I actually thought that what Zwift was doing with zFTP matched what intervals did using the 2 parameter model you can select on intervals for calculating CP.
Intervals.icu has multiple choices for how they calculate eFTP - where the default is Morton’s 3 Parameter CP Model - If I recall correctly (which I know is a big “if”) a few folks mentioned that their zFTP matched what intervals showed for CP when they chose the M&S 2 parameter model in intervals.icu (which requires two max efforts - one 3-5 mins in length, and one between 10-30 mins in length - which does sound a lot like what Zwift requires for an accurate zFTP too).
Yeah, I don’t think it lined up with the 3 param model. Are you able to change it to the 2 param model? I think that was the one that most closely resembled it.
There’s a chance things might have changed if they made changes to zFTP since the original discussion we had on it a while back.
In both cases however those algorithms assume you’ve done max efforts within the points they use, so for instance the 2 param model probably won’t be that accurate if you don’t have a max effort between 3-5 mins long along with a max effort somewhere above 10 mins etc…
The 2 parameter calls my value 325 instead of 330, so it’s closer, but still rather from 300. Not that it really affects me at the moment. All of them put me in C (I weigh 100 kg), and, from there, it’s a matter of me doing my best against the people around me. I wish I had the 342 FTP (max 20 minute power 362 W) I had before getting Covid, but it’s kind of nice being in the front group and affecting the race now that I’m back amongst the Cs. Except when I was in the lowest division of B in ZRL, I mainly only dreamt of pack finishes as a B.
My Intervals account is set to the default single max interval plus Morton’s 3 p.
Usually Intervals and Zwift are very close (within 4) but if there was ever a sign difference, it was always Zwift being higher by 15 or 20 for no apparent reason then a correction 1 week later.
I don’t know why.
I’m pretty close right now between 2-param and zFTP (3 points difference), but yeah, something seems to be different given a delta between 325 and 300. Could be a difference in algorithm, or perhaps time to decay results (90 days vs. 60 days etc.). I guess either way nobody from Zwift has confirmed what model they are using, so it’s probably they started with a current model and made changes.
Personally I’m a fan of seeding a ranking algorithm with the modified compound score that zwiftracingapp uses. After Tim analyzed the data from a large amount of Zwift races the modified compound score seems to be very balanced across racers of a wide variety of weights, and race types. It only requires a 5 min best effort, which is better than requiring a max 10min+ effort like zFTP uses given how few people really do a max effort above 10 mins naturally in Zwift.
Thanks again, Paul. It makes sense. The new algorithm keeps flip-flopping me between B and C. I wish it would settle at B so that I can remove the decision anxiety when I prefer a trophy over a workout. Although I haven’t raced in C since I was first moved up to B, I have been very tempted to do so