Power data is totally incorrect. Are you serious, Zwift?


This month (not so sure about previous one, but probably also), I found that I have more 1st places in zwift-races than before + I feel that race is definitely easier for me. Success? - I do not think so. I spent a lot of time comparing numbers from my neo vs SRM - numbers were not totally the same, but dispersion was not so huge + all this numbers were ~ near my expectations.

I checked my SRM, checked my neo - I compared in another programs and application - recalibrated/checked my SRM. When I keep constant power - numbers ~ near like I want, but in race I do small sprints all the time - and I found that it helped me a lot.

Today I decided to record race on all devices I have: neo + SRM on zwift-pc, android, wahoo-bolt, srm-head. But I suppose that the only comparison from zwift-pc and wahoo-bolt would be enough to show that zwift is totally wrong in result power-curve.

Please find attached image, sorry for the quality - it is necessary to zoom-in to see the numbers, but zwift has is significantly higher numbers than wahoo.

Unfortunately, I am totally disappointment with zwift - looks like I spend last months training on a wrong power number. And I disappointment twice - because all the 3d and GUI is not so important for me like the correctness of the number. And the company with a lot of developers and managers cannot calculate and check just one number - it would not be the problem if it would be number of “drops”, but the number is very important for all results of all races which happens :frowning:


I thought the Bushido was a smart trainer the broadcast it’s watts to Zwift. If this is true then the issue is with the trainer not Zwift.

The only time you use Zwift’s power curve is when you are using virtual power (using a speed sensor).

It also looks like there’s is a time difference in the images.

1 Like

I am pretty sure it was SRM, all previous times it was SRM. I have SRM for power+candence and bushido for controllable-trainer. Unfortunately, I did not point special attention to what exactly paired today, but all previous year it was SRM only + I suppose that zwift pairs the same devices like previous time.

The power-curve is not from zwift - is it from strava - from the same ride: one from zwift, another from wahoo bolt.

Ok, I’m not sure what you are getting at. Zwift does not manipulate the wattage that is being broadcast to it in any way.

On Zwift and the Bolt do you have them configure the same way for watts? For example are they both set to 3 sec average? Looks like that could Ben the difference you are seeing.

3 sec avg cannot create 40w diff in 10mins and 76w diff in 1min.

… and 3sec usually just for display, not for record in fit file.

Well some smoothing is going on if you look at the images.

No matter what the case, though, Zwift is not calculating your power if you are using any sort of power meter. Zwift is only taking the data that is broadcast from the power meter. So, your beef should not be with Zwift, but with one (or both) of the devices you are using to measure power.



It is the same device: SRM, just recorded on two different devices: zwift-pc and wahoo-bolt.

I do not know why I mentioned bushido (my prev trainer), because my trainer is tacx neo now. Changed initial text.

I do not think that any smoothness is necessary to show power-curve, yes - they are different for zwift and wahoo.

@Nigel_Tufnel @Paul_Allen
You assumptions are not correct, because SRM does not transmit any power - it sends it in CTF format. CTF it not direct power number, it is just a torque number which depends in delta in time and zero-average. On android all the logic is covered by ant+ library - and even it is not correct is calibration numbers are not correct. What is going on zwift-pc and how it works with SRM and power is unknown, but there are a lot of points when you can produce incorrect power number.

I found a lot of topic here about SRM calibration in zwift - probably something changed in it, but again - no one knows.

I don’t understand the statement “SRM does not transmit any power.” I have understood that SRM power meters do transmit power readings. I also read this on the srm.de website:

It is a bit of technology: SRM does not transmit power in watts, it transmits torque.

Power watts = [(Sensor output in Hz)-(Zero Offset in Hz)] / Slope) * angular velocity rad/sec)

    Please find that it is necessary to calculate angular velocity with good time precision, it was one of the problems of early quard d’zero powermeters.

Also, if you would check specification on https://www.thisisant.com/developer/ant-plus/device-profiles#521_tab site, you can find there are two modes for power transmission: PWR and CTF. I described CTF above.

Another day - another race.
Chat: “B” is too strong isn’t it?
I explained situation and pointer to the forum.
What I did today:

  • I checked that the only SRM is paired like power and cadence source.
  • Found another topic in the forum and followed advice: to unpair, then pair SRM, then do not touch pedals for 20s - probably it will zero calibration offset. I did, but it does not change anything:

Zwift 1m:

Zwift 10m:

Wahoo bolt 1m:

Wahoo bolt 10m:


How does the data from your smart trainer compare?

Be interesting to see if they reported different data or if they matched one of your current sets

I did visual comparison of my neo (connected to phone) and srm (displayed on wahoo) and numbers were pretty close.

Did not check neo in the mentioned races. All my races I did before were near ±300w/20m. and today’s wahoo shows 306w for 20m - which is pretty close to my expectations. Zwift shows 355w.

And it is the same data from the same one power-source-srm. I do not think any reasons to think that neo will show these high-numbers and wahoo is totally wrong and show low - I did a lot of tests on different powermeters indoor and outdoor.

Last night I did race recorded on 4 devices - I am quite sure that I found the root cause of the problem. I will post and update about it a bit later, it is not related to zwift only - you can show very high numbers in strava on the air too.

Also, I poster it in our local chat, and a lot of people helped to test different PMs, that is why you can expect this cheat in races, but I have to warn that probably someone would like to test it in official/important race, because it is not a secret bug anymore.


Similar issue with SRM Origin and latest Kickr. Power reading between the 2 is fairly similar normally, but if i use the SRM for power on Zwift instead of the kickr the numbers are way out. Zwift will be giving me 50W higher reading than what the SRM is actually reading on the head unit. No idea why, must be the method of data being sent as mentioned. Wasted money to buy a 4iii HR strap to use as ant+ bridge