ctak

My guess would be around 20%. From what I've seen in cat (a) Zwift races with 15 or more participants, there's often at least 2-3 riders sustaining pro-level output... this has been quite humorous to observe, particularly as these users are often in the sub 60kg weight range according to their ZwiftPower profiles.

Zwift threw out De Gendt from a race last year for pumping 7wpk during segments of a short 10-mile course before realizing who he was (and re-instating him), but they have yet to create a simple auto-flag formula for superlight randos capable of the same or similar...I'll never forget Armstrong's famous epo-infused climb up Alp d'Huez at 6.6wpk / 465w for 38min. Or Pogacar's 6.5wpk up Peyresourde for 25min, with 10min at nearly 6.7wpk. When I see noobs hitting numbers like this on Zwift, its either faulty / improperly calibrated trainers or blatant cheating and the impetus in either situation ought to be on them to appeal if they want to be included in race results.