This is last night Activity…you can see Cadence droping also when power is droping.
So yes he is using Start/Stop pedalling.
@James_Cavell_Amsterd is there a way that you can record yourself while doing a race?
To show your style of pedalling or racing.
I also race with Assioma Duo Pedals and my Power looks totaly different.
As others already wrote, a video would make it easier to understand how this happens. I don’t see how you can get such signals without actually coasting a lot.
I’ve been racing with Kickr and with Assioma Duo for dual for a few years, not in this exact course but I have done e.g. the Titan’s Grove from north many times, and neither Kickr nor Assioma shows anything similar to your graphs.
Below is a quick comparison of your race report and my analysis for the same Watopia segment, from the rollers and up to the banner at the Titan’s Grove Reverse KOM. My recording includes both Kickr and Assioma. My lowest cadence reading was about 61 rpm and the lowest power was about 120W (on some rollers), in a rather big group (ride or race depending on your opinion 
), with the usual surges.
In the comparison, notice that both graphs are with time (not distance) on the horizontal axis, so they should be rather comparable even if they are from different types of zwiftpower reports.
I’m pretty bad at holding steady power, but when comparing with your readings, my power looks pretty stable. Even if none of my devices have “race mode” and I always experience some lag.
(Edit: I just noticed Dejan wrote something similar right above)
you can see Cadence droping also when power is droping.
So this is OK, right? If it’s ‘cheating’, shouldn’t power not drop when cadence drops?
Cheating happens 2seconds before that…when you push and stop and the same POWER/CADENCE is reported to Zwift for 3 more Seconds but you ain’t doing the power at that time.
Do you see your power similar to this video 
This is really annoying.
I wish I hadn’t bought the sh*tty Zwift Hub One then tried to compensate for it’s lack of accuracy with the Assiomo pedals.
Now all my performances and improvements are tarnished.
This is so demotivating.
So you have 2 possibilities, assuming this is indeed hardware problems, and you want to resolve it properly rather than just vent.
The trainer is indeed under-reading by 8% at higher power output. Get in touch with manufacturer, and figure out how to get that replaced.
The pedals are over-reading by 8% at higher power output. Get that sorted via Assioma, and in the meantime switch to trainer power and begrudingly accept lost power & being in too high a cat. [fwiw, I had this problem 12 months ago when I switched from Juusto to Kickr bike and discovered my Juusto had been pumping my ego up and I forgot I had many months before offset my pedals to the trainer for indoor/outdoor matching! A few months spent getting my a** handed to me in As followed]
And if you wish to continue using the pedals, the other thing to check, are you displaying power as 3s average or 1s average? With 3s average there will be a slight delay in the game showing an increase in your power from say 250w to what you ‘know’ to be 300w, for example.
Oh, and the ERG mode issue you are having is a result of using a different power source to the resistance control. The trainer is trying to resist at 200w, for example, but pedals report 210w, so the trainer drops etc etc
Honestly Zwift racing is a joke. Hardware shortcomings mean you can unintentionally cheat while the A grade is full of riders who weigh 50.0kg and produce World Tour w/kg.
If you ride with a PM, the latency makes it really difficult NOT to end up in a spiral of accelerating (too much) then coasting.
PMs are more accurate than most trainers and their use should be encouraged.
Yeah, I am using the 3s recording.
The Zwift Hub is known to underreport at above 300w one it gets warm. At that price range it’s never going to be as accurate as the Assiomos.
I trust the calibrated and updated Assiomos. The Hub is not accurate.
In which case give the 1s power reporting a go next time you’re on, it could help to smooth your riding, save you energy & avoid future issues.
I can ride with my Assioma’s without seeing a delay (I use 1s) in power shown on screen, so if you continue to have issues, it could well be that you do have an a hardware issue either with connection or the pedals - in which case next step would be to try running zwift on a different device, see if problem repeats. Then fire up something like MyWhoosh and see if the delayed output is present there too.
Gotta drill down to find the cause before can find solution
PS - If that was me, and you can prove it is underreporting [assuming it’s not old & still within], then i’d be looking to get that returned/refunded as not fit for purpose, even though boxing/returning a trainer really is a pain in the hole
If you refuse to use Zwift Hub as the primary source, my advice is to keep your cadence above 50 rpm regardless of surges. Coasting only when you want to supertuck, never elsewhere. You would never coast when racing uphill outdoors, so no reason to do it indoors either.
If you still get such power readings, make a video of that.
Agree.
Could it be that some of this is because you have game setting to show 3-second averaged power (instead of instant)?
Thing is, I hit the climb with far too high a wattage, couldn’t hold it, started going backwards, started to think ‘I’m getting dropped’, coasted, then recovered, so accelerated again.
keep your cadence above 50 rpm
Outside of going downhill I am usually keeping it high.
Not at all. That only affects the number on the screen, not the recorded power nor your avatar.
(edit: Had quoted the wrong text)
When you look at it in Zwift Companion, it’s at a scale where you cannot see the near-zeros. Notice how different it looks in the race report:
It has jumpy power throughout the race; uphill and downhill.
Dejan already plotted the cadence further up this thread, and it drops in the same spots.
My response was to his observation he had that if he starts pedaling (ie. from 0 watts), the game screen takes 2-3 seconds to start showing realistic watts. This could well be because it’s averaging input of the past 3 seconds before showing a value (so 2 of those 3 seconds it’s using a 0 value).
Yeah I wouldn’t be putting up with a trainer than was underreporting by 8%. I’d be focussing my efforts getting in touch with the manufacturer.
Its quite easy to buy kit that you regret. I started out on a C2 bikeerg because i thought my wife would use it…that was £1k and ended up gifted to my father in law. 
Was this mention buried in one of the posts – how has it been determined that the trainer underreports by 8%? If comparing to an Assioma session where sticky watts was used, could that be the 8% right there?
could be, but I would guess that could give a lot more than 8% difference… 
It would be easy to tell if we just had a dual record from the Zwift Hub.
