Adaptive PLs and IRL/Outside Rides

Any idea when TR Adaptive Training will credit IRL rides? My 1-hour TR ride at IF=0.66 on Friday increased my ENDURANCE PL +1.3. Then my 130 mile IRL ride on Saturday at IF=0.65 didn’t give me any PL changes. That should be pretty easy to do, since TR has the power meter metrics and knows the IF and TSS and power profile. That is probably my #1 wish. Other than that, having just started TR a couple weeks ago, I am all-in using this platform - totally love it.


This is the $million question and nobody has any idea…given what TR have said about the challenges of doing this and the lack of any visible progress for a long time, the general view is not to hold your breath…

do it as an outside workout and you’ll get credit. Honestly though, I would just accept the fact that the ‘Endurance’ PL isn’t a representation of ability. Do you do actual endurance work(not 1hr) on the trainer?

1 Like

@Cory.Rood - I’ve scheduled repeated OUTSIDE 200km rides every Saturday. I did an outside ride on Saturday (130 miles). That IRL ride automatically pushed to TR and showed up on my TR calendar. So Saturday shows my IRL ride (completed) PLUS the scheduled 200km Outside ride (not yet completed).

What is the trick to get my IRL rides to apply to the scheduled Outside ride? Thanks!

Find an endurance ride, there is at least one that is 6 hours in the TR Library. Make sure it has an outside version. Add it to your calendar on Saturday and set it to Outside. After the ride you can match. That was working a few months ago.


What challenges did they bring up? I am sure it is not an easy feature to develop but wouldn’t it be as “simple” as awarding PL increases based on times spent in each power zone?

It needs to be a TR workout, not just a ride you put on the calendar.

Have a search - there are literally hundreds of threads about it. Its nothing like as simple as that, as obviously they firstly need to get the data back from the head unit (and they cant get it frorm Wahoos for example), then understand not just TiZ but how long each interval was, rest periods etc etc. Clearly a load of 30 sec periods in z4 across a 2 hour ride is a totally different ride to a 2x20 steady z4 effort etc.

If it was easy it would have been done months ago.


Also there’s no “target” for IRL rides in the same way as for an indoor workout. AT works on you being told “ride for X minutes at Y power” then analysing how well you did so. Where’s the target for a Sunday group ride? AT doesn’t know what you were trying to do so can’t work out whether you actually achieved it.

As above, schedule a long TR workout that has an “outside” option and then associate the ride with that. Not perfect but it’s the best we have at the moment.

1 Like

@bobw thank you. Makes sense. But that approach seems limited and “backwards”. The way you describe seems like every workout has to be tagged as awarding “n” PL points for each of the seven zones if you finish the workout and follow the prescribed power profile.

It seems to me a MUCH more flexible and scalable approach would be to analyze any ride you do based on total duration and sustained duration in various % FTP power bands. And then award PL changes based on your performance profile. Then any IRL ride and any TR workout would just work.

I’m happy with the TR plans so far. Maybe some day…

1 Like

TR are definately looking at awarding PL progression for outside rides - but haven’t given any timescales.

I’m wondering if they could at least impliment an intrim stop gap solution where at the very least outside rides stops progression level decay?

1 Like

but you also have to remember that AT is designed to analyse your performance vs plan, and adapt the following workouts based on the result. AT is NOT primarily about PLs, its about tracking and evaluating your performance and adapting. PLs are just a label that allows workouts to be compared in terms of difficulty.

Yes this is a common excuse, but I think it’s a complete flaw in the design, yes it can’t match to what you were trying to achieve, but it can determine what you did achieve, and surely this would be a far better way to have gone (as in how Xert, Join and Breakaway work)

The fact that if you intend to 2 * 20 @ threshold, but only achieved 2 * 20 @ Sweetspot, isn’t picked up AT, but you still achieved a good training session, and should be credited with that, infact the current system is more give you the wrong credit for the wrong group

So with outside workouts it should be the same, who cares what you were trying to achieve, if I go out with the intention of doing 2 * 20, but end up doing 4hours of Z2, who cares what the intention was, what did I actually achieve

After AT went live, and people were complaining about outside rides, I did point out that Nate said in the AT launch video in Feb 2021 that he has a working version of outside rides, TR replied a said that version has been scrapped, as there were unforeseen challenges, and a completely new version

I agree that’s a more flexible approach and is, as I understand it, what TR are ultimately aiming to do but it’s not trivial.

It’s some years since I did ML/AI but in “training” the system you need to be able to categorise both positive and negative outcomes so while it’s easy to say that an IRL ride contributed X minutes to a particular power zone (in fact TR already do this, see - ) you can’t say “why” or indeed if it was desirable. Then you need to get the data into the system, outdoor rides are inherently “noisy” data wise - is that 30 seconds over threshold intended or just the rider dealing with a slight rise in the road or a bit of headwind? So the data needs to be “cleaned” in a way that rewards intended effort whilst to a certain extent ignoring the noise but still taking into account that it’s occurred.

I suspect (but really don’t know so it’s just a guess) that Nate’s original “working” prototype was something fairly simple along the lines of “X minutes in Z1, Y minutes in Z2, etc.” and they found that it messed up the ML. There’s a good argument that it shouldn’t but…

But you should be able to say if it had an effect, it may not have been intended, it may not have been desirable, but it may (or may not) have had and effect, you shouldn’t need to clean anything, standard deviation could be used if you are looking to see what the base line of the session was, but that shouldn’t be required for the training effect of a session, cleaning is basically what TR are doing when only allocationg PL from what you intended

Unfortuantly I can’t see your link, i’ve left TR again (and having training gains again)

Even Whoop can look at my training yesterday, and detemine some sort of (basic) training gains, without needing to know what I intended

But that’s the problem with discussions, I suspect / believe / understand, aren’t based in fact, all we know is that the first version didn’t work, the second version isn’t ready for release

i’ve said it before, I believe endurance PL’s aren’t a thing most of us shouldn’t be all that concerned about. It’s an area that for most plans doesn’t really get a lot of adaptatons/progression (generally they’re for easier rides). I’ve done what I’m sure is 10 level work on a few rides this past year and didn’t have them associated with anything, it really would have made zero impact on my training plan, considering key workouts are generally sweet spot or higher. Plus, big rides don’t currently cause future workouts to scale back.

I think progression levels can be a bit of a trap/vanity metric, esp for endurance. Many cyclists with a reasonable training level can go out and do a pretty high level z2 ride. And as an aside, I really have doubts whether a lot of unstructured rides outside of races even contain enough work to actually count toward positive progression levels in a lot of areas.

1 Like

Agreed, but not all outside workouts / races are endurance/z2, most saturday mornings I do a 1hr time trial, thats one hour of ss/threhold, if that isn’t included in the PL’s, then all other workouts would be affected

well that’s why I used the term many, and not all or most lol

but even an hourish of sweet spot is easy enough to associate with a workout and not be ‘unstructured’ or at the very least one can be aware of how their hourish efforts and accept/decline/sub workouts

I think it’s the random kitchen sink rides, where maybe you’ve got like some sweet spot, maybe a couple vo2 efforts that people are getting worked up about and they probably don’t contain enough efforts to really register very highly on the respective progression levels

1 Like

That is why I used "simple’ in quotes: I didn’t think it was easy. I was just more so curious on the challenges developing it would require.

Thanks for answering.

Totally agree with you, I was just a little confused (not by you) that effect could only be measured if you know the intent, but totally agree the the effect might not be worth measuring … and that it’s not easy