Normalised power

Ok then, I cant get my answer elsewhere so thought I would try here.

NP or average??
1. If my NP shows 326 and my average is 315 for an hours ride, what is my FTP!
2. On my other training rides and there is a significant difference between the two (as like today when my 3 hours ride av was 228 but the NP was 268) which is most important?

Thanks

That's tough to say, since we don't know whether you went "flat out" or not.

Normalized power should be more indicative of the physiological strain.

I was trying. It was a hilly course

Meaning that you were at least trying to go "flat out"? If so, then I'd say that 325 W is the better estimate of your functional threshold power.

Thank you for replying.

You're welcome.

well its nice to be nice to the nice....

__________________
Chann McRae Cycling Coach
Velossimo Racing Team
Racing Blog
john@cmcoaching.com

I'll probably make a fool out of myself as usual, but I have to ask....

If you don't go "flat out", isn't it probable that you could have sustained a higher AP? So maybe NP will overestimate but FTP would still be higher than AP for the ride?

If you don't go "flat out" (assuming that by that you mean a maximal effort over the duration) then by definition both the NP and AP will be less than your FTP.

But wouldn't that mean that NP is always a better estimate than AP? Which from acoggan's answer does not seem to be the case.

I assumed that there was a universal definition of "flat out" that everyone else but me knew about. I would assume that "flat out" would mean going as hard as you can for the entire duration (no variable pacing, no rest periods).

While what you say is technically possible, it would be tough to generate an "NP buster" sort of ride w/o going close to "flat out". That, plus the fact that beldon's normalized and average powers aren't very far apart, says to me that his functional threshold power is closer to 325 W (i.e., his normalized power) than it is to 315 W (i.e., his average power). Indeed, it is possible that his functional threshold power is greater than 325 W, although it certainly sounds like he was giving it "full stick".

In this case, the low VI (1.035) indicates that it was a very steady ride, just not perfectly steady.

In the bigger picture, there was a less than 4% difference between AP and NP, and even NP's creator will tell you that +/-5% isthe rough margin of error.

I'd have split the difference and gone 320 to cover my bases.

Maybe I should rephrase my question:
When is AP the better estimate?
If you don't go hard at all then then FTP>NP>=AP. You state (and I believe you) that if you go hard NP is a god estimate of FTP. But when is AP the better estimate of FTP?

Thanks,
J

For normalized power to not be the better estimate of functional threshold power would require that it overestimate to a greater extent than average power underestimates. That can happen, but usually the opposite is true. For example, even restricting the analysis to "NP buster" workouts (during which normalized power is, by definition, more than 5% higher than functional threshold power for a ~1 h effort), normalized power is closer to functional threshold power the vast majority of the time (i.e., for 23 out of 27 cases that I just examined).

EDIT: Some other stats for these 27 "NP buster" files...

Average duration (h): 0.99+/-0.13
Average average power (W): 227+/-53
Average normalized power (W): 308+/-39
Average VI (as already mentioned): 1.38+/-0.28
Average IF: 1.08+/-0.04

Average error if you used average power to estimate functional threshold power: -19+/-12%

Average error if you used normalized power to estimate functional threshold power: +8+/-4%

Correlation between average power and functional threshold power: R^2=0.69

Correlation between normalized power and functional threshold power: R^2=0.90

Remember, this is for a selected group of files for which, by definition, the normalized power algorithm failed to live up to it's billing...

I think the best answer to that question is "rarely" (but not "never").

Note: the average VI for the four files in which AP was the better estimate of functional threshold power was 1.21, which is actually less than the average for all of the "NP buster" files I've collected (which is 1.38). IOW, the degree of variability doesn't seem to be a distinguishing factor. Similarly, I have been unable to identify any common characteristic of these files - or for the 27 "NP busters" that I have collected - that would explain what makes them different.