goose70

Your easy weeks appear to be nearly as hard as my hard weeks, and your hard weeks are way off my charts. Makes me wonder if I really should be building in recovery weeks at all.

Carbonfiberboy 

In 2020 I did a 3:15 ride with an IF of 1.00. I got a lot of flack for that back then, retested my FTP a couple times, but nope, if anything my FTP was optimistic. I happen to have had an unusual high end ability. I could outsprint anyone in the group of stronger riders I was with. The route had a lot of short steepish hills. I was on the wheel of a stronger, younger rider, and I wouldn't let him get away. So whatever the mechanism, it's possible. One just has to have that talent and the training, in my case years of chasing stronger riders on long hilly rides. I had 10 PRs, all on familiar roads, just had one of those days.

So it's really the artificiality of NP. My average power was only 74% of FTP, best 20' power 97%, weighted average power 90%.

PeteHski

Yes, my NP is often 50-60W higher than my average power on a lumpy ride. So that pushes IF considerably higher than a more steady state effort. I've also noticed that Strava weighted average power is always considerably lower than NP. So they must be using a different method for their equivalent of NP. I actually think the lower Strava weighted power is a better measure for me - in terms of matching my perceived effort. For example this afternoon I did an easy 2 hour ride with average power 135W. This gave NP of 195W and Strava WP of 173W. It certainly felt more like a 173W steady-state effort to me, but I'll happily take the higher NP!

RChung

I've noticed that, for my own data, Strava's weighted avg power is never different from the Skiba xPower calculated in Golden Cheetah. xPower is like NP, but rather than use a 30 second "box" running mean to smooth the data before doing the vector norm, it uses a constant decay smoother. Then it uses the same vector norm as NP. I don't remember what the constant decay rate is for xP; I suppose I could look it up.

[Edited to add] Ah, this seems to explain it: https://science4performance.com/tag/xpower/

GhostRider62

My feeble brain remembered, it was Dr. Philip Skiba. The attached paper is a bit complicated. Essentially the recovery of W' is curvilinear despite its expenditure being linear.

I was interested to understand an observation I had in my own performance and Dr. Skiba (on ST and Wattage, IIRC) did some posting that seemed to explain what I saw doing the TransAm bike race, where lots of short punchy hills would give a high IF seemingly too high. And then after the race, I could just smash hill and hill after hill for hours and get IF over 1 like carbonfibreboy did. So, I guess it comes down to how useful the math is to describing reality. I messed with TP, XERT, and GC a lot. Somehow, I just feel GC works best for me. Or, I could just be cheap.

https://journals.humankinetics.com/v...icle-p1561.xml

MoAlpha

Interesting. My IF can easily be 90 or a little above on an ill-advised 4-hr ride spent trying to keep up with stronger riders, and I bet it was in that range in races, based on how I used to feel. The FTP accuracy caveat always applies, but this never made sense to me.

RChung

That's a nice paper. I'd had a couple of exchanges with Phil and Dave about W' renewal prior to that, so I'm a little biased, but I think the general topic of "how fast can you spend down, and how fast can you renew W'" is pretty interesting. I think when Andy originally came up with the idea of NP as a way to "normalize" extremely variable efforts, we considered it a quick hack but a useful one. Almost immediately some riders presented data files that broke NP (that is, showed implausibly high NP for plausibly long durations). Andy called these "NP busters" but since (back then) no one foresaw how an entire ecosystem of training metrics and training prescriptions would eventually come to be based on NP, it was kind of interesting but NP was better than anything anyone had done up to that time, and we didn't want to let the perfect be the enemy of the good, so we lived with it. But the NP busters were always micro-bursty rides, so Dave and especially Phil wondered whether micro-bursty rides revealed something about the algorithm, or whether they revealed something about physiology and the dynamics of W' renewal. So this is a little long and convoluted, but the overall story is kinda fascinating, and encouraging: Andy knew that micro-bursty rides "broke" NP, so he just said NP shouldn't really be used for micro-bursty rides (in the same way he's said NP shouldn't be used for rides shorter than maybe 20ish minutes). Phil saw that there was a pattern in when NP was broken, and asked himself whether it revealed something interesting about micro-bursts. Pretty cool.