Training based on TSS instead of HRS



RChung said:
I suspect that's it: I believe Andy lives in a pretty flat place so he and his wife probably rarely see rides with longish coasting intervals.

Not flat, but not mountainous, either. Thus, although you need a 39x25 for many of the hills, if you ride by yourself you don't waste much time coasting because the descents are short.
 
djconnel said:
I'm not sure how Andrew got only a 1% difference: the math is straightforward, unless the coasting intervals were so short the 30-second rounding was affected.

That likely explains it, because I don't think I chopped more than 1-1.5 min at a time from the file.
 
RChung said:
I suspect that's it: I believe Andy lives in a pretty flat place so he and his wife probably rarely see rides with longish coasting intervals.

A deeper issue (that I usually try to ignore) is that the entire implementation of TSB depends on the additivity of TSS.
do you mean intra-ride or inter-ride additivity of TSS, Robert?
 
A deeper issue (that I usually try to ignore) is that the entire implementation of TSB depends on the additivity of TSS.
do you mean intra-ride or inter-ride additivity of TSS, Robert?
You trying to get me not to ignore it?

By how long must segments of a ride be separated before they become additive?
 
RChung said:
http://www.cyclingforums.com/showpost.php?p=3101406&postcount=18
Well, maybe it wasn't designed to be linear across time segments, but that doesn't change the fact that there are ambiguities with this. For example, a 1000 km brevet is (1200km?) presented as an example in Allen & Coggan. How many rides is this? With the F-term approach I presented earlier, there's no ambiguity.... after several hours of rest, you may as well split it; it no longer matters, as F is close to 0. With a linear model without a "state" parameter, it doesn't matter either, obviously.

Dan
 
djconnel said:
Well, maybe it wasn't designed to be linear across time segments, but that doesn't change the fact that there are ambiguities with this. For example, a 1000 km brevet is (1200km?) presented as an example in Allen & Coggan. How many rides is this? With the F-term approach I presented earlier, there's no ambiguity.... after several hours of rest, you may as well split it; it no longer matters, as F is close to 0. With a linear model without a "state" parameter, it doesn't matter either, obviously.
Exactly the reason why I try not to think about it. I have several friends doing Paris-Brest-Paris in August.
 
acoggan said:
Steve Seiler's comments might give you something else to think about:

http://www.mail-archive.com/[email protected]/msg00350.html
My conclusions from this (which I'd actually already read, having found it with google, interestingly) from this:

1. 4th power NP works fairly well, in his opinion, although he makes an error in his "3rd power gives less than AP" statement.... I simulated his intervals (I used 10 reps of 4-on 4-off) and got:

n NP
4 311
3 307
2 303


AP = 300.

2. TSS linear with time makes sense, and hard intervals versus sustained power at threshold can give comparable stresses, but there's some wierd (time-indendent?) results for sub-threshold work at fixed power.

There's nothing in the letter which gives preference to any of the TSS formulas which have been discussed here.
 
RChung said:
Hmmm. I'm trying not to think too deeply about it. However, I did notice this:
The sentence immediately prior is:
"When we measured RPE during two studies employing similar 6 x 4 minute interval sessions, RPE increased linearly with each work bout." He is addressing time-linearity, which ironically is the problem I (and others) have attempted to address with the existing formula

Interestingly, the statement you quoted by him was incorrect: NP is not multiplied by time; it's squared and then multiplied by time. However, doing as he described wouldn't help, either.... it would need to be NP^4 * time to be linear with time.
 
djconnel said:
The sentence immediately prior is:
"When we measured RPE during two studies employing similar 6 x 4 minute interval sessions, RPE increased linearly with each work bout."


And is followed by:

"However, we recently examined autonomic recovery (using Heart rate variability measures) following controlled time exercise bouts performed
under VT1 (1mM lactate, 60% VO2 max), between VT1 and VT2 (3mM lactate, 86% VO2 max) and over VT2 (intervals at 95% VO2 max and 7mM lactate). We found that in these highly trained runners 1) increasing work
duration from 60 to 120 minutes at 60% VO2 max had essentially no impact
on session RPE or the rapidity of autonomic recovery after the bout, but
2) as soon as intensity increased above VT1, autonomic recovery was
delayed significantly but similarly ("identical" time course) after
lactate threshold and hard interval training sessions. THIS data
suggests to me that maybe the time factor used to calculate Total
Training Stress (TTS) should perhaps be different for low intensity
exercise in highly trained athletes. But what should the weighting
factor be? Arbitrary."

...which brings us full-circle to the last 3rd-to-last and 2nd-to-last slides of my presentation to UK Sport: at the present time, IMHO nobody - and I mean nobody - can claim to really know the answer to very simple questions such as "just how long is long?" and "just how hard is hard?".
 
Okay -- that was the wierd time-independent thing I referred to a few posts back. I admit to having problems parsing that part :). He suggests the "time factor used to calculate TTS (ie TSS) for low-intensity exercise in highly trained athletes". Well, sure -- this is why NP isn't multiplied by time, but is instead multiplied by an intensity factor, which is essentially also NP. However, the wierd part is where he says PE and recovery were unchanged by going from 60 to 120 minutes of aerobic exercise. Well, everyone knows that recovery rides are supposed to help, and there's an optimal duration for these, so they represent an obvious limitation on the TSS concept: basically negative ATL (to the degree ATL takes away from "race-readiness", or TSB) from positive TSS. Adding the same TSS by an extra interval the day before would have had a very different effect. I don't think TSS is designed to handle the recovery ride concept. A small error, as recovery rides tend to be small TSS.
 
djconnel said:
[Seiler] is addressing time-linearity
I think the time linearity is a borderline red herring.

Interestingly, the statement you quoted by him was incorrect: NP is not multiplied by time;
Yeah. No one can accuse me of not taking the cheap shot.
 
I'm still trying to work out if this is a serious issue or you're just nit picking. (I can appreciate the "fun" being had);)

Are you saying I'm cheating on TSS since I have regular recovery level riding in my schedule (recovery days and between intense efforts)?:)

Whatever, why is it then that the PMC still works so well?
 
Alex Simmons said:
I'm still trying to work out if this is a serious issue or you're just nit picking. (I can appreciate the "fun" being had);)

Are you saying I'm cheating on TSS since I have regular recovery level riding in my schedule (recovery days and between intense efforts)?:)

Whatever, why is it then that the PMC still works so well?
Andrew referred to a section of Peter's email in which Peter referenced a low-intensity experiment in which perceived exertion was not increased by doubling the effort. My point was simply this is a commonplace occurance -- the whole point of recovery rides is the least possible work isn't the best-possible recovery; sometimes going longer even results in feeling fresher. So I was simply commenting on conclusions from Peter's email: the experiment didn't reveal anything each of us hasn't already experienced. (Cheating on TSS? How about "optimizing" on it? :) )
 
Alex Simmons said:
why is it then that the PMC still works so well?
From the perspective of TSB, it's possible that any monotonic transformation of TSS would work just as well. However, some monotonic transformations may have better behavior over other criteria.
 
RChung said:
From the perspective of TSB, it's possible that any monotonic transformation of TSS would work just as well. However, some monotonic transformations may have better behavior over other criteria.
Consider the following transformation: TSS' = TSS^2/100

Then consider the following workout schedule:
T 100
W 100
R 100
Su 300
total: 600

One week I skip R, but do a bit more on Sunday:
T 100
W 100
Su 400
total: 600

This works if I use TSS, but with TSS', things look MUCH different:

normal week:
T 100
W 100
R 100
Su 900
total: 1200

Then on my modified week:
T 100
R 100
Su 1600
total: 1800

Dan