Ride Time by Training Level and NP

RapDaddyo said:

Some of you may know that I have built an application for parsing a ride file by training level (e.g., Andy Coggan's L1-L7). This is a rather complex app and I'm not looking to go into overload but I do have a question for those of you who might eventually want to break down your rides this way.

An obvious issue is whether to use AP or NP to categorize a high-intensity effort (e.g., L4-L7). Presently, this is defined by level. IOW, if you want to use AP for all efforts, you can. If you want to use NP for all efforts, you can (with the exception of L7 because the L7 durations are too short). Or, if you want to use AP for some efforts (e.g., L5-L7) and NP for other efforts (e.g., L4), you can do that too. But, another approach (much more complex) would be to use a duration cut point rather than by level. IOW, let's say you want to categorize efforts >= 20mins by NP and all other efforts by AP. Would you want to do it that way? I hope not, because it will be a pain in the ass (don't ask why), but I thought I would pose the question before I ship this little sucker off to Andy to try and break it (I have little doubt that he can break it or at least "trick it"😉. Then I'll fix it and we'll see what happens next.

RDO,

For training, I would want to see it with AP.

For racing, I would want to see it with NP for anything >=5 minutes, though it really wouldn't matter, since I dissect race files pretty thoroughly...

Jim

otb4evr said:

For training, I would want to see it with AP.

Easy.

otb4evr said:

For racing, I would want to see it with NP for anything >=5 minutes, though it really wouldn't matter, since I dissect race files pretty thoroughly...

At the moment, it's parsed by level. I'll look into parsing by duration. When you think about it, it actually gets pretty hilarious to parse it by NP or AP as a function of duration. Let's say that you have an effort that drops below the AP lower boundary at 4mins, yet it remains above the NP lower boundary until 5mins. So, what is it? It's a 4min effort based on AP, which is the controlling criterion at 4mins. It's a 5min effort based on NP, which is the controlling criterion at 5mins.😕

RapDaddyo said:

Easy.

At the moment, it's parsed by level. I'll look into parsing by duration. When you think about it, it actually gets pretty hilarious to parse it by NP or AP as a function of duration. Let's say that you have an effort that drops below the AP lower boundary at 4mins, yet it remains above the NP lower boundary until 5mins. So, what is it? It's a 4min effort based on AP, which is the controlling criterion at 4mins. It's a 5min effort based on NP, which is the controlling criterion at 5mins.😕

Let's take the hypothetical person that puts out, oh, let's say, 320 watts at FTP. They are in a pretty hot race from which they were dropped.

I am parsing the file after the race to see what is up...

I see an interesting point during the file where their 4 minute power average is just under their CP for that amount of time. Immediately following that "interval", we have a very brief lull and then an immediate attack. At which time the athlete was dropped.

If you look at a situation like this, the NP for the 5 minutes would be above CP, whereas the AP would probably be lower.

We could argue at what would prevent the drop next time, but my thought would be to work on the VO2Max area to allow the athlete to be able to live in this zone a little more.

Like I said, I would love to be able to see this data the way that I would like, but I would find the issue anyway.

When all is said and done, of course, 4 min and 5 min are using the same energy systems, so it probably doesn't matter.

While I am rambling, I might as well say that zone 4 and below would be nice to see as NP, whereas above zone 4 would be AP...

:o

Jim

RapDaddyo said:

Some of you may know that I have built an application for parsing a ride file by training level (e.g., Andy Coggan's L1-L7). This is a rather complex app and I'm not looking to go into overload but I do have a question for those of you who might eventually want to break down your rides this way.

An obvious issue is whether to use AP or NP to categorize a high-intensity effort (e.g., L4-L7). Presently, this is defined by level. IOW, if you want to use AP for all efforts, you can. If you want to use NP for all efforts, you can (with the exception of L7 because the L7 durations are too short). Or, if you want to use AP for some efforts (e.g., L5-L7) and NP for other efforts (e.g., L4), you can do that too. But, another approach (much more complex) would be to use a duration cut point rather than by level. IOW, let's say you want to categorize efforts >= 20mins by NP and all other efforts by AP. Would you want to do it that way? I hope not, because it will be a pain in the ass (don't ask why), but I thought I would pose the question before I ship this little sucker off to Andy to try and break it (I have little doubt that he can break it or at least "trick it"😉. Then I'll fix it and we'll see what happens next.

I think that by level would be sufficient as we mostly "judge" our rides this way (or at least I do) and from a coaches perspective it would be easier to explain and use with our athletes (it always helos if they can relate to and understand a tool and I have only just finished explaining match app and then you hit us with another one 🙂 )

peterwright said:

I think that by level would be sufficient as we mostly "judge" our rides this way (or at least I do) and from a coaches perspective it would be easier to explain and use with our athletes (it always helos if they can relate to and understand a tool and I have only just finished explaining match app and then you hit us with another one 🙂 )

Hey, you can't stop progress.😄 What do you mean "by level would be sufficient ..."? The app breaks down a ride file by level (in consideration of both power and duration), that's its purpose. My question relates to the use of AP or NP for assigning the efforts. It has some user control built in at present, in the form of a schema table that the athlete (or coach) defines that has the training schema. This table currently provides for a choice of AP or NP as the criterion for assignment of efforts. What it doesn't have is a duration cut point for assignment of efforts (e.g., <5mins use AP, >=5mins use NP). Does it matter that it doesn't have this feature?

RapDaddyo said:

Hey, you can't stop progress.😄 What do you mean "by level would be sufficient ..."? The app breaks down a ride file by level (in consideration of both power and duration), that's its purpose. My question relates to the use of AP or NP for assigning the efforts. It has some user control built in at present, in the form of a schema table that the athlete (or coach) defines that has the training schema. This table currently provides for a choice of AP or NP as the criterion for assignment of efforts. What it doesn't have is a duration cut point for assignment of efforts (e.g., <5mins use AP, >=5mins use NP). Does it matter that it doesn't have this feature?

It's been a hard day.(got to pay attention on this forum or you get f..kd up by RD)

I mean... NP/AP chice by level is fine and no need for duration cut point.. (please don't hit me any more... you tapering ?)

otb4evr said:

While I am rambling, I might as well say that zone 4 and below would be nice to see as NP, whereas above zone 4 would be AP...

Well, the AP/NP choice by level is already done. That's the way it works now. But, what the app does not now do is to allow you to assign L4 efforts <20mins by AP and to assign L4 efforts >=20mins by NP (as an example). At the moment, you would have to "force" the split by subdividing L4 into two parts (e.g., 10-20min durations at AP and 20+min durations at NP).

peterwright said:

I mean... NP/AP chice by level is fine and no need for duration cut point.. (please don't hit me any more... you tapering ?)

Okay, got it.😎 And, nah, I'm done with tapering (maybe forever). It makes me too cranky.

RapDaddyo said:

Well, the AP/NP choice by level is already done. That's the way it works now. But, what the app does not now do is to allow you to assign L4 efforts <20mins by AP and to assign L4 efforts >=20mins by NP (as an example). At the moment, you would have to "force" the split by subdividing L4 into two parts (e.g., 10-20min durations at AP and 20+min durations at NP).

Gee, I guess it will have to do...

😄

Seriously, though...I can't wait.

Thanks for your efforts into helping us understand another little part of ourselves...

Jim

otb4evr said:

Gee, I guess it will have to do...

😄

Seriously, though...I can't wait.

Thanks for your efforts into helping us understand another little part of ourselves...

Jim

Well, it's basically done. I'm just adding a few final touches (you know, the ones that break it 😄 ) and documenting it (that's the hard part) before shipping it off to Mecca. Then we'll all await for it to be blesssed.

Oh, one more thing. I'm computing TSS by level by ride. I found this really helpful since some of the key levels are so broad in power terms (e.g., L4). I have moved from minutes by level to TSS by level to summarize my rides from an adaptation perspective. It's pretty neat data.

RapDaddyo said:

The app breaks down a ride file by level (in consideration of both power and duration), that's its purpose.

I'm having a hard time understanding the issue here, since I thought you were looking at parsing the file based on time in level which meets or exceeds the minimum useful time in level for training stimulus. If that's the case wouldn't the time factor be pretty important in your analysis?

One of the things about NP is that the 30-sec averaging is based on the *half-life* of the metabolic processes, rather than the complete metabolic response time (ie, 5 or more half-lives). IOW, if power is high for only 30 seconds, the NP model calculates 4th order factoring for the whole period, but the metabolic response only has time to build up to half of that level by the end of the period. The longer each excursion is, the more closely NP models the actual metabolic response, while the shorter the duration (>30sec), the easier it would be to trick NP into overstating the output. Ideally, you would use AP when the peaks and valleys are less than a couple minutes each, and NP when the excursions are longer. But, it's still probably fine to use AP or NP to categorize a particular period of time into a level. I'd probably use NP for L4 and below, and AP for L5 and above, with the understanding that L2-L4 are really only dealing with periods longer than about 10 continuous minutes. The shorter-term periods are going to evaluated against the L5-L7 ranges, and that's where the bursts are likely to be shorter and NP is most likely to overstate the physical demands of the effort.

Short answer: I'd use NP for L4 and below - AP for L5 and above.

frenchyge said:

I'm having a hard time understanding the issue here, since I thought you were looking at parsing the file based on time in level which meets or exceeds the minimum useful time in level for training stimulus.

You are correct in your understanding of what the app does. It breaks down a ride file by time in level. For L3 and below, there is only a power condition, so that is nothing more than a frequency distribution of the ride file observations. For L4 and above, there are two conditions, power and duration. So, an effort is categorized in one of these levels only if it meets both conditions. There is no problem with using either AP or NP for all efforts in a given level (e.g., NP for L4 and below, AP for L5 and above), because there are never contradictory results. But, when you have a duration cut-point (e.g., 20mins as Andy has expressed on a few occasions) for using AP or NP, you can get contradictory results. For example, you ride a variable power L4 effort in which NP remains slightly above 91%FTP for 21 minutes. But, AP remained above 91%FTP for only 15mins, then it dropped below and remained below 91%FTP for the remainder of the effort. Based on a 20min cut-point for use of AP or NP, the AP test says that it is a 15min effort yet the NP test says that it is a 21min effort. What is it?

frenchyge said:

I'm having a hard time understanding the issue here, since I thought you were looking at parsing the file based on time in level which meets or exceeds the minimum useful time in level for training stimulus. If that's the case wouldn't the time factor be pretty important in your analysis?

One of the things about NP is that the 30-sec averaging is based on the *half-life* of the metabolic processes, rather than the complete metabolic response time (ie, 5 or more half-lives). IOW, if power is high for only 30 seconds, the NP model calculates 4th order factoring for the whole period, but the metabolic response only has time to build up to half of that level by the end of the period. The longer each excursion is, the more closely NP models the actual metabolic response, while the shorter the duration (>30sec), the easier it would be to trick NP into overstating the output. Ideally, you would use AP when the peaks and valleys are less than a couple minutes each, and NP when the excursions are longer. But, it's still probably fine to use AP or NP to categorize a particular period of time into a level. I'd probably use NP for L4 and below, and AP for L5 and above, with the understanding that L2-L4 are really only dealing with periods longer than about 10 continuous minutes. The shorter-term periods are going to evaluated against the L5-L7 ranges, and that's where the bursts are likely to be shorter and NP is most likely to overstate the physical demands of the effort.

Short answer: I'd use NP for L4 and below - AP for L5 and above.

If AP and NP are significantly different isn't it safe to say that you didn't spend a lot of uninterrupted time actually riding at the power output of that zone. It's pretty hard to get a high VI (particularly in the lower zones) unless you spend significant time in other zones. Plus there are many different ways that one can get the same NP for effort with different adaptations. Although I think the latter is accounted for.

To be honest Im confused with how it works. So the above might not be significant issues.

gvanwagner said:

If AP and NP are significantly different isn't it safe to say that you didn't spend a lot of uninterrupted time actually riding at the power output of that zone. It's pretty hard to get a high VI (particularly in the lower zones) unless you spend significant time in other zones. Plus there are many different ways that one can get the same NP for effort with different adaptations. Although I think the latter is accounted for.

To be honest Im confused with how it works. So the above might not be significant issues.

Oh, I just though of something- if it uses rolling AP or rolling NP that would smooth out the line to tease out the adaptation a little better and also would make deviation less so the longer the effort go on. Assuming it goes back a bit then it would also say that 2x20 with 5 min would be equal to 40min because rolling AP/NP hasn't dropped enough.But wait that brings up other problems in other situations. Now really confused.

gvanwagner said:

If AP and NP are significantly different isn't it safe to say that you didn't spend a lot of uninterrupted time actually riding at the power output of that zone.

Ummm... that's a lot of negatives, so I'm not quite sure if I agree or disagree. 🙂 A high VI definitely means that the ride was very variable in magnitude, but doesn't necessarily address the duration of each variance (as long as they are >30 sec). I think the accuracy of the NP model depends more on the duration of each variance, rather than the number of occurrences or the magnitude of each.

gvanwagner said:

If AP and NP are significantly different isn't it safe to say that you didn't spend a lot of uninterrupted time actually riding at the power output of that zone.

But, NP reduces a variable power ride to its constant power equivalent from a physiological stress perspective. So, categorizing an effort as L4 even though a significant percentage of the total effort was at an L3 power is perfectly valid because the purpose is to classify the ride time by level for adaptation purposes.

RapDaddyo said:

But, when you have a duration cut-point (e.g., 20mins as Andy has expressed on a few occasions) for using AP or NP, you can get contradictory results. For example, you ride a variable power L4 effort in which NP remains slightly above 91%FTP for 21 minutes. But, AP remained above 91%FTP for only 15mins, then it dropped below and remained below 91%FTP for the remainder of the effort. Based on a 20min cut-point for use of AP or NP, the AP test says that it is a 15min effort yet the NP test says that it is a 21min effort. What is it?

Yeah, I understand the issue there, and I'd be okay using AP or NP based on level rather than duration. Have you worked out the 'intervals within intervals' issue?

Here's a question. When WKO+ changed the minimum NP duration to 5 minutes, did the NP calculation then change to a 5-min rolling power average, or is it still at 30-sec? If it now uses the longer averaging, then I might be more tempted to use NP for L5 as well (but not L6/L7, since the durations would be too short).

frenchyge said:

Ummm... that's a lot of negatives, so I'm not quite sure if I agree or disagree. 🙂 A high VI definitely means that the ride was very variable in magnitude, but doesn't necessarily address the duration of each variance (as long as they are >30 sec). I think the accuracy of the NP model depends more on the duration of each variance, rather than the number of occurrences or the magnitude of each.

Well, if you spend long enough durations over/under total NP while keeping in the same level then NP will *always* end up in that level. In practice though it's hard to bust out a high VI (so as to make the decision for NP or AP affect the outcome) without spending significant time away in different zones.

frenchyge said:

Yeah, I understand the issue there, and I'd be okay using AP or NP based on level rather than duration. Have you worked out the 'intervals within intervals' issue?

Yes.😄 That was one of the major problems. Actually, my solution is quite elegant (if I do say so myself) in that I give you the option of seeing efforts either "raw" or "net." IOW, if you do a 20min L4 with 2x1min L6 efforts in the middle of the L4 effort, you can view the L4 effort isolated (20mins) or net of higher efforts (18mins). Which raises an interesting question for Andy. What are the adaptation implications of "stacked" high-intensity efforts such as described above.

frenchyge said:

Here's a question. When WKO+ changed the minimum NP duration to 5 minutes, did the NP calculation then change to a 5-min rolling power average, or is it still at 30-sec? If it now uses the longer averaging, then I might be more tempted to use NP for L5 as well (but not L6/L7, since the durations would be too short).

NP is still computed based on 30s rolling average, they just don't show it to you until total duration is 5mins (i.e., they "hide" it).