Wow. This generated a lot of good discussion. Anyway, I'd say this was the post that I'd most wished I'd summarized in my own original post.

In economic terms, people respond to incentives, and if you express the idea that CTL is basically proxy for "fitness," then obvious people will seek to maximize CTL.

I would say that my objection to this is really confined to triathlon. I think the concept of, especially, sTSS (swim TSS) is incredibly malformed. rTss (run TSS) is also pretty bad, though it's more/less bad depending on terrain. So in that sense, I would say that I do actually say that TrainingPeaks does, to a certain extent, guide people towards training in a certain way, and one that is not particularly well grounded in reality WHEN IT COMES TO TRIATHLON.

In that sense, I think the extension of the concept of TSS away from a strict powermeter-based concept was a mistake. And perhaps a disservice to athletes...

That said, I found the graph to be interesting. I found the discussion it generated to be even more interesting. As always, I appreciate Dr. Coggan chiming in. I have seen tremendous value in many of Dr. Coggan's cycling-related metrics. The extrapolation of those metrics - or, really, of the concept of a numerical TSS value based around a "threshold pace" value - to swimming and running, in practical every day terms, is something that I disagree with.

Anyway, did not expect to see almost 50 replies at the end of the day. Thanks to many of you for also giving me some good stuff to think about.

To continue to discuss, one thing that I've been toying with has been ignoring CTL and instead looking solely at the slope of CTL over given periods of time. The idea being that your body responds to change. So the slope of the CTL curve matters more than its absolute value. This was/is reinforced by the paper modeling injury risk as a function of ATL/CTL. A ratio of ATL/CTL of >1.5 seems to dramatically increase risk of injury. http://bjsm.bmj.com/...rts-2015-095788.full

What's interesting is that Paulo has been able to replicate these findings with his athletes, only Paulo uses longitudinal RPE tracking (with a separate constant for each sport). So he tracks A*RPE*duration where A is a constant that varies by sport. And that gives him a training load score for that workout. He then tracks that over time, using some basic math - similar, I believe, to Dr. Coggan - to represent ATL as roughly 7-day load and CTL to be 6 weeks plus.

I'd say that looking for a slight - but positive - slope to your CTL curve is more important than the actual value of CTL. I think that's the sort of thing that TrainingPeaks could offer more of and that might encourage more intelligent training.

I haven't yet played with it, but given that I've found value in the derivative of CTL - slope, I'm not curious if you could look at "fitness" (loaded term) by examining the area under the CTL curve - the integral of CTL - over varying periods of time.

---

"Non est ad astra mollis e terris via." - Seneca | rappstar.com | FB - Rappstar Racing | IG - @jordanrapp

I guess you get points for diversity in your twitter feed for following a self-promoter high performance guru like Mike Young. Filliol must be proud.
But get give credit where it belongs, the graphic was by Derek Hansen of strengthpowerspeed.com, a Charlie Francis protege. Easily visible in the bottom right hand corner.

I may get flamed for this, especially by Dr. Coggan, but I think looking at the slope of CTL invites the same type of misuse of these metrics described by jbank with respect to absolute CTL. The 'idea' behind your hypothesis is that the body responds to change, but it doesn't necessarily respond to volume changes (as quantified by TSS-based metrics) on a relative basis either. It is entirely plausible--although I don't necessarily expect Dr. Coggan to agree with me on this extreme case--for an athlete to spend 6 months increasing CTL, increasing CTL slope, and actually getting no faster at all (as measured by threshold). This function would look something like this:



The "no faster at all" bit is the blue line, the athlete's FT. Your underlying "idea"--that the body responds to change--is a good one. But what type of change? CTL change? No, not necessarily.

Everyone has said that these are valuable tools to help manage training, and of course they are...we can thank Dr. Coggan for that. But my ongoing challenge to him is basically as follows:

If we agree that there are covariates here that are extremely important, what are they and how important are they? Maybe this exists somewhere and has been done and I just haven't really seen it, but what sort of impact would there be in this equation if we incorporated training mix composition correlates? Could we evaluate the efficacy of these metrics with respect to FTP improvement, irrespective of CTL, or CTL change? Aren't there perhaps variables/metrics that are just as vital, or more vital in some cases, to the ultimate end goal, which is increasing functional threshold? Can the science/math figure out a way to better incorporate them into our analysis?

P.S. I am primarily talking about cycling, mainly because I find the TP bastardized metrics of rTSS to be absolute garbage, especially for my training (trails, hills, etc.). Clearly this all becomes much more complicated when you incorporate running into the equation.

As I pointed out to UK Sport when they had me over to talk as part of their build-up to the London Olympics, I don't think that there is a lot to be gained by trying to improve the input function - it is the model structure itself that is the greatest limitation.

Also I have pointed out before, it is the above conclusion that started people on the path that eventually led to WKO4...

I would tend to agree, which is why when I introduced the PMC (http://home.trainingpeaks.com/...-performance-manager) I wrote things like:

"CTL is a relative indicator of changes in performance ability due to changes in fitness, not an absolute predictor"

"within the logical constructs of the Performance Manager, performance depends not only on TSB, but also on CTL (in keeping with saying that “form equals fitness plus freshness”). The “art” in applying the Performance Manager therefore lies in determining the precise combination of TSB and CTL that results in maximum performance. "

"in the Performance Manager concept, an individual’s CTL (and the “composition” of the training resulting in that CTL – see more below) determines their performance potential (at least within limits), but their TSB influences their ability to fully express that potential."


Despite some people seeming to have suddenly discovered the idea, the notion that the rate of change in training load is indicative of both training risk and training reward has been around as long as the PMC has...longer, in fact (e.g., the 10% rule for runners).

I also find it amusing how much attention Gabbett has received for simply "dumbing down" my already "dumbed down" (by necessity) version of Banister's impulse-response model. Kudos to him, however, for applying the concepts of CTL, ATL, and ramp rate to other sports.


As I said above, there is really nothing new in this - in particular, Carl Foster first came up with "session RPE" as a simplified way of scoring training, and I have repeatedly suggested that people may wish to apply it to triathlon or team sports, e.g., soccer.

Perhaps some day I'll have a job where I can justify spending time working on such things (hint, hint). Until then, I will just have to continue to contemplate them as a hobby. (Note that I first pointed out the distinction between modeling stress and modeling adaptation on the wattage list almost 10 y ago. At the time I also pointed out what shape I thought an aerobic training adaptation score should have.)

Just an FYI: https://www.ncbi.nlm.nih.gov/pubmed/19910822

I haven't read the full study, and would probably not understand half of it. I would - however - add that from my perspective (probably as a typicak WKO/TP-user: amateur self-coached) the problem with PCM in a triathlon-perspective is that the rTSS-metric rareley hits the mark estimating the real stress of a run-workout. Workouts on the bike measured with a PM are spot on - the NP from a ride with my PM correlates very well with the actual stress from the workout in question. For cycling the TSS/PCM is an easy-to-use tool (and quite understandable, if aware of the limitations that have been underlined numerous times above).

When it comes to running, my subjective experience is this:

1.
The NGP-metric is nowhere near as usefull/accurate as NP. In other words - when TP is trying to account for terrain etc affecting how stressfull a run is, it often misses the mark. The problem increases the more difficult the terrain is, and when f.ex. running mountain/trails the metric can be disregarded completly. F.ex when running a mountainous trail my HR can go through the roof and I do a super-hard 1.5hr run. NGP from the session is - however - still very low. I guess this is mostly because the rTSS - metric is not designed to monitor this type of runs. If there existed a PM for running that was as scientifically exact as PM's for running, I guess this problem could be eliminated (but lets not steer this thread over to a Stryd-debate!)

2.
I guess if purely running flat/track, the rTSS is an accurate/usefull metric. If only running this kind of terrain, I feel the PCM/TSS to some extent gives a good reflection of ATL vs CTL/fatigue. However, where I live, its hard to come by a route that would give me anything less than 200+metres of elevation gain/loss pr 10k of running. This often skews NGP to much, IMO. To be precise - i feel the NPG-metric hits the mark much better going uphill than downhill. The strain of going downhill is not that much less than flat - especiallyif it gets too steep downhill (an extreme example to make my point: most my runs start with a descent going down at 10% grade. When running down this at 4 min/km pace, TP works this out to equal NGP 10 min/km. I'd argue that does not quite hit the mark)

3.
I still find it difficult to combine cycling and running TSS in one PCM, and get usefull information out of the combined CTL/ATL-numbers. This is the part where it gets to scientific for my level of understanding, but I feel that some parts of the fatigue is "sport-specific" and some fatigue carries over from running to cycling or vica-versa. My subjective feeling of fatigue across sports is also grossly affected by my fitness in that particular sport (not surprising!). IOW - if I have laid of cycling for a month, starting to train both cycling and running again will leave me more fatigued than if my body is used to the "combined" load of two sports. The fists cycling-session will "hurt" my later running sessions alot more than if I had better cycling-fitness to begin with. This applies even if I account for the fact that I have to adjust f.ex. my cycling FTP as a result of lost cycling fitness during the off-period (iow - the increased fatigue is not a result of me trying to hold training levels that are out of date due to lost fitness).

In short my take-away from this is that the PCM is a great tool for cycling, and probably also for running if only running flat (if running hilly terrain it would be better to substitute the rTSS with a number based on HR/percieved exertion).

For the two sports combined, I'm unsure of how usefull the numbers are. In any case I would argue one needs alot deeper insight into training/physiology etc to get the same use of PCM for running and cycling combined, than just for one sport. I should also add - I use WKO/PCM and am happy with it. Its nice for tracking training-load in the longer term (i.e. how much did i train this winter vs last winter).

I'll leave the swimming sTSS to someone else. I dont even float!

This was my thought as well, and when it comes to training stress and response the funny thing is that it is both simple and complex. The simple aspect is that your body is remarkable at adapting to what you ask of it. Want to more performance doing x (where this is something like sprinting, pursuit, TT or endurance), then do more x and be mindful of recovery and diet. It is like magic and is just plain works, until it doesn't. Everyone who has been doing this a while knows the feeling of the plateau, so in order to break through you need to create more intense training stress and mind recovery even more.

If there is any lesson about the chart it would be that time in the middle zone won't along get you to your peak ability. I take issue with the chart implying that methods 1 & 3 yield infinite and ever upward progress because they all contribute to fitness. In my observation the first chart represents what fit people need to do to keep making gains.

While I respect your knowledge on this sport I don't see how this chart can be used for anyone that is not training to their maximum potential or ability. Recognizing fatigue is individual, this chart just puts down an approach to training but doesn't say how to know if fatigue is overlapping.

I see the TSS score as being a way to quantifying training stress but that doesn't mean that higher or a steeper ramp rate is better for any one person. Blaming someone's failures on their limited understanding of TSS is like blaming 7 for being greater than 6.

So the question is, why are you attempting to do something that I have repeatedly warned people against? :)

I don't blame 7 for being greater than 6, but I do understand why 6 is afraid of 7. ;)

I noticed that triathletes are now chasing higher CTL numbers like chasing higher run mileage for a pure runner. This leads to burn out, injury, etc. A good example is that I've seen athletes run on treadmill at 0% slope instead of the road or higher slope to get a higher rTSS. The only real way to test fitness for swimming/running is to test. If after two or three tests, you should be getting faster if you are training right.

I was wondering if someone was going to bite on that :)

Yes, I do think some people game the numbers and doing that on a treadmill is a way to do it.

Yes, I feel testing on the road (or Pool) is a very good way to see if you are improving.

Personally I've changed my training this season because I hadn't seen much improvement last year. Now it could be that I'm fairly close to my limit or that the stress was too much for me, I'm not getting any younger and it appears I'm starting to need more recovery. Since I've gone back to self coaching I haven't given myself the ups and downs like I did when I was with Mike P. Right now I'm looking for the best results, if that means a lower CTS then that's what I'll do.

I saw what you did there.

Thanks. I should have just cut-n-pasted the contents of Mike's tweet since he credited Derek in it. Apologies.

Joel follows lots of people he disagrees with or otherwise doesn't necessarily see eye-to-eye with. what matters, and we've had this discussion, is can you learn something. A lot of what Mike posts I disregard. But he does also post things that I have found interesting. Like this.

---

"Non est ad astra mollis e terris via." - Seneca | rappstar.com | FB - Rappstar Racing | IG - @jordanrapp

 

TrainingPeaks does like to default to showing multiple sports in their PMC chart. I have to change that back regularly on the website. :/

nice graph thanks for sharing.
since you talk about paulo, i think in this podacst they talked aobut that, he said that for running he preferes an more sweatspot approach.

I am sure Jordan will speak for himself here, however I will just sum up what I took from his post. I agree with his assessment to a great degree.
Your TSS/CTL model serves single sport, cycling very well. I agree that carrying your concept into swimming and running clearly skews both metrics to the low side of things. I am not going to argue here why I think that, I can start with swim TSS being bogus entirely, I can say that NGP is inaccurate at best as it never accounts for winds, ie, pace is not a metric like power that does not depend on winds..........let alone GPS watches that do not have baro altimeter input......so elevation gains are inaccurate (yes I know about the normalizing that with a button in TP)....
Further, your TSS model is skewed far more toward the duration than it is to intensity, ie. short course triathlon training has very little value with that model. That is your explanation of the algorithm used in calculating TSS.....
On top of all of that, triathletes have no capacity to test all three sports T-pace, FTP and Threshold pace respectively to keep the TSS/CTL semi accurate.....Not like cycling and cyclist that can use all the methods you describe to arrive at FTP estimates. Most athletes rarely do all out efforts for 20min or longer when IM training at 20-24hrs per week in our amateur world. Not enough time to recover.
Yeah I know everybody is using some kind of estimate, but your model depends on accuracy of benchmarks. We all know that FTP changes daily based on your recovery.....
There you have it.

I think this is a nice example of why I said I think there's a "TrainingPeaks/TSS model." Dr. Coggan explicitly says he doesn't recommend combining these numbers, and yet the primary default graph on the website and on the mobile app is exactly that. And on the mobile app, there's no way to change it. And the app, each day, displays combined CTL/ATL/TSB. And you cannot turn that off.

If you are going - as TrainingPeaks has done - to dictate that CTL/ATL/TSB, as calculated - and combined - by their metrics, MUST be displayed (at least on the mobile app), then I think that's not exactly being "agnostic" as to how people train...

@pk - Yes. Paulo - and Joel too - both advocate more "sweet spot" training on running. But that's at least in part because the risk of injury goes up substantially with max and supra-max type training when running. But Paulo is also a big fan of hills - as is Joel - which is one way to introduce additional load for a given speed...

---

"Non est ad astra mollis e terris via." - Seneca | rappstar.com | FB - Rappstar Racing | IG - @jordanrapp

Just an FYI: typical ground-level winds have very limited impact on the energy cost of running.

You are conflating two things here, i.e., 1) the tendency of some to chase CTL without regard to training composition, and 2) the desire of many triathletes (served by TrainingPeaks) to add TSS, rTSS, and sTSS to arrive at an overall measure of their training load. It was the former to which I was referring when I said the approach is agnostic (although note that I don't recommend that people do either #1 or #2, and never have).

This formula has been successful

Probably because if the PCM were a hammer, I´d assume I can use it to bang in both nails and screws! :)