kkapdolee

If im understanding this correctly, and assuming i got my FTP down correctly, the IF and TSS formula dictates that if i maintain a power output of twice my FTP, i can go for 15 minutes and feel the same amount of physical stress. I can and i have gone FTP for an hour or more before but there is no way i can go FTP * 2 for 15 minutes.

Also, im a little skeptical about how power outputs are weighed in NP formula. First raised to the fourth power and summed up then taken a fourth root? My VI is very high sometimes greater than 2 but i guess that could just be because im a noob. By the way, im using Powertap C1 as my powermeter which i hear is pretty good.

In your experience, are NP, IF and TSS reliable indicators?

valygrl

Yes, they are reliable measures. No, you can't do 2xFTP for 15 minutes. Don't worry too much about the NP calculations, those are how those measures are defined, so you can't just decide they are wrong. VI can be high, it just means your ride is really variable.

You should get the Hunter & Allen book Training and racing with a power meter to understand how to use and interpret the data you get from your powermeter.

If you are using a correct FTP, your TSS should be a pretty decent descriptive measure of how much training stimulation your ride provided - but it may still feel different depending on how you rode - long & slow vs short and fast, may have the same TSS.

RChung

Your understanding isn't quite correct.

First, NP isn't terribly reliable for shorter intervals. Coggan recommends that you don't use NP for intervals shorter than 20 minutes (though that may be a bit conservative). TSS depends on NP, so TSS for short intervals probably isn't reliable, either.

Second, NP was originally envisioned as a way to approximate "equivalent" demands of a steady state power output with a highly variable output -- like the efforts for a crit race. The proof of a pudding is in how it tastes, and the NP for an hour-long crit race where riders say they've been going *hard* is very close to the steady state FTP -- much closer than the average power for that race is. So the empirical evidence is that in this kind of situation, NP is closer to FTP than average power is.

Are you familiar with how to calculate a standard deviation of a random variable? It can be described as the square root of the mean of the squares of the values. In statistical parlance, there is something called the Lp-normalization where you take the pth root of the mean of values x^p. So the standard deviation is the L2-normalization. Coggan's algorithm is the L4-norm of the power (actually, it's the L4-norm of the rolling 30 second mean of the power), which is where the name "normalized power" comes from.

Lots of people have tried different p's in for the Lp-normalization. Lots of people have tried different smoothing intervals than the 30 second rolling "box" smooth Coggan originally came up with. (For example, Phil Skiba uses a fixed decay smoother in xPower). There's nothing sacred about using a different p or a different smoothing window, but p=4 and 30 seconds puts you in the right ballpark.

So, when your FTP is well-measured, and if you don't try to infer too much from very short intervals NP, IF, and TSS are pretty useful summaries, imperfect though they may be. Try not to let the perfect be the enemy of the good.

Heathpack 

Hmm. I knew of course that NP is not really reliable with short intervals. And I knew that TSS is based off NP. But I never really put it together to conclude that lots of short intervals makes your TSS unreliable. Makes sense.

I'm not terribly caught up in the TSS concept but still out of curiosity, I'm wondering the magnitude & direction of the deviation.

For example, say workout 1 is two twenty minute threshold intervals, 5 min rest between sets, and a total of 30 min warm up/cool down. So X TSS over a 1:15 workout.

Then workout 2 is also 1:15, same warm up/cool down but this time 30 min of over/under intervals (say 1 min each at 107%/93% FTP) and 10 min intervals at 1 min 150% FTP/1 min spin with 5 min rest between sets.

Then workout 3 is 1:10, same warm up/cool down but now 3 x 10 x 112% FTP with 5 min RBI.

The TSSs for workouts 2 & 3 are over estimated relative to workout 1? How far off are the TSS values from 'true'? 5%? 15%? Does this question even make sense, given this is just a mathematical formula trying to describe a non-mathematical biological thing in which there is no 'true' known training stress?

PepeM

RChung basically covered it all. Just want to add that if you have gone FTP for more than an hour then, by definition, your FTP value is wrong.

revchuck 

You're reading something that isn't there. As stated above, NP isn't necessarily accurate for short intervals, but gives an estimate of your training stress as it varies over time. If you do an hour's ride that includes hard intervals, recovery time and endurance pace riding, your NP for that hour is going to be a good estimate of the training stress *for that hour.*

Seattle Forrest

That's an extreme edge case. If you ride normally for a month, you'll find that the NP/TSS/IF of your rides will generally be a good approximation of how much effort you put into them.

kkapdolee

I don't think that's right.

According to TrainingPeaks, who I think invented the TSS (or at least the current variation in question),

"While individuals will tend to differ in how much training they can tolerate, depending on their training background, natural abilities, etc., the following scale can be used as an approximate guide:


TSS less than 150 - low (recovery generally complete by following day)
150-300 - medium (some residual fatigue may be present the next day, but gone by 2nd day)
300-450 - high (some residual fatigue may be present even after 2 days)
Greater than 450 - very high (residual fatigue lasting several days likely)"

You are saying TSS that is greater than 100 is impossible because 1 hour sustained effort on FTP is IF of 1 and TSS of 100.

RChung

Well, maybe my post didn't make sense. I'm not saying that intervals per se make your NP (or TSS) unreliable. I'm saying that NP was designed as a quick way to summarize a variable series of data that is more accurate than looking at the most common quick summary of variable data -- that is, the mean -- to characterize its "center." It's kind of an ad hoc summary (that's loosely based on a physiological model), but empirically it works pretty well in the sense that when you know your FTP well, the NP from a *hard* variable-effort ride is almost always closer to your FTP than average power for that hard variable-effort ride will be.

The extension from NP to TSS is, once again, an ad hoc model that, empirically, works pretty well. You can do extreme things that will test the limits of the NP and TSS models. For example, TSS isn't additive but we mostly treat it as if it were -- and that usually works reasonably well as long as you don't throw extremes at it. It's possible to "break" the NP, IF, TSS, and PMC algorithms if you really want to, but the NP algorithm turns out to be fairly robust so you kinda have to try hard.

PepeM

No that is not what I am saying. What I am saying is that if you were to pedal constantly at what you thought was your FTP and went longer than one hour then, by definition, that would not be your FTP since your FTP is the power output you can keep for an hour.

kkapdolee

Thanks for the explanation. The Lp-normalization was very informative. What do you mean by 30 second smoothing window? How is it calculated?
I'm new to the whole power meter thing so I'm not exactly sure how often power is measured on my Powertap C1 unit. Is it measured a few times per stroke so the higher my cadence the higher the frequency of measurement?

If I recall my Physics correctly, power is defined as Work (Nm) over Time (s) which is the same as Force (N) x Distance (m) over time (s). By 30 second smoothing, do you mean that power data is combined every 30 seconds so the total work done in the first 30 seconds is divided by 30 seconds and that is the first data point while the next 30 seconds of work divided by 30 seconds is the second data point used in L4-normalization?

RChung

No, you're confusing TSS with FTP (and IF). Trainingpeaks is talking about TSS. You can certainly do more than 100 TSS points in a ride. However, it's very hard to do more than 100 TSS points *per hour*. PepeM was commenting on your claim that you've done FTP for more than an hour. If you can do more than 105 TSS points *in an hour* then your estimated FTP is *probably* off.

PepeM

What head unit are you using? Is it a Garmin? If so, set it to 1 second recording (smart recording off) and no autopause for best accuracy.

kkapdolee

So you're saying if I keep pace at my "correct" FTP, then it would be impossible to go more than an hour "by definition"? That was my original understanding of what TSS was before I checked what Training Peaks had to say. If you see my quotes above, Training Peaks says a TSS of less than 150 is considered a low TSS session. According to your definition of FTP, a TSS of 150 shouldn't even be achievable.

kkapdolee

I see. That makes sense now. Thanks for the clarification.

PepeM

It shouldn't be achievable in an hour. But you can ride as long as you want and the longer the ride the higher your TSS will be.

kkapdolee

Thanks. I have Garmin Edge 1000 with smart recording off and 1 second recording already on. I do have autopause when stopped but I guess that would impact the NP calculation.

Seattle Forrest

You can get 100 TSS in 1 hour. You'd do that by riding at FTP for the whole hour, or averaging out to FTP. Of course that assumes that you know your correct FTP; if the computer says you got 500 TSS in an hour that means your FTP is too low.

Also you can get 150 TSS in 2 hours. Or some other number below 200.

You'll be able to keep riding after 100 TSS in 1 hour, but you'll have to back off in terms of effort. After an hour riding at FTP you'll be cooked and have to ease up.

PepeM

Autopause will impact all of your calculations, you really need it off to get the best out of the data.

RChung

Coggan was looking for a simple way to capture the physiological observation that the response to a change in stimulus has a lag with a half-life of 30 seconds or so. That is, you can do a sudden spurt of power for a few seconds without much consequence to your aerobic system (short spurts are mostly fueled non-aerobically, while NP is supposed to capture something about aerobic output). So he used a running 30-second mean of power to smooth out those spurts that he's not interested in. Then he used the L4-normalization on the sequence of running means.

The C1 samples torque at (I think) 100 Hz but the ANT+ protocol that transmits between the ring and the head unit at only 4 Hz -- and most head units only keep 1 out of every four packets so when you download data from the head unit it comes out at once per second. There are some power meters that let you get access to the higher frequency data but they're messy to use. The simplest way to think of what's happening with the C1 rings (and SRMS, Quarqs, Power2Maxs, and other crank or crank-spider based power meters) is that they sample torque at relatively high frequencies (like, 100 Hz) but then bundle up those torque samples with some estimate of crank rotational speed to get power, which is sent out at a maximum of 4Hz.

Heathpack 

Well I guess "unreliable" was not exactly what I meant. More that intervals will cause your TSS to trend in a consistent direction (when compared with the TSS generated from more steady-state workouts).

But this is becoming a somewhat circular question. TSS is higher with short interval workouts because the model predicts they create more physiologic stress, so its expected I guess that the trend would consistently be towards higher TSS than steady state workouts, which I guess is my feeling from actually doing both types of workouts. So... forget it.

kkapdolee

A follow-up question that relates back to my original question of how I cannot do 2 x FTP for 15 minutes:

Here are the relevant formulas I have.
NP = L4-normalization of 30 second mean power data.
IF = NP of a session / FTP
TSS = IF^2 x number of hours in the session x 100

Assuming I have my FTP down correctly, I can do no more than 100 TSS in one hour.
Assuming any NP, IF, or TSS under 20 min time is unreliable, let's say I do a 30 min sustained power session.
Maintaining a power output of sqrt(2) x FTP for 30 minutes should yield 100 TSS.
( (sqrt(2) x FTP) ^ 2 x 1/2 (hrs) x 100 ) = 100 TSS

But can you guys really keep sqrt(2) x FTP for 30 min straight?
Sqrt(2) is about 1.4 so if your FTP is 200w, can you maintain 280w for 30 minutes?

..

Actually, I guess if "not exceeding 100 TSS in one hour" is the only constraint, I don't have to do sqrt(2) x FTP for 30 minutes "straight". I can have rests in between. I could do a 10 minute ride, 15 minute rest, another 10 min ride, 15 min rest and the last 10 min ride for the same 100 TSS in one hour... But then, wouldn't keeping your bike computer on (autopause off) during rests lower your NP and you will have to do greater than sqrt(2) x FTP for each 10 minute ride?

I know I couldn't do 1.4 x FTP for 30 minutes straight but maybe I can do 3 ten minute sessions of 1.4 x FTP with 2 fifteen minute breaks in between. I'm still not sure about the affect of breaks on TSS.

Seattle Forrest

We can all go over our FTP for a limited time, it's called burning a match. I can do 1.4x FTP for maybe 5 minutes. I'd have to check the data to know for sure.

If you're using Training Peaks or Golden Cheetah, look at the Critical Power chart.

PepeM

I am not quite sure what you're trying to get from TSS exactly, but it seems to me that you're going the wrong way about it. As you seem to have figured out already, you won't be getting 100 TSS in a thirty minute session. You shouldn't be basing your workouts on what TSS you think you would like to get.

RChung

It's better to think of NP (and the quantities based on it) as descriptive of what you did than predictive of what you can do. (FTP, on the other hand, is kind of predictive of what you can do.) Just because your mean income for a year is $100,000 doesn't mean you can work for six months at a rate of $200,000 per year. Just because there are lots of ways to get to the same TSS for an hour doesn't mean all of them are possible for you.

BTW, I've done more than 100 TSS in an hour. I think my max is something like 103 TSS. Coggan has said that he rarely sees a data file for someone with a well-estimated FTP where the NP is more than 105% of FTP for an hour effort (but they exist). A 1-hour NP of 105% of FTP means an IF of 1.05, so since TSS = IF^2 * duration in hours, that could mean a theoretical TSS of 110. But that's rare, and usually if you get a TSS over 105 it's a signal that your FTP is *probably* underestimated.