BikeTechReview.com

  • Increase font size
  • Default font size
  • Decrease font size
Home Forum
Welcome, Guest
Username Password: Remember me

Rick Murphy's Normanalized Power Thread
(1 viewing) (1) Guest

TOPIC: Rick Murphy's Normanalized Power Thread

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24230

  • Kirk
  • OFFLINE
  • Administrator
  • Posts: 1102
  • Karma: 2
djconnel wrote:
It seems to me at some point lactate production has to become proportional to power, as it's a byproduct of processing anaerobic energy, right? Therefore, the following sort of model might make more sense, where P1 == aerobic power, P2 == anaerobic power, and P == total power = P1 + P2:

(rude & crude):
P2/P1 = (P1/P0)^4 (for some P0, which has units of power, assuming such a power model is desired)

lactate production would then be proportional to P2.

For a given P, P1 and P2 can be solved numerically, but it's clear for P << P0, P2 is proportional to P ^4, while for P >> P0, P2 approaches P.

You could use other relationships between P1 and P2, for example exponential, but the key is you need to think about conserving total power.


This is an interesting idea. One issue to consider is that power isn't dependent on lactate (or rather H+) accumulation much below about 20MP (with variability). Essentially, it's either "fill" or "raise the left" if the duration is significantly long enough.

Another issue is cumulative "on" time.

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24233

  • rmur
  • OFFLINE
  • Platinum Boarder
  • Posts: 1088
  • Karma: 0
kraig wrote:
rmur wrote:

I suspect if you looked hard enough that it's out there in the public domain now ... that's about all I can say. Except the bit about the duck of course


Well, then, it should be pretty easy for you to point the BTR readers to the full disclosure of the data and protocol (and your perceptions of the applicability of this data to the task at hand) and let the data speak/let psuedoscience run it's course, eh? So why aren't you doing this? I'm curious to know, honestly, what is holding you back. I mean, you and those that you are heavily invested with aren't actively censoring this information, are they, Rick Murphy?

As for ducks...The last time I asked the duck to speak, it squawked, complained, generated a random number, gave up and then finally said "quack, why does the math equation dole out so many bonus points at lower powers?"...


Kraig Willet: why do you keep pushing that comment in my face? I'm 'invested' with no one.
GIZMO marketing specialist

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24237

  • djconnel
  • OFFLINE
  • Gold Boarder
  • Posts: 203
  • Karma: 2
I assume the reference to "low powers" is to TSS. As I noted, I think TSS is a disaster, because it's non-linear in time in an irrational way. In other words, it isn't a set of integrals over time -- it's the multiplication of two integrals. It's obviously a kludged equation which sort-of works but is so trivial to break it's embarrassing.

The easiest example is comparing three workouts:
1. A 20 minute effort @ FTP
2. Two 10 minute efforts @ FTP, separated by 40 minutes rest (ie two separate workouts in the same day)
3. Two 10 minute efforts @ FTP separated by 40 minutes spinning at 1 watt (ie a single workout).

Neglecting smoothing, the TSS scores are, assuming I didn't make a math error, which I almost always do

1. 33.3
2. 33.3
3. 57.7

Just senseless.

This goes to my point that it's a mistake to apply an NP formula to TSS. Focus on one thing at a time. I have my own TSS formula which does what I want without these anomalies:

? 50/hr * [(P*/FTP) + (P*/FTP)^4] dt

Dan

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24240

  • Kirk
  • OFFLINE
  • Administrator
  • Posts: 1102
  • Karma: 2
Low powers refers to virtual power's behavior at a time when one can also vary power more. Of course, that is then given an additional boost when applied to virtual stress.

Your virtual stress example seems similar to what we've discussed regarding how virtual stress tracks closely with kj's with deviations being driven by soft-pedaling.

Interesting formula of yours. Out of curiosity, where does the ^4 come from?

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24242

  • kraig
  • OFFLINE
  • Administrator
  • Posts: 3285
  • Karma: 4
rmur\n

why do you keep pushing that comment in my face?


because your behavior is consistent with the definition?

So, Rick Murphy, why won't you provide the protocol, data, and your perception of the applicability of the data? In the absence of these things, I reckon I, and those reading on the internets, are left to form their own conclusions - one of which could be that you and those you are heavily invested with are actively censoring the information.
-kraig

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24244

  • djconnel
  • OFFLINE
  • Gold Boarder
  • Posts: 203
  • Karma: 2
Kirk wrote:
Out of curiosity, where does the ^4 come from?


The ^4 comes from Coggan's fit to lactate data of the form lactate = L0 + L1 P ^ 4 described in Coggan & Allen. I assume there is a stress component proportional to lactate levels over time. Valid? I don't know: I see you've been debating the applicability of these fits. But it does kick in a nice superlinearity for extended efforts over FTP. You could make it exponential, as well. Note in this formula the derivative versus power of the time rate of change of training stress at FTP = 5 times the same derivative in the low power limit. So doing that extra watt @ FTP is worth almost 5 times more than doing a watt in Z1. This is basically arbitrary, but seems sort of right to my amateurish judgement

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24248

  • rmur
  • OFFLINE
  • Platinum Boarder
  • Posts: 1088
  • Karma: 0
are actively censoring the information.


You're decrying censorship?
GIZMO marketing specialist

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24252

  • kraig
  • OFFLINE
  • Administrator
  • Posts: 3285
  • Karma: 4
rmur wrote:
are actively censoring the information.


You're decrying censorship?


Well, I think that reply speaks volumes.

There are word replacers in place here, and the word replacer topic was discussed early on in this thread. There is no censorship that takes place here.

Thanks for your input on the topic, Rick Murphy!

<rhetorical question>I wonder, do those involved in doing "real science" have different views on censorship or selective presentation of data?</rhetorical question>
-kraig

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24253

  • rmur
  • OFFLINE
  • Platinum Boarder
  • Posts: 1088
  • Karma: 0
you're most welcome Kraig. I am pleased to be of service.

Now ... perhaps censorship isn't the correct word: it's the derogatory/demeaning manner in which "words" are placed in people's mouths. When I make a "real" post, it gets lost in the clutter ... so I wonder why I bother?

Meanwhile, whilst using those terribly flawed concepts, I'm happily chugging away ...

AFLAC D D
GIZMO marketing specialist

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24257

  • DMC
  • OFFLINE
  • Platinum Boarder
  • Posts: 373
  • Karma: 0
Any chance of re-posting the graph.
That Average power thing you used is deceptive to me.

Could you re graph by Gizmofied Joules per relativity adjusted second?

just messing. sorry

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24258

  • kraig
  • OFFLINE
  • Administrator
  • Posts: 3285
  • Karma: 4
rmur wrote:
<snip>


I would encourage folks to check out the "maximum" thread here on BTR to see how people are making progress with various approaches to training - it's pretty impressive how robust "improvement" is when folks are free of global health issues.

Furthermore, for those that put their faith in the various "top down" developed gizmo power tools, I'm curious to hear feedback on the following plots:

[attachment=2:28g5eotq]<!-- ia2 lower_equals_bonus.png<!-- ia2 [/attachment:28g5eotq]

[attachment=1:28g5eotq]<!-- ia1 doesnt_seem_quite_right_2.png<!-- ia1 [/attachment:28g5eotq]

[attachment=0:28g5eotq]<!-- ia0 new_physiology_2.png<!-- ia0 [/attachment:28g5eotq]

If I'm not being clear with the presentations of my plots, please let me know, and I'll do my best to explain them in more detail.
-kraig

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24259

  • Kirk
  • OFFLINE
  • Administrator
  • Posts: 1102
  • Karma: 2
kraig wrote:
rmur wrote:
<snip>


I would encourage folks to check out the "maximum" thread here on BTR to see how people are making progress with various approaches to training - it's pretty impressive how robust "improvement" is when folks are free of global health issues.

Furthermore, for those that put their faith in the various "top down" developed gizmo power tools, I'm curious to hear feedback on the following plots:

[attachment=2:67aiejpe]<!-- ia2 lower_equals_bonus.png<!-- ia2 [/attachment:67aiejpe]

[attachment=1:67aiejpe]<!-- ia1 doesnt_seem_quite_right_2.png<!-- ia1 [/attachment:67aiejpe]

[attachment=0:67aiejpe]<!-- ia0 new_physiology_2.png<!-- ia0 [/attachment:67aiejpe]

If I'm not being clear with the presentations of my plots, please let me know, and I'll do my best to explain them in more detail.


Those are interesting plots.

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24260

  • Kirk
  • OFFLINE
  • Administrator
  • Posts: 1102
  • Karma: 2
rmur wrote:

Meanwhile, whilst using those terribly flawed concepts, I'm happily chugging away ...

AFLAC D D


Congrats on your recent improvements. Perhaps you would be up for sharing your training ideas in the thread Kraig suggested.

"Terribly flawed concepts" are certainly only your words. I'd say, they have issues and errors which are either not recognized or acknowledged.

My impression, mostly from 3-4 years ago, is that these products are being presented/marketed as a solution to some kind of problem people have in evaluating their training. Essentially, these products are promoted to coaches/athletes as some kind of diagnostic/prognostic tools which will enable some kind of positive outcome. Ya know...something along the general line of "hey, this is the proper, scientific, and most effective way to use your powermeter...and if you don't use it, you are behind the times." ...and marketing works.

The thing is, for that "this is better" theme to be valid beyond the marketing, one first must identify that folks really do have a problem managing their training. IOW, what is the prevalence and magnitude of this problem in what population of power users. Once that is established, then one must identify the diagnostic/prognostic error of these tools. Without recognizing/acknowledging/quantifying those parameters, it's very similar to trying to apply a new serum test for disease X without sensitivity, specificity, or prevalence data. WIthout that package, one cannot make evidence-based diagnostic/prognostic claims about such a tool. It is therefore not a tool (I think your choice to use the term concept is a good one).

After it is established that something is an evidence-based diagnostic/prognostic tool, it then has to be determined if it actually provides new information which preferentially enables a superior outcome compared other methods. Without this entire package, it is not an evidence-based entity or advantageous. This does not mean that using the concepts will result in inferior results though (there are many different ways to the same performance), it's just literally not evidence-based. Heck, the reality is that even if one believes in the GP related stuff, it doesn't matter whether one plugs kj's in the manager thing or gizmo score anyways.

Over my years of experience, I've developed the opinions that training is "robust" in terms of overall philosophy (basically, what Kraig said)...that there can be many different paths to the same performance...and after that, what really matters is what one actually does in training based on one's individual limitations and in response to the unknown things in life outside of training.

I appreciate being corrected and hearing other opinions if folks have thoughts.

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24261

  • djconnel
  • OFFLINE
  • Gold Boarder
  • Posts: 203
  • Karma: 2
Kirk wrote:
This does not mean that using the concepts will result in inferior results


Although it may. Suppose I read Coggan and Allen and decide, hey, this is hot stuff. I'm going to use it to prepare for championships. According to the book, T-S-B is an indicator of readiness. So I decide to guide my training to maximize T-S-B.

T-S-B = C-T-L - A-T-L.

C-T-L = sum of T-S-S exponentially amortized with a 42 day time constant (by default)
A-T-L = sum of T-S-S exponentially amortized with a 7 day time constant (by default)

So what's the response over time of an impulse of training at time 0? Basically:

delta T-S-B = exp(-t / 42 days) / 42 days - exp(-t / 7 days) / 7 days

I want all my training to contribute positively to T-S-B. So I don't want to train too close to the event date. How long should I recover before the event?

delta T-S-B = exp(-t / 7 days) / 7 days - exp(-t / 42 days) / 42 days = 0
exp(-t / 42 days) = 6 exp(-t / 7 days)
-t / 42 days = ln 6 - (t / 7 days)
t (1 / 7 days - 1 / 42 days) = ln 6
t = 15.05 days

The result? I should train as hard as possible up to 15.05 days before my event, then HARD REST for the remaining 15 days. Any additional work I do in those 15 days will hurt my T-S-B at the event.

This doesn't suggest these aren't useful tools. But they're clearly over-simplified, and even at a given level of simplicity, not necessarily optimized.

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24263

  • rmur
  • OFFLINE
  • Platinum Boarder
  • Posts: 1088
  • Karma: 0
djconnel wrote:
Kirk wrote:
This does not mean that using the concepts will result in inferior results


Although it may. Suppose I read Coggan and Allen and decide, hey, this is hot stuff. I'm going to use it to prepare for championships. According to the book, G-S-B is an indicator of readiness. So I decide to guide my training to maximize G-S-B.

G-S-B = C-T-L - A-T-L.

C-T-L = sum of GSS exponentially amortized with a 42 day time constant (by default)
A-T-L = sum of GSS exponentially amortized with a 7 day time constant (by default)

So what's the response over time of an impulse of training at time 0? Basically:

delta G-S-B = exp(-t / 42 days) / 42 days - exp(-t / 7 days) / 7 days

I want all my training to contribute positively to G-S-B. So I don't want to train too close to the event date. How long should I recover before the event?

delta G-S-B = exp(-t / 7 days) / 7 days - exp(-t / 42 days) / 42 days = 0
exp(-t / 42 days) = 6 exp(-t / 7 days)
-t / 42 days = ln 6 - (t / 7 days)
t (1 / 7 days - 1 / 42 days) = ln 6
t = 15.05 days

The result? I should train as hard as possible up to 15.05 days before my event, then HARD REST for the remaining 15 days. Any additional work I do in those 15 days will hurt my G-S-B at the event.

This doesn't suggest these aren't useful tools. But they're clearly over-simplified, and even at a given level of simplicity, not necessarily optimized.


Aah come on now Dan. Everyone knows that C-T-L and common-sense can't be ignored! Form = fitness + freshness. Depending on the target event, folks need to trade the two off. Nothing novel about that - PMC allows folks to quantify the formerly purely qualitative.

I like 'dem numbers and all but using tsb in isolation is clearly whacked
GIZMO marketing specialist

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24264

  • Kirk
  • OFFLINE
  • Administrator
  • Posts: 1102
  • Karma: 2
rmur wrote:

I like 'dem numbers and all but using GSB in isolation is clearly whacked


Yeah, especially since kj's, gizmo score, and either a positive or negative GSB can all have different meanings/outcomes depending on what one actually did in their training.

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24266

  • rmur
  • OFFLINE
  • Platinum Boarder
  • Posts: 1088
  • Karma: 0
Kirk wrote:
rmur wrote:

I like 'dem numbers and all but using GSB in isolation is clearly whacked


Yeah, especially since kj's, gizmo score, and either a positive or negative GSB can all have different meanings/outcomes depending on what one actually did in their training.


oh I think I agree with that. It's not Big Brother. Just IME, an extremely useful tool ....
Could it be sharper? Probably ... and I'm looking fwd to that ...
GIZMO marketing specialist

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24267

  • Kirk
  • OFFLINE
  • Administrator
  • Posts: 1102
  • Karma: 2
rmur wrote:
Kirk wrote:
rmur wrote:

I like 'dem numbers and all but using GSB in isolation is clearly whacked


Yeah, especially since kj's, gizmo score, and either a positive or negative GSB can all have different meanings/outcomes depending on what one actually did in their training.


oh I think I agree with that. It's not Big Brother. Just IME, an extremely useful tool ....
Could it be sharper? Probably ... and I'm looking fwd to that ...


Yeah, you just might agree. Kind of all over the map eh? Given the wealth of information which already exists from power files, the athlete's/coaches' own intuition, and noting that the actual training itself is only part of the bigger supply picture, I think it would be a challenge to demonstrate that something like GSB provides new information that will independently make a difference even if one had specificity, sensitivity, and relevant prevalence data to describe the associated concepts.

Even if one places value on these concepts, it doesn't seem to matter how a GP cucumber and an AP orange (assuming a variable power effort) compare to a particular maximal power-duration banana when they are plugged into the GSB thing anyways eh?

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24270

  • rmur
  • OFFLINE
  • Platinum Boarder
  • Posts: 1088
  • Karma: 0
more uniformly graded fruit in == better jam out
GIZMO marketing specialist

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24271

  • Kirk
  • OFFLINE
  • Administrator
  • Posts: 1102
  • Karma: 2
rmur wrote:
more uniformly graded fruit in == better jam out


Hey, doesn't everyone selling special jam machines believe their jam machines are needed and are the best? I just go...why the heck do I need a jam machine, let alone one where I can't tell how hot or cold my jam and jars are?

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24272

  • Kirk
  • OFFLINE
  • Administrator
  • Posts: 1102
  • Karma: 2
Kirk wrote:
rmur wrote:
more uniformly graded fruit in == better jam out


Hey, doesn't everyone selling special jam machines believe their jam machines are needed and are the best? I just go...why the heck do I need a jam machine, let alone one where I can't tell how hot or cold my jam and jars are?


Ya know, I still have to inspect the fruit and check the temperature manually anyways eh?

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24276

  • djconnel
  • OFFLINE
  • Gold Boarder
  • Posts: 203
  • Karma: 2
I admit my post maximizing T-S-B was misrepresentative, as I don't believe it's ever recommended someone maximize T-S-B for optimal performance, rather that it be tuned to balance fitness and freshness. Since fitness is measured in C-T-L, this suggests each sort of event has a particular linear combination of C-T-L and A-T-L which should be maximized, where the weighting on C-T-L is positive, the weighting on A-T-L is negative, and the magnitude of the C-T-L coefficient is greater than that of the A-T-L coefficient. But I've already gotten myself into enough trouble with these idle mathematical speculations lately, as tempting as they may be.

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24281

  • Ron Ruff
  • OFFLINE
  • Platinum Boarder
  • Posts: 894
  • Karma: 0
kraig wrote:
Yes, I agree - the variable duty cycle exercise is another good way to demonstrate the non-physiological basis of the math equation - especially when placed in the context of really long efforts.


Or like last weekend's 52 mile ride. I've done this loop a bunch of times as a "TT". Best AP was 205W. On Sat I rode with a couple of other guys and there were a lot surges and attacks, drafting, etc. Lots of easy riding too. AP was 187W... but GP was 235W! Guess I just got a lot stronger all of a sudden.

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24282

  • Ron Ruff
  • OFFLINE
  • Platinum Boarder
  • Posts: 894
  • Karma: 0
Kirk wrote:
After it is established that something is an evidence-based diagnostic/prognostic tool, it then has to be determined if it actually provides new information which preferentially enables a superior outcome compared other methods.


The thing that bugs me after I get into this a little more, is that there is the assumption that the best we can hope for a crude curve-fit that matches a particular set of data... and then we extrapolate that to regions that are obviously different, then make up another factor to take a wag at stress... and call it good. The idea of adjusting the weighting factors, exponents, whatever seems equally silly. Why not try to model the actual physiology?

I don't know anything about physiology, but I'm certain I could do a little research and come up with a model that would do a better job of fitting reality than GP and GSS. At various levels of effort the body will use different fuel sources, which will be replenished at different rates, waste products will accumulate and be cleared, muscles will break down and be repaired, etc. The model would keep a running status of these based on what the athlete is doing and has been doing, and predict how much he has left in the tank.

For a universal model you could use the rider's power duration curves to estimate what processes are occuring at a given power level. You can "track" as many parameters as you like... anything that you'd expect to be important and can be at least estimated. Once you've decided on the basic model, there are plenty of power files around that you can use to refine it. Processes with a long time constant will carry over to later days (training stress), so that is a part of it as well. Modifiers based on things like "how do you feel?" would be easy enough to include as well.

Re: Rick Murphy's Normanalized Power Thread 9 years ago #24283

  • djconnel
  • OFFLINE
  • Gold Boarder
  • Posts: 203
  • Karma: 2
I developed a model like this, but didn't have the time to pursue it. It was based on the 3-reservoir model, but instead of hard on-off valves between the reservoirs, it uses nonlinear functions to govern the relative flows. To calibrate the reservoir capacities, I'd run the model against power data, and try to have reservoirs essentially deplete at points in the ride where I ran out of gas.

It was all hackery. It would have been fun to try to go further with it. I could send you my notes if you want.
Time to create page: 0.90 seconds

Poll

Which type of tire is more aerodynamic?