Originally Posted by An old Guy
You really have to stay on point here. I made no comments on the idea of impulse-response. I made comments on Dr. Banister's model of impulse-response. His model is defective. The fact that you can list a series of papers is not remarkable. What is remarkable is that you cannot find a paper that points out that Dr. Banister's model make outrageous predictions. That is an important paper. As I said before one cannot prove a theory; one can only disprove it.
I find impulse-response to be a reasonable idea. Dr. Banister's model has problems. Your model of it, TSS, has serious problems.
Let me recap what you have said here and in the past.
You have said: TSS is best thought of as a indicator of glycogen depletion. You have also said: not so. But you have not indicated any other "better thought."
The basic idea of TSS is an implementable model of impulse-response. You have no data to show it models anything.
You say that TSS does not predict performance. My understanding is that Dr. Banister's model is good enough to compare proposed training schedules and determine if one is superior to another. True or false does not matter here. The importance here is that one cannot use TSS in that manner.
You say that no amount of proof a single person could amass would disprove TSS (actually you switched words and used NP.) But you keep asking for my numbers. As though you would accept my numbers as a valid disproof.
I have provided you with workouts that shoudl cause you to doubt TSS, but you dismiss them.
You claim TSS is better than any other easily implemented method but you provide no proof. It should be easy to compare TSS averages (CTL; ATL) to something like boxcar averages of daily work or even riding time and show that TSS is better. But since TSS is not predictive, what doe better mean?
I just don't know how to deal with a religious zealot like you.
1. No question, the impulse-response paradigm (which, as applied to modeling the relationship between training and performance, is inseparable from Banister) has some significant limitations, some of which are practical and some of which are related to the structure of the model itself (e.g., the prediction that you should stop training entirely when tapering for an important competition). I discussed many of these in my article; others can be found in the literature (e.g., the Fitz-Clarke paper does a good job of illustrating the implications of the time constants). Nonetheless, it is the most studied model, works rather well in laboratory-based studies, and has significantly advanced our understanding of how the body responds to training even if it has limited use outside the laboratory.
2. You are confusing the input function (e.g., TRIMP, TSS) with the structure of the model (e.g., Banister's impulse-response, my Performance Manager).
3. The reason that people keep asking you to post some data is because your claims are simply unbelievable, something that would be true even if I had never developed the normalized power algorithm. The only way forward in this discussion, then, is for you to prove that you're just making things up.
4. In fact, on the wattage list I've previously compared the use of TSS vs. total work as the input to the Performance Manager, and the former outperformed the latter for the case study I examined. Of course, that is entirely logical, since only TSS and not total work takes into consideration the intensity of training. It would be nice if somebody did a formal scientific study, but that's neither my interest nor my concern.