I don’t know. It’s all relative. Say you do a long form test that was 40min at 330W and 60min 320W. You might have been working up to and completing 60min intervals at 320W, but 330W could still arguably be your FTP but a lower TTE and room to grow “out”.
Not necessarily. In this example you’re talking 3%, 97% is still a Threshold interval.
However I think the point is also, it’s not a switch. You don’t reach a test and one day you can do an extra 10W for all the same durations. Progressive overload either increase the Watts or increase the interval duration. As described above.
Perhaps the decrease needn’t be as big, but that depends on someone’s ability to endure a new Training Power. To say that there isn’t a need for a PL decrease doesn’t leave much margin for error. It’s a lot safer for the system to prescribe you a lower workout which is the same “amount of work”, than assume you can do the same workout at a higher output and have it kick your butt.
As athletes get more familiar with their abilities and the system they can more quickly dial up the appropriate level. My assumption is that as the system gets smarter it will also be able to pick up on individual trends as well as population ones @Nate_Pearson ?
ETA: Having re-read your post I think I agree with the second half of it 
.
4 Likes
if you get sick, it probably makes you weaker. At that time, your potential is probably impacted and therefore your FTP for that day is almost definitely lower. The auto runs after every workout so as soon as you recovered from your illness, and perhaps rode a few days to get back up to speed, your FTP would self correct.
I assume TR is smarter than Garmin but the Garmin auto FTP also auto-ran after every workout and if you took a short break by riding in recovery for a few days, it would lower your FTP. That was annoying bc it suggested the device had no idea how hard you were actually trying. eg, the output may have been low but was it an incredibly low effort?
In the end tho, as soon as you ride to your potential again, it should all be back inline with your actual numbers.
I used my heart rate post surgery to estimate FTP. I guessed with a drop of 20% and that seemed right 4 weeks off the bike. I did 2 light weeks to start before I Ramp Tested. I then watched my HR response to rides of 1 hour or more to guess where FTP was based on % max HR and was able to see gains of 4, then 3% based on previous FTP being 100%. I did a ramp test after 2 weeks (when the surgeons cleared me to go hard) and was right at 13% down. I can see how you could do this with AI, so much so, I was toying with the idea of never needing another Ramp Test.
I don’t have a pic but the sign should read “PL makes no sense outside the scope of structured workouts”
1 Like
PLs are not a measure of fitness, strength, physiology or any aspect of our riding ability. A PL is just a robots estimate of our realative progress in some aspect of our riding. They’re personalised and essentially have no meaning other than trends over time. Even then they are only meaning with the context of FTP movement over the same time.
It goes without saying that when we do training or activities the robot doesn’t understand or only understands poorly the robot will produce less meaningful estimates of our progress. This reflects a weakness of the robot rather than a weakness in the training or lack of progress by the rider.
While structured training has its place and many riders find it a useful tool for achieving their goals others can reach their goals using unstructured training alone.
Unfortunately estimating the impact of unstructured training especially is an intrinsically harder problem and the robot is always going to struggle with it. The robot is better than it used to be at understanding unstructured activity but will probably never understand it as well as it understands structured activity. When it comes to drawing meaningful conclusions from unstructured activity without power I think the robot will always fail.
TL;DR - PLs are a metaphor made up by a robot. If you find them useful you can make the robots life easier by including more structured training with power in your training regimen.
Personally I view Progression Levels as correlating to increases in fitness. In general, as I progress to higher levels I am challenging myself with workouts that have a greater Intensity Factor for my given FTP. Thus, if I can successfully continue to push to higher levels, it tends to indicate a higher level of fitness. At least that is my view.
Currently, I’ve chosen to stay at my current FTP (though the AI has suggested raising it), as my desire is to work to higher PL levels in both Threshold and VO2. As I break through higher PLs, I certainly feel that I am gaining both greater fitness and endurance to maintain a given power level.
3 Likes
If the robot does its job properly you should see strong correlation between your Progression Levels and your fitness. As you push yourself harder your body adapts making it capable of performing more intense workouts in the future. This improved fitness also makes it likely you will need to increase the intensity of your training to see future gains.
In general when we are training sensibly with proper recovery and nutrition our PLs should be good predictors of future. However I view PLs as more a measure of past workload than current or future fitness.
When I do structured training the robot does a fantastic job of adjusting my workout to maximize my gains. I’m a fan of the way PLs contribute to this. When I focus more on unstructured outdoor training the input from the robot stops being useful.
Your choice to target higher PL rides at a lower FTP is a valid one. The tradeoffs involved in this choice is one of the things that makes me view PLs as a tool to fine tune your training rather than a true measure of your current fitness.
2 Likes
General Q, possibly for the TR guys (@IvyAudrain ) - if I have a ride where my PM was massively overreading, and I exclude it from my PRs, will it still count to AI FTP and possibly warp it?
I think it was overreading by around 15%, which is pretty substantial… apparently I did 350 watts for 20 minutes, which would be pretty unprecedented!! A little over 300 would be about right (and on a level with the other guys I was with).
Also would it be possible to edit the power data in TR somehow like you can on intervals.icu? I have applied a 15% reduction there to bring it back to where I think it should be. Unfortunately I can’t see an option to download that from Intervals, or I would delete the ride off TR and replace it with the Intervals reduced version. (edit: I adjusted it using fitfiletools and reuploaded to TR so should be all good now…)
@Nate_Pearson Thanks for the clarification on AI FTP detection and the update on unstructured rides counting toward progression levels. Feeling the love…
2 Likes
One of the key aspects of RPE is the acknowledgement that we all have good, bad and indifferent days. Sometimes things happen that may lead to shortening a workout, giving up altogether, equipment failure, for example the West Australian summer we just had, had periods of extreme heat, and despite best intentions, it was unrealistic to complete say a couple of weeks of full on Threshold.
I wonder how large a FTP change is worthwhile? Is it better to complete a series of workouts, or better to under achieve at a harder FTP target?
Related to increasing the training load and intensity, is the ability to tailor auto plan adjustments; I say this because often the auto upgrade is increasing workout duration (Volume). I don’t equate Volume with Regularity. In fact I think they are very distinct and ought to be recognised as different plan characteristics. Hence, a low volume plan for short events, may require workouts for 6 days, ie, high regularity, but all low duration, some being endurance, some Intervals. Does that type of plan equate to a low or a high volume, clearly the term doesn’t make sense.
The TR plans require two separate questions, ie, how many sessions per week, secondly low/medium/high duration. Within these Intensity adjustment is also required. FTP, the focus of the proposed auto change, fits in here somehow?
1 Like
Love it! Great idea, got for it.
It’s great that y’all are trying to anticipate our concerns and find plans to address them, but here are my only issues:
- Do the FTP changes, whenever and however they occur, help modify my plan (or Train Now recommendations) to provide the best workouts for me?
- Does the AI system take unstructured outside rides into account?
- Does the AI system take reasons for failure/skips (sick, vacation, equipment failure, etc.) into account?
So I voted auto and manual however, I have a ramp test tomorrow leading into speciality phase and i’m now swaying towards just auto,
(In my limited expereince) It could save me draining a day, achieving a number to base my workouts off, where I could just get on with my training. The ramp tests aren’t that taxing but that time could be better spent fullfilling a productive (swim/bike/run) workout.
Having said that Im new to TR (started in Feb) and as Ive seen my FTP grow exponentially I feel like I don’t spend enough time in the higher PL numbers because the ramp tests intervene and bring my numbers down after each test.
I welcome the update but a manual over ride would still be desirable for me as I can’t see the need to refresh so frequently. I’d manually over ride my FTP referesh to only update after a block.
I don’t think AI FTP adjustments mid phase of a block help, AI Detection between phases is more than enough and I dread the thought of users getting obsessed over marginal gains (especially so early on in training - speaking for myself)
I appreciate the developments though and love the feedback loops connecting the forum to the developers.
It’s a sick App.
4 Likes
I think the option to ‘check’ your estimated FTP is a good one.
If you’re coming up to a ‘not “A” race’ event and want to get a gauge on a target power to test where you sit it might be handy.
6 Likes
- people can’t drive manual anymore.
3 Likes
I feel like people are now just messing with Nate by asking the same questions over and over.
1 Like
Wasn’t posing the questions looking for answers. Like you said, they’ve been asked and answered before. It was more meant to illustrate which questions users (or least this user) worry about. Probably could have phrased it better.
Imagine being new(er) to structured training, doing SSB, having your FTP increase every 1 or 2 weeks, your PLs reset, and be stuck doing the same few workouts over and over (at least until the next phase). Even doing ramp tests every 6 weeks, I would have appreciated more variety in those workouts. I’ve found it more fun to experience the variety of workouts as I increase in PLs. You really don’t want newbies getting bored early on.
2 Likes
Now that you’ve edited that workout to reflect an accurate power profile and re-uploaded, your AI FTP Detection will work as expected.
If you were to leave that workout as-is and just remove it from your PRs, it would still count towards AI FTP detection with that super high reading in mind. We don’t have a way to exclude workouts individually from the AI FTP Detection algorithm yet outside of deleting the ride from your career entirely.
Checking in with the team to see if we have plans to isolate activities and remove from AI FTP Detection logic. I’ll update if new info arises!
3 Likes