Sheridan has all of the details so he can provide more info if needed, but he and myself are looking for clarification on what appears to be an interesting ZP bug/anomaly

This is not my screenshot but apparently the circled rider has a processed activity. There is also another example in this screenshot just below the person with red VI:

If you can answer, do you know what is causing this? I suspect internet dropouts, but I don’t know

Thank you brother

i’ll send a search party out for james hodges if mr bailey doesn’t know. i kinda just think it’s an internet dropout quirk though. iirc ZP takes the fit file from your zwift feed and the actual power from your live data (in case of private activities etc i guess?) i guess that could lead to discrepancies. i probably don’t recall correctly though


good idea, but i don’t have a fax machine. just an annoying personality

They deserve you. I have experience taking dictation and will draft a letter. Send me a cassette tape.

1 Like

definitely something wrong with zwiftpower. the VI is taken from the blue bolt. i aint sending no email though