My rides consist of long endurance rides, say six hours around LTP or long alpine climbs.
The first issues i am running into is that after a 140km ride, my form will stay fresh where in real life i am pretty tired. Xert will suggest a nice hard interval session which will be impossible to complete in that state.
The other issue is that when i do a max effort on a long climb (one hour or more), i almost always ride just below TP and will be completely wasted at the top. The MPA graph of the workout is nice and flat and my form is again not tired, but it was a max effort. In this case, the data is also not matching with the real world.
Am i using Xert the wrong way or is the used model just not working for my style of riding?
I would think that for example XSS would also be a parameter for form and workout suggestions. I have for example two (for me) exactly the same rides, 330xss, 140km, fresh status at start, etc. After the first ride, my form stays fresh, after the second very tired, where the only difference is a small sprint in the second ride.
Hi, yes these points re: long endurance rides, and hard and slightly below threshold efforts are a known challenge within Xert. Basically anything below threshold is assumed to be ‘easy’ and helping you recover, without (fully) taking into account intensity and duration. As you say, that’s not actually the case.
I have heard Xert are working on a solution but not sure of timing.
On this point, are the two rides far apart or near each other? Sprints won’t make you go ‘red’ / very tired unless your overall form gets too far negative. They will make you go ‘yellow’ / tired, since you are putting strain on your high intensity and peak ‘systems’. If you are red it just means your total fatigue level is too high relative to your training load. Here’s a chart that explains it visually
Search this forum for “always tired” or “always fresh” as the two are related.
The model does not currently handle day after day of long slow distance or ultra-distance that rarely exceeds TP and seldom includes max efforts above TP.
As @wescaine mentions Xert is working on an update to address this gap.
For your type of riding you’ll want to use the Freshness Feedback slider when form prediction does not reflect how you feel. Use as often as needed.
I’ve had similar issues with Xert and long distance riding, worst example was after a 600km audax, it had me as fresh within 48hours and was recommending supra-threshold and VO2 style workouts.
There are a couple of issues that I’ve run into with the model, due to doing quite a lot of long distance stuff, my training load can get me up to 3 (sometimes touching 4) stars, this results in XATA suggesting workouts that are too intense for me. To remedy this I use the filter to put a cap on the difficulty of the workouts that XATA suggests.
Secondly, again I suspect due to the training load, I find that my TP in particular gets over elevated, which then increases the issue of being constantly ‘fresh’, to remedy this, I test my power durations and input those into the fitness signature calculator to set my signature, rather than completely 100% trust the system. This seems to result in a better signature for me (others will be different) and my needs, quite often my TP is lower than what a breakthrough gives me, especially if one comes through activities such as 30/15s, so that in turn means my freshness rating and difficulty of my longer rides tends to better reflect how they feel.
Wouldn’t the accumulation XSS (including the ratio of workload on the 3 energy system) influencing the MPA be the answer to this? A sort of MFT Maximum Fatigue Tolerance factor that would influence back stage the MPA value throughout the workout.
High performance athlete/high volume phase of XSS would be less sensitive on their MPA and untrained athlete would be very senstive to it. That would explain why some athlete can produce world-class MPA effort even after many hours of prior efforts.
Fairly sure they won’t post details before it’s released to protect IP and prevent copycats, but interesting to speculate
What’s missing fundamentally is deterioration due to work done below threshold (and agree potentially just deterioration with work done in general, as you suggest). It will likely vary with training load. It will also vary with nutrition and other factors which I think is what makes it more complicated
I also think a differentiation between lower threshold and threshold is needed (beyond just xssr, and the way LTP is used in workouts) in modeling fatigue, both within a ride and over the longer term… perhaps with a revised way of estimating lower threshold or even using / estimating VLamax, given they kind of have the components already - that would be a strong differentiator as you currently need to pay a lot for a metabolic or lactate test, and even that’s not perfect.
I actually wonder whether ‘threshold’ is even needed… there are other models where it’s the result rather than the driver… and more importance is given to the lower threshold.
Just speculation, and do appreciate it’s not at all easy.
this is possibly due to xert no taking into account mental factors ie motivation willing to suffer etc the number that xert gives is theoretical. i suggest you work on some mental techniques to help with this
the video above is great and has some techniques that i have used to great effect
hopefully @ManofSteele can create some development’s for xert to take this into account(long efforts below tp)
I also think that accuratly estimating fatigue in relation with nutrition must be pretty complicated, and in my sense pretty much out of the scope of Xert. It’s sort of in the same category as HRV is for recovery. I think Xert can give great guidelines (like they did recently on YT/Insta on suggesting when to take a Carb Gel when you’re on free ride just by looking at your Carb data field) on all of those aspect that can contribute to the feeling of “fatigue”. But those causes are more contextual fatigue that happens because an athlete habits decision rather than perfoming data, such as MPA.
So I also agree with you @wescaine that this metric should be mesure from acculated XSS under LTP that deteriorates MPA over time.
And Granted, it’s really just fun spectulation I’m still a fan of MPA as it stance right now
Thanks for all the feedback. My subscription just expired so i think i will go the intervals.icu with a workout player route for now. The FTP/TSS model is also not perfect but maybe better for now until Xert has a way to handle this. I will keep checking the forum and be back when possible.
The interesting thing is, is that this is already in place, the difficulty data in the activity graph is pretty accurate.
I appreciate the feedback but this is not the case. As stated in my first post, i do max efforts on long climbs and am completely toast at the top. The issue in my view is that the model does a lot of things above TP, but almost nothing under. In real life this is not such a hard line but a gradual increase of used energy.
Overall your point on mental state is a good one, a survey after a workout or a RPE field for every activity could be very helpful in determining the fatigue level.
Many professional and amateurs work on fueling their efforts so that their fitness parameters (TP/FTP, HIE and PP) don’t erode over longer rides. Using a fixed signature is what causes MPA to stay elevated and predicting the influence of this fatigue on your signature is what would be required for MPA to show a greater decrease over longer rides. As a thought experiment, what would your FTP be after doing a 20 minute FTP test? After 5 hours of riding?
We could reduce your TP/HIE/PP as you perform and thus increase XSS for longer rides and thus assign more fatigue to your rides but a dynamic signature isn’t a simple problem. Having said that, a model for dynamic signature values is in the works with Xert 2.0.
Every other system, for the most part, is trying to pinpoint FTP, generally by a regression or AI on months worth of data. There is really no thought or possibility of adjusting FTP (and therefore TSS) over the course of a longer rides. That’s a long way away. Getting a daily FTP is also beyond their current state.
I would suggest you first think about your fueling and why you’re fading as much as you are during longer rides. This can be trained and improved.
I’ve used various, climber, gc contender and sprint time-triallist. I’ve also tried to use Continuous Improvement, but found it gave me too much intensity (probably down to incorrect signature and the elveated training load from long distance).
I’ve recently decided to give it another go using the signature generated by inputting the results from power duration tests, and it seems to be behaving much more in line with how I feel. I think just loading a load of data into the sytem and letting it do it’s own thing doesn’t seem to work for me, as either TP is too high or starting training load is off. Hopefully, now I seem to have it running okay it will stay that way as when it works it’s really good for me as I don’t work Mon-Fri with weekends off so my training needs to be flexible, which the majority of canned plans simply don’t account for.
As a thought experiment, what would your FTP be after doing a 20 minute FTP test? After 5 hours of riding?
I always wondered if the HR metrics that are interpreted could somehow be used to inform the power data in terms of fatigue, XSS and potential performance. I feel my HR data is fairly representative of my fatigue over longer rides and XERT already is using this for activities where power data is not available. As an example if my TP is 250 and I go on a ride at 225W for 4 hours my HR drift provides an indication of the strain induced by the activity and provides some indication of potential performance at that point. Definitely in better shape than riding at the TP point but significantly more fatigued than riding at 200W.
Definitely looking forward to a XERT 2.0. I’m still in my 30 day trial but will be signing up for a year to play with things and to see how the automation works for endurance events over 4 hours.
Thanks @cmroy HR can be good but can also be pretty crappy from user to user. Using it without having an eye on data quality can result in poor results for a meaningful portion of the user base, especially given the sensitivity that would be required. Xert is the only platform that uses day-by-day, even second-by-second feedback from your data to inform current fitness settings and training decisions so errors can mess things up. Wish everyone was using a quality HRM.
Whether or not zones are used I find correlating power and HR very interesting. I see this is where AI/machine learning analysis over a period of time could really help identify the most appropriate efficient and effective training. Will be interested to see what XERT 2.0 brings to the table in terms of artificial intelligence/machine learning or training suggestions based on algorithms. So far I love what Xert has to offer. And as soon as I’m ready to start training in January will be signing up.