AI FTP, RLGL and Outside Workouts without Power

Hey folks,

Firstly, let me say, I really love Trainerroad, especially the massive choice of workouts and how well it syncs with the Kickr and Garmin.

I can see how well AI FTP and RLGL works with power data too

However, I am finding, possibly, that it just doesn’t seem to always make sense when I only have HR data. I only have one outside bike with a power meter. I have just got a new gravel bike, but the SRAM quarq has died, got sent back, I got a new one, and that died too. Most of the time at the moment I am on the graveller, or the big bike which also is without a power meter. So I am relying on HR data and perceived effort for how hard I am working for the majority of my rides.

No one wants to sit on the trainer in this weather, do they?

The last month I have seen some weird decisions by the software that I am just not sure about. I’ve been mostly training outside and it often gives me massive TSS scores for what are, in all honesty, rather gentle rides. Other times it gives me much lower TSS than I would expect, or a yellow day when it really should be red in my eyes, or a yellow day when the day before I barely did anything.

Sometimes the TR software totally contradicts my Garmin - I don’t use my Garmin for training but it is interesting to see. Today Garmin told me I should do V02, which I had already chosen to do as it definitely felt like the right thing to do today within my training plans for the month. Trainerroad gave me a yellow day today. A yellow day when, all I did yesterday, was pootle around with a friend at endurance pace or, tbh, recovery pace for a lot of it. The system gave me 112 TSS and a yellow day today! I could have ridden at that pace for days and days on end. So weird.

Yet a few weeks back I did an absolutely savage Audax with several off road sections and hours in z4 threshold heart rate (I know this is not the same as threshold power, but it was still effort). Crazy HR numbers for over 9 hours of riding (22 mins in z5, 4.39 hrs in z4, 3.36 hrs z3, the rest recovery) and the system gave me 220 TSS, then RLGL gave me a yellow day after that. Surely to goodness this should have been a red day. How can a ride of pretty hard effort for that long only have 220 TSS when a 2 hr pootle at barely out of breath pace gets 112? It makes no sense.

AI FTP has also just put my FTP down, again, even though I feel stronger than ever and just easily got through some 30/30s. Honestly, they felt the easiest they ever have. I did have some sick time this last month (after the audax, unsurprisingly - I had to ride through fields full of cows at the end just when my immune system was probably at it’s lowest), but still, I have got a good deal of outside work in over the weeks since the last detection. I still accepted the FTP reduction because I am trying to trust the process, but, I was wondering, is anyone else finding this who has to rely on HR/PE for their outside rides?

Recognised this potential issue early in my training ‘career’ and the only power meter I ever bought were power pedals - can move to any bike or device. They’re a lot cheaper now that they were in 2016, could be worth your investment.

1 Like

Thanks JoeX, but I run flat pedals on all my bikes so power pedals are not an option, unfortunately

1 Like

Hey there,

Totally get that you don’t want to sit on the trainer when the weather is nice!

I think it would be worth checking that your max cycling heart rate is set as accurately as possible. You can check what value your max cycling HR is set to on this page.

More instructions and FAQs on HR TSS estimation can be found in the following article:

Getting those settings dialed in (if they’re not already) might help you get more accurate HR TSS estimates.

That said, RLGL and AI FTP Detection will both function at their best with power data. AI FTP Detection in particular needs power data to give you accurate FTP detections. If there isn’t any power data being fed to it, then it won’t have any information to use when calculating its FTP detection for you.

It may be best to simply not run AI FTP Detection while you’re training without power data. It won’t have enough data to give you an accurate reading. If you end up training for several months without a power meter, we’d recommend doing a Ramp Test to re-establish your power zones to ensure they’re accurate. After that point, if you’re able to continue training consistently with power, you’d be able to use AI FTP Detection more reliably. :slight_smile:

Without power data, we’d recommend leaning on RPE a bit more than HR. HR can vary considerably day-to-day, while RPE tends to remain more stable.

Taking note of your yellow/red days would still be a good idea, but also stay in touch with how you’re feeling. If you do another big 9-hour ride, for example, and your HR TSS estimate doesn’t reflect how you’re feeling, certainly feel free to take some extra rest.

Hope this info helps clear things up – feel free to let us know if you have any additional questions!

Thank you for getting back to me, you guys are always so helpful!

I’ll start with some good news - SRAM have now sent another crank for the gravel bike - so far it seems to be third time lucky as it actually worked today. Hopefully that will get me back on track with data soon. I still won’t have a power crank on the big bike though, which I tend to do most of my hard weekday riding on.

My Max HR is somewhere close to the 188 it’s set at. Might be a touch higher.

I do, however, have more questions

  1. If I stick with the current reduced AIFTP figure, will it adversely affect my training? I mean, the VO2 max workout I did before the system dropped my FTP felt relatively easy. It then dropped my FTP by 3.8% or something. Not much, but surely this will just make my workouts easier, so does this mean I just won’t be working hard enough in VO2 and Threshold stuff? Or is 3.8% negligible and I am better just ignoring it and letting the system do it’s thing? If the workouts continue to feel relatively easy, should I manually amend the FTP figure, or should I just run with it and see if I can improve my PLs over the next 28 days? I really don’t want to be relying on ramp tests if I can avoid it.

  2. Can you explain why the TSS of the easy 2 hr ride (112 TSS) and the higher paced 9 hr audax (220 TSS) are so mismatched? I literally cannot fathom how the system has worked out the TSS for these rides.

  3. Should I untick the AIFTP detection in the career inclusion box for each individual ride I don’t use a power meter, or should I just ignore AI FTP detection completely until a certain percentage of my rides are being done with power. If the latter is the case, what percentage should it be? Over 50%, 80%?

Thank you!

1 Like
  1. Use the automatic AIFTP and talk to support if you are worried about a problem. Aim to do enough workouts with power that the system will work.

  2. I would not worry about TSS, it’s an output that isn’t important. My hrTSS rides are sometimes double the TSS of my powered rides.

  3. TR will know if the ride has power or not, so I would not bother altering it.

Mostly you just need to know if you’re training regularly, and getting a little bit harder each week of key workouts. With a lot of outdoor unstructured riding that is going to be hard on paper, so you’ll need to be honest with how you feel, when you’re going too hard and when you need to rest.

1 Like

Thanks @JoeX

Can I just ask a couple more questions

  1. How many workouts with power do you think I need to do for the system to work?

  2. Interesting perspective on TSS. I don’t disagree with you, however, it seems that the system does utilize it though, as part of adaptive training/RLGL? If it does, surely the TSS generation needs to be correct, if it is feeding in to the training plan overall?

Thanks for your input, appreciate it :slight_smile:

1 Like

No worries

  1. You need ten indoor workouts with power before it can start estimating your FTP from workouts rather than doing ramp tests. From then it will adapt FTP from any ride with power.

My experience has been that I stop progressing FTP if I stop all indoor workouts, but you may well be different. I’m trying to do one indoors and two outdoors now per week.

  1. How it works exactly is ‘secret sauce’, but I think it’s reasonable to conclude they are working on the training data as a whole rather than any meta data like TSS.


Since you are working with limited data, it’s important that the data you are working with is accurate.

My Max HR is somewhere close to the 188 it’s set at. Might be a touch higher.

Let’s make sure that you get an accurate measure of your max heart rate.

all I did yesterday, was pootle around with a friend at endurance pace or, tbh, recovery pace for a lot of it

Likewise, I can see that you didn’t submit a survey response for this ride. Make sure that you submit a survey response for all of your rides. Without power data, other information carries greater weight so for the best results, it’s important you give us all of the information you have available.

the system gave me 220 TSS

Can you please point me toward this ride please and I’ll look into it and see what might be going on here?

I agree with Zackery here. I’d suggest doing the ramp test to calculate your FTP until you are more consistently using power.

Once we make sure that we have all of the information we need, you’ll start to see the power of RLGL!

1 Like

Hi @SarahLaverty thanks for your response

With regard to HR, my Smart watch tells me it’s 191, however, I am not sure I have hit that at any point since I have had the watch and it’s a guesstimate. I do hit 187-188 when working hard outside though. I rarely get close to that on the indoor trainer, I just an seemingly unable to get my HR up on the turbo. I’d say 188 is probably about right, it was 212 when I was 20, 202 when I was thirty, around 192 when I was 40, and I am now 46. I only remember these figures because the steady decrease of 10 watts per decade appealed to my geeky side.

Ah, I didn’t realize I hadn’t submitted a survey response for that ride, I guess that, because it doesn’t pop up when the ride is automatically imported from Garmin, I just don’t remember. Something I shall pay closer attention to in future.

Anyway, looking back the bigger ride also didn’t have a survey submitted - although that may change the TSS, it still doesn’t explain why that TSS seems to be low, when the other seems to be high for the easy ride. Both are based on HR data. I can’t edit the survey response for the big ride now as it is after 7 days.
The ride is activity 302312384

In terms of using AIFTP/Ramp Tests…I now seemingly have a power meter on the gravel bike, so as long as that keeps working, I should get more power data from that, and the XC bike. However, getting a working power meter on the enduro bike seems unlikely at the moment as the only workable solutions are very expensive, and, to be honest, I am not keen in investing in another SRAM power meter until the third replacement for the gravel bike has proved itself reliable. So I am always going to have my harder technical weekday ride, and the occasional weekend big day out, without power at the moment. I will, however, have all my longer rides, XC races, outdoor 5x5s and indoor turbo figures with power. Will that be enough?


Okay great stuff. It sounds like your Max Heart Rate is reasonably accurate then!

You are in good company :nerd_face: :wink:.

Great stuff! As I said before, the more data you can give us, the better.

I’m struggling to find the ride. Can you please send me a link to the ride on your Calendar? You can do this via dm if you wish. Or you can just tell me what date you completed the ride. Whatever works best for you!

Yes, that is enough for AI FTP detection and RLGL to work well! The more rides you have with power, the better. However, it is less important to have power on technical rides than it is on your harder, more physically demanding Workouts that are both going to move the needle in your fitness and stress the system.

I think you are now in good standing to really see the power of Red Light Green Light, AI FTP dection and Adaptive Training as whole.

I’m excited for you!!

Let me know if you have more questions :+1:.