Adaptive Training Advisor nonsense

Yesterday, after my ride (Endurance), I hit refresh in my browser and the Adaptive Training Advisor said I was Tired and recommended a ride with 30 XSS, Active Recovery 60. I didn’t think to copy the exact text. So that was my plan for today - an easy ride for about an hour. Out of curiosity, I hit refresh again before I was going to ride. Now it says this: “Your current Training Status is Tired and should consider a Endurance activity or workout, generating about 276 XSS of overall strain.” WTH? So it thinks I’m Tired yet recommends 276 XSS workout like SMART - Iron Man - 180? I don’t understand…

That advice was generated immediately after your activity and was valid at that time.
Note the status stars color represents predicted form as of now while color of the arrow in the pacer needle represents your predicted form 24 hours from now.
If your browser is left loaded with a cached page you should always refresh the page as the circumstances change hour by hour.
XSS goal factors in what you did last week on same day that you may choose to ignore this week unless you are purposely following a weekly pattern.
The workout XSS is determined by your current signature. View that workout in Workout Designer and you’ll see lots of time is spent below LTP with brief bursts into red. IE, not a hard workout but a long one. If not practical for the day use Filter to change duration and cast a new list of recommended workouts.

Reference –

Understaning the Training Advisor - General - Xert Community Forum (

Training Advisor recommendation (focus and target) - General - Xert Community Forum (

1 Like

OK, I appreciate the attempt to help, but what’s your point? What advice is the ATA providing?

I finished my last ride at about 6 PM yesterday. I checked Xert ATA a few hours later to see what I would be doing tomorrow. It said Tired and recommended 30 XSS. Made total sense to me that after 180 minutes (117 XSS) Endurance workout that I would do an easy Recovery ride. Less than 18 hours later, with no input from me (I don’t touch the Freshness slider), it is recommending 276 XSS despite the fact that it still thinks I’m Tired (which I am). That simply makes no sense. The ATA is providing no clear actionable advice in this case.

Wouldn’t it make WAY more sense if the last line of text in the ATA advice actually provided actionable advice? Advising me to perform a workout 276 XSS is just simply not useful no matter how you decode it with the secret Xert handshake and decoder rings.

If I should just do what I feel like, what’s the point of paying for Xert? Shouldn’t paid advice be a little more clear than mud?

What’s your average Tuesday ride over the past 6 weeks, looking back at your calendar?

Point being the guidance isn’t random nonsense.:wink:
The model is driven by your historical data on file (to determine status and form) and the settings you’ve configured at this point in time (to set your goals)…
TL (XSS) and focus (strain ratio) are the two main components required to stay on target (pacer needle) based on your current settings and form. From there any choices you make depend on whether you decide to ride indoors or out and the time you have available to train today.

If you’d like a deeper dive explanation of “today’s advice” refresh your browser and post screenshots of the Training tab page including current profile (Star Status/Form and Signature), ATA, ATP, and Recommended Workouts.
Also select XO, Activities, Table and capture a screenshot of the last 7 days.
(We’ll assume you have at least 30 days of consistent power data on file to feed the model.)

1 Like

If you want to see the training advice for a future date, go to the planner and click on the date you are interested in.

1 Like

I have rarely been riding on Tuesdays - only twice in the last 6 weeks one with a 68 XSS and the other with a 466. Maybe that’s why?

Well it looks mostly like random nonsense but given what’s been offered here so far, I see why the models may be having trouble. Most of my rides for the past few months don’t have power data (mountain bike). I only have power data when I ride the trainer indoors which is usually winter months, this year being an exception starting last week because I want to get in a legit training block before an upcoming mountain bike race.

I still think it would be nice if the ATA actually gave a simple clear recommendation that wasn’t obviously nonsense regardless of the inputs. I don’t think most users want to see what is basically the raw output of some formula.

I tried Xert last year and got frustrated for the same reasons but also because it wasn’t considering rides without power data. It’s nice to see they’ve at least made an effort to include those using Derived metrics. I do normally wear a HR monitor outside, just don’t have power meters. It would be extra cool if Xert had a way to set the ride type to road or mountain - like add 50 watts across the board or whatever for off road rides.

Having power meters would be nice but they’re expensive and my bikes are a bit old so I’m not willing to bother at this point. Too many crank “standards” and too much crank abuse on the off road rides I enjoy.

Now that’s helpful! Thanks! I seriously had no idea I could do this! I’ve looked at the planner but never actually tried clicking on the calendar.

I used a powercal HRM for several years, bought it on eBay. It was developed by cyclops. It can be used as a power meter. See dc rainmaker write up. It worked well for me for 30 sec avg and above. Good for ftp, avg pwr, nominal power but not sprint efforts. I calibrated the algorithm every 20-30 watts improvement. This should help you on your mtn bike to use Xert.

It can also be used for measuring power during running as it only cares what your heart is doing. Ultimately your heart drives your whole engine

You will probably need to use a Strava add on periodically to edit erroneous pwr readings under 10 sec. takes a couple minutes to do this.

This is a cheap solution, using a pseudo pwr meter but it will work. I checked often against my trainer pwr meter. Again not perfect but keeps you in the game

Just checked online - no longer available. I would sell you mine but I use it for mountain biking….lol

I wonder how the Xert Derived power data compares to what you see from this HR Power meter. You’d think they’d be very similar. Does Xert update your Fitness Signature with data from this device? I assume it does but I wonder if the result is good or bad. I suppose it would be better if you still use your trainer often enough but maybe not so good for stretches where you only use the HR based one.

I am not sure how they would compare. I am not familiar with Xert’s derived power calculation to comment. I am assuming this is a post ride calculation, like the one on Strava. I don;t see an App for it on connect IQ.

The benefit of the powercal power meter was you could see your power real time (30 sec avg +) on your outdoor rides and observe your overall effort… including breakthroughs. Based on my own experience periodic power calibration was needed as your fitness improved for the powercal power meter device to remain accurate.

Last year I bought a real power meter but I still use the powercal to record other activities (running, elliptical, mountain bike, etc…)

I just reviewed Xert’s page on Heart Rate Derived Metrics. The information provided is well developed based on what I have researched in the past. I would run a test on a trainer and compare results between the file from the trainer and your bike computer (wahoo, garmin, etc…). If the match looks close this is a good solution. (I imagine the QA/QC work has already be done by XERT but it would help you understand the strengths and weakness of this method and adjust accordingly.

I think Xert could +create an APP on connect IQ that calculates derived power real-time :slight_smile: :smiley: The fitness signature would maintain the algorithms accuracy with no calibration required.

PS: when I say real-time I should say 30 secondish lag time depending on own heart responds to an effort… I can know see this might be problematic if you tried to incorporated it into a fitness signature :frowning:

HR data is far too wonky (scientific term :slight_smile: ) to be used to derive power in a real-time sense. You could derive XSS and Focus in real-time but power … not so great. Deriving power looks at HR rate of change and this calc is too sensitive to error. We know because MPA and essentially Xert evolved from doing precisely this (deriving power from HR and cadence) so we have a very good algorithm for it. We also know how hard it is when there are errors in HR data.

We’ve thought about creating a datafield for XSS and Focus based on HRDM but that math is difficult to port over to ConnectIQ. At the moment, it’s a pretty computationally heavy operation that runs on the server. It may make it to Xert EBC since there are greater flexibilities and possibilities when developing in Android.

1 Like

I can’t make sense of the XATA recommendations the other day either. I did a pretty big mountain bike ride on Sunday (277 XSS according to Xert). No power meter on my MTB so XATA shows DERIVED.
Planned to do a recovery ride on the trainer Monday evening, about an hour Active Recovery. Out of curiosity, I checked to see what the XATA recommended. It said I was Tired and recommended an Endurance ride with about 178 XSS.

I didn’t end up having time to ride Monday but I did on Tuesday so I prepared for a ride with at least 178 XSS, about 2 hours on the trainer. I checked XATA and now it said I was Fresh and recommended a Breakaway ride with about 78 XSS. I don’t understand why the recommendation would go down since I didn’t ride at all since the big ride on Sunday.

Maybe it’s because you recovered (yellow to blue presumably) and therefore recommended focus changed to high intensity, which typically have lower XSS values. Getting the right focus seems higher priority in this case?

Could also be linked to ‘usual activities’ on that day… where normally you ride longer on Mondays than Tuesdays? (Personally am not sure that usual activities on day X thing adds a lot… schedules change enough that most people ride different days each week… even short vs long for weekday vs weekend is not always the same - vacation, public holidays, long summer evenings etc)

I considered the usual activities thing but my schedule has been pretty inconsistent with all the wildfire smoke out west.

I was in Build the week prior and the week after, so I’m not sure a change in that type of focus explains it either. So far I’m not getting much for actionable suggestions from Xert but I also don’t claim to understand it very well yet either.

TrainerRoad was easy - I just did what it said. The problem is I got slower using it probably due to over-training probably because neither it nor I gave consideration to my outdoor rides without a power meter. This stuff has so much potential if it actually worked.

Even without changing focus in build Xert is monitoring how much high intensity (above TP, whether from powermeter or HR) you are doing. It sets your status to yellow when ‘tired’ (colour of stars, see planner for history and projections) and only recommends endurance focus (lower intensity) when it thinks you’ve not recovered. Once you are blue status again, it recommends your targeted focus. That’s how it achieves some degree of polarization

1 Like

It simply used your average Tuesday ride XSS since it assumes you are going to do something similar this Tuesday as other Tuesdays. If you have more time, filter for a longer workout.

Focus changed because of Tired to Fresh status.

It works perfectly once you understand what it’s trying to do. There’s no magic.