Zwift recording incorrect all time PB numbers and possible consequences

yesterday i did 616w for a minute in a zwift event, confirmed by strava/zwiftpower/trainingpeaks etc. but zwift itself is showing this:


i have never hit 623 or 625 so i am not sure where it is getting those numbers from. it’s only out by a few watts so it’s not a big deal but i’m flagging it because for racing purposes even 1 or 2 watts could make a difference to CE enforcement over a longer interval and i am not sure if this is just a visual bug or a reflection of the (wrong) numbers that zwift may rely on

Hi @S_A_Cestria_CC, I’m sorry that you’re having unusually high 1 minute power readings. I did notice your 623 watts were reported to our servers. There is something I noticed with your trainer setup that might be contributing to this. I sent you some more details about it to your email.

Thank you very much for response, I’ll try that out. To be clear, is it simply having the “controllable” paired to BLE that could be the hypothetical cause of this, or should I switch the power readout to FE-C also?

Hi @S_A_Cestria_CC, it could be a cause. For troubleshooting purposes make sure your controllable is paired via FE-C (ANT+) instead of BLE.

When I checked our servers, your connection method was pretty consistent, but on that day with that had the high wattage I noticed your controllable connection method change. So eliminating this as a possibility would be a good first step.

I would hold off changing how you pair else for the meantime. If you experience the high wattage again, then I would recommend checking your firmware, calibrating, and if those are okay then start using FE-C exclusively for pairing your trainer.

good old elite trainers over reading

for the sake of transparency i’ll clarify the actual problem: it didn’t actually overread because on the zwiftpower graph for the event (3105836) and in the .fit file on my public strava the correct value is shown, 616w. to paraphrase the staff response it’s possibly a visual error caused me having the controllable setting on BLE for the event instead of the usual FE-C setting i use

616 is a number i have hit outdoors, it’s in my profile. i haven’t hit 620+ yet.

I dont think its a visual bug, I think that most of Zwift’s data shown on ride reports is live data which is not necessarily the same as FIT file data. Example from a ride yesterday:

Zwift ride report - 1020 watts
Garmin (1030 fit file) - 975
Strava (Zwift fit file) - 968

I only know the ANT+ protocol but BT won’t be too different - ANT+ devices default to producing data at a rate of about 4Hz, that’s about 4-6 data points a second (no doubt some manufacturers “tune” this). Fit files capture at 1 sec intervals. I more often than not, get different numbers between Garmin and Zwift due to both devices capturing the 1 sec snapshot at slightly different periods. Both read from my Kickr directly.

Without empirical data, I am speculating that Zwift is showing “live data” in your instance, but it’s logical to assume that your trainer is sending more data than your fit file captures, and if your fit file is only logging say 1 in 5 data packets. It’s also worth noting that in the past if people have had data dropouts and fit file corruption, live data is still available in activity listing for them, so we know Zwift use and capture live data (CE discussions here have validated this).

I may be way off on this but the same issue stuck out like dogs bollocks for me yesterday as I’m peeved that Garmin didn’t capture the 1000w which for me, is unobtanium.


Thanks for the insight Dean, I’ve experienced similar where I see my peak power during a sprint and find out the actual recorded peak was maybe 20w less, though in my case my strava and zwiftpower are always within a watt or so which I assume is just a rounding difference, it’s only the zwift app that doesn’t seem to agree. I came to roughly the same conclusion as you, just a difference in sampling rates.

Never had a case like mine in this thread where zwift has recorded more than I actually produced before though… but your explanation is reasonable enough that I suppose it could happen.

In reality these differences mount up to maybe a fraction of a percentage of power over short distances so I’ll just chalk it up as “one of those things”, switch my controllable back to ANT+ as advised and mark this as solved

1 Like

Yes heaven forbid you have to ride in A races with those numbers

1 Like

Life’s too short and too easily lost to be spending it arguing with people on the internet Rich. Anyway, topic resolved