Hello,
Month ago I created topic that something strange with zwift. I was not really sure what it is, but after last race I can say that the root cause of the bug was found:
My initial topic: : https://forums.zwift.com/t/power-data-is-totally-incorrect-are-you-serious-zwift
I found the same report ~1y ago: Power Output When I Stop Pedaling
and the same ~2y ago: https://forums.zwift.com/t/race-cheating/2484/13
Yesterday I shared results of my investigation in our local chat and after it most of the guys were able to finish in top of A-cat using this bug!
Protocol to reproduce is quite simple: Better to do it out of saddle: increase rotation, push the pedal hard (1 stroke of sprint) and stop pedaling immediately. You will find the zwift does not drop power immediately - your power output will be the last high value for ~3 more seconds (more on android). … relax during these 3-5+ seconds, repeat again (not necessary to sprint very hard to extract quite good advantage from it).
Even if you do not use the bug, you can try to catch the group - sprint far away from group - and you will catch it without pedaling. Someone during race mentioned that something wrong with draft - It feels exactly the same, but it is not draft problem. The worst case - you can repeat this bug all race long.
By the way - the problem affects not zwift only - most of the head-units affected: SRM, Android, Bolt (less than others). Right now I can show very high numbers in strava outdoors.
Everything was calibrated, and recorded from the same SRM on 4 devices, you can find the final protocol extracted from fit files. Bolt do the same - but drops power much faster - it is better, probably Zwift can fix it this way.
Also, it is easy to understand why it was necessary - to approximate power if you have ant+ drops, but it is a serious bug and cheat, even if you do not know about it - it can increase your power.
Regards,
Alexander.
–added–
Just to be sure about calibration on zwift-side, checked Log.txt and it is full of “NEW Calibration offset = X” with X from 520 to 530 - which is correct value for my SRM offset.