It seems odd that my zFTP, which is described as, “the power you can sustain for a long period of time, i.e. > 40 min,” is higher than my 40 minute peak power.
Is this some kind of special zMath?
zYes.
I guess the calculation thinks that your power curve indicates that your 40min result isn’t the best you could do?
This isn’t odd. What would be slightly more odd, and there have been some examples, is when your zFTP score is significantly less than your 40 minute power.
If I understand correctly, Zwift’s estimates stem from a critical power model. Consider the jagged line representing your best ever power at various durations on your fitness page. Those are what you have done, but your power at every duration is not necessarily what you physically could do. For example, I’ve been doing sweet spot training for a couple months and I’m only just starting to add threshold. It’s likely that my 1-2 min power is higher than what I have done recently.
A critical power model is an estimate of what you could potentially put out at various durations. The estimate stems from a mathematical model. It is not a perfect estimate because models are not perfect. If you want details, you should Google, but be aware you may find multiple CP models. I know the one Zwift uses takes 2 inputs, that is it will find your best recent short term (4-6 mins or so) and longer duration (12-15ish?) max powers, and calculate the curve from there. So, it’s not just something that Zwift cooked up out of their heads. They had to do at least a bit of research.
You can see this forum post for a comparison between an empirical power curve and a model-estimated potential power curve.
Anyway, back to the question. Perhaps you are actually capable of higher 40 min lower than you think. Or perhaps that power estimate is off. My zFTP (model based) is lower than my recently tested FTP - I did that test, and now I’m doing sweet spot intervals off that test, and my default power settings for sweet spot and threshold intervals are 90 and 95% anyway. So, for the last 42 days, the model has not been receiving data that are reflective of my best potential efforts.
it used to, back in early 2022. i guess it still sort of follows the same principle, but it aint cp anymore