How Are You Using Xert’s Forecast AI Training Plans?

@ManofSteele Can we have a way of forcing a day to be low intensity only but still allow the Xlss be adjusted by the Forecast AI?

1 Like

Thanks for your suggestion! It’s definitely something we could consider implementing, but it would require a bit of effort and might only provide a minor incremental benefit.

Currently, you can achieve a low-intensity day by pinning an endurance workout that suits your needs. For example, if you have three hours available, you can find a three-hour workout or ride that matches your desired intensity. While having Xert’s Forecast AI choose the duration for you could be useful, it may also add unnecessary complexity to the user interface and algorithm, which could divert our resources from other important features.

We do appreciate the continued thoughts & input from users though!

It would be nice to be able to move workouts around within a given week.
For example, the AI is telling me to do a workout today that could be completed easily indoors while tomorrow is over 2hrs and would be preferable outdoors. It’s a beautiful day today but forecasted for high winds tomorrow so i’d like to swap the order of the two workouts without adapting the entire program.
Is there any way to do this?

1 Like

Yes, more or less. Start by finding a workout for both days using the standard Choose Training dialog. Once you’ve identified suitable workouts for each day, open the dialog again and select the workout recommended for the other day. Afterward, pin both workouts and click Adapt Forecast.

Note that this is not always recommended, so check the predicted training status before each workout. For instance, it might be okay to do 100 XSS today and 300 XSS tomorrow, but if you do the workouts in the opposite order you might start with a red training status on the last day.

I’m using it for the first time to prepare for the Amstel Gold Race tourride (150km version) in April.
Restarted cycling after years of inactivity last year august. Started with unstructured 1h/2h rides. Then followed KNWU Fondo (dutch cycling union) program as prep for the AGR 125km. After that I dabbled in using apps such as join.cc and trainerroad… before finally trying Xert as the outdoor season ended and it’s become too cold and rainy for me.

  • Program: event goal for 19th of april.

  • Adjusting for availablity: I set it to max 8 hours. Mostly I’ve been riding between 4 and 6. But I could do a bit more than that realistically. Just not more than 8.

  • Program setings: I left this mostly to default… 4:1 polarization, 100 xss etc.

  • Final result: We’ll find out in april I guess. :slight_smile: im especially curious because KNWU Fondo which i used before has a lot of longer rides planned… like 2h,3h,4h endurance rides in the weekends. And even the shorter intensive rides are like 1h30 minimum. Trainerroad before and XERT now have a lot more <1h or 1h~1h30 workouts, with bit higher intensity than I am used to. Even zone 2 is like upper end zone 2. So I’m very curious to how effective it will be.
    Also my forecasted TP from XERT in april 2025 should be 240, I’m at 215 now. When I restarted cycling in 2023 I tested FTP in december 2023 (190), after the AGR 2024 (210) and after Limburgs Mooiste in May (220).

1 Like

Adapt Forecast looks at the week ahead to rebalance the plan. If that isn’t viable the routine looks further up the ladder to make an adjustment.
As you approach your target date the ability to rebalance becomes limited.
Adapt Forecast is optional since shuffling days of the week may not adversely affect the plan. You can decide when to run it rather than when the red dot first appears. To determine why it appeared view the forecast chart and hover over any magenta bars to read the explanation.

I wouldn’t rule out long endurance rides on weekends if your schedule permits them.
If anything that will improve your readiness for the event.
They can also be low end Z2 if you want below LTP.

1 Like

I have had mixed results experimenting with XFAI, but after tweaking a lot I can force it to provide what looks to be a reasonable training plan. But I keep going back to the XATA instead. My consideration is that rather than defining the endpoint and letting XFAI calculate the slope of fitness increase, I prefer to use XATA “Continuous” mode to simply define the slope instead. I can then extrapolate out to predict the endpoint, and I find it’s much easier to use this way. When I want to have a recovery week, I just change the ramp rate to maintain, and after that week, I change it back to where I want.

Kind of like driving a manual vs. automatic transmission :slight_smile: Just my preference…

4 Likes

Sounds like a great plan! Bust of luck with your preparation!

Regarding the longer rides, you can definitely get out for some longer rides on the weekend if you can - maybe find a group to ride with! That will help generally prepare you for riding multiple hours in the saddle.

It doesn’t surprise me that Xert has some shorter, harder workouts. Since we can prescribe intervals using your whole fitness signature (rather than just %FTP), there are many workouts in our library with pretty intense efforts! And, because many of our workouts are fairly intense, it means the training is time efficient!

I am registered for the 7-day BC Bike Race the first week of July 2025. I have many years of MTB CX racing as a masters athlete, but have never done a multi day stage race. So leveraging the Xert platform should be quite helpful. I have set the Forecast AI race goal based on the hardest stage, but I know that is not really reflective of a 7-day effort. It would be great if the Forecast AI race goal settings could accommodate a multi day event. (If it already does, I apparently missed that functionality.)

I used the Xert training program to train this year toward a goal race and I think it provided a good training framework. I had a good result and was pleased with my fitness. There were certainly days last spring when I listened to my body and modified what was suggested. But overall I was pleased with how things went. I hope to have a similar experience next July at BC Bike Race.

100% I can’t get an achievable plan otherwise and still… always in deficit… always fresh… (yes, even if slow or lower improvement rate)…

I like to compare the AI’s predicted results for a maximal hour plan to what I actually achieve following my coach’s plan that has a lot fewer hours.

Currently, I’m 2 months into a 3 month plan and I have just hit the AI’s prediction for the end of the plan. I hit the prediction with a LTL of 84 in comparison to the final prediction of 146. HTL 3.4 vs 10.1 and PTL 1.0 vs 0.3.

Congrats! :partying_face: I guess you’re an overachiever!

When was your last Breakthrough? How frequently are you achieving breakthroughs? The projections for training load/signature are based on your own historical data.

Thanks. My previous breakthrough was 7 days earlier and I typically see a breakthrough with ‘No Decay’ every week or two.

IMHO my Low Training Responsiveness value is questionable. The estimated value is 0.4, but for this season I’d put it in the 0.7 - 1.0 vicinity.

If Low Training Responsiveness is estimated over my entire training history of 14 years, then that may explain why it is on the low side.

I finally found a solution to make the forecast work: I stopped using it. There is something wrong in the way it generates a plan. If it was working, we would be able to reproduce goal already achieved with something more realistic than 10h a week (if that goal was achieved with 8)

Idk, so far forecast has been pretty accurate in predicting my progression 1 month in. I guess it’s different for everyone.

Or when there is invalid data in the progression or the user’s signature isn’t up-to-date. That can happen at times.