Beta Forecast AI - Super interested

This should work better now. Any ride you do on a day should go against the forecasted training activity unless you have a planned activity before it and record an activity at that time. You can mark the forecast as a rest day if you don’t plan on doing it.

1 Like

This should also work better now.

Indeed, now it almost matches:

@xertedbrain Could you please comment on which of the settings sliders currently have any effect on the forecast calculation? Versus which maybe will have effect only on future or also never?

More questions:

In case of going for “personal best”, is there any time frame you would recommend to target? I tried now until July (vacation trip) and I needed to give it a few tries (lowering my target of choice) until my wish goal was found to finally be achievable.

And how is it handled if during the forecasted training period I face BTs which doesn’t align with the assumption of the forecast AI regarding my improvement potential? Let’s say, mid of the training period I get a BT which shows that I’m getting stronger faster than what was assumed while calculating the inital plan.
Would the “goal” move to an earlier date?
Or would the estimated target signature adapt?

1 Like

All but freshness feedback affect forecasts (taper adjustment only for events). More documentation and help is forthcoming.

I wouldn’t set polarization to 90/10 since that will make your high intensity workouts the most challenging. 80/20 still will give you very hard high intensity workouts (some can be done while still yellow status if that’s what works best for your schedule).

Periodization only takes effect for target dates more than 120 days away. Lower it if you want more training all the way through.

Leave taper where is it for now. More info on it coming…

You can experiment with these by changing them and running the set up guide. Just don’t save the forecast and return them to their defaults.

I played a bit with the sliders. Now the calculations won’t terminate. I.e., the save button doesn’t become active, even minutes after the progress bar reached 100%.
The graph still jumps a bit up/down every few seconds. So looks like it stucks in some endless loop.

Now looking at this graph I also wonder why focus power would first lower a tick before starting to climb up?

Long training plans take longer to compute. We’re working on that.

If you turn off periodization, you won’t get the dip in focus power. You’ll also notice that until about 120 days before your target date, all you have is base/low intensity. That’s a long training plan to be dedicated to.

1 Like

This seems to be working now. However, there does not seem to be a way to set maximum number of days per week. For example I am available for one hour a day every day except Saturday and Two hours on Sunday. That said I only want to workout a total of 5 days per week - any way to set that? Right now It is basically scheduling something on every day I’m marked as available.

First lets get the ability to actually take availability into account and still optimize a plan. This is immensely challenging task and the variations and possible combinations are hard to fathom. Once we know we can get it to work, we can make it easier to input the information. Hope you understand? :smile:

2 Likes

Of course! I’m just trying to figure out how to use it in my real world.

Right now when I change my plan from continuous to an AI generated plan it is populating the calendar for every day of the week and significantly lowering xss value daily workouts. Let me try marking one more specific day as unavailable.

Best tool keeps getting better!

That worked - I now have T-F +Sun as available. 5 Days total.

Daily XSS recommendations are now way below what I have been actually doing when set to “continuous, moderate -1” - Granted, I would probably not get to the 8.5 hours a week that is forecast under that plan, but I am working harder on a daily basis than the AI generated plan. I am doing ~75XSS/Hr on slow days and 100xss/hr on HIIT days - plan is recommending 60/hr and 75-80/hr respectively. Plan has 5 set days/week and actual past history varies, so total avg XSS/day seems to be same but spread over 5 days/week under AI plan.

I feel like I’ll lose out by switching?

What happens if you increase your goal or bring the target date up closer?

Oddly enough - now looking at the Training tab, I see this:

Target had been set for 255 Watts and 220 TP

(same result with goal at 215 TP and date a month sooner)

Usually is red and says unattainable

Try again. The algorithm has been updated.

I still see mostly 60XSS days

If all you have is a fixed 1 hour time, that’s pretty much all you’ll get in terms of XSS. If you have more time, increase your availability and/or change it to flex.

Why is that? I regularly do 80 to 100 xss an hour.

I’m getting a very strange bug, I am setting my ‘goal’ then setting my daily availability but when I forecast the system seems to be ignoring the rules I set entirely. My availability schedule is as follows:
Mon - 1hr
Tue - 1.5hr
Wed - 1.5hr
Thur - Unavail
Fri - 1.5hr
Sat - 3.5hr
Sun - Unavail

You can see my set goal here.

This is the weeks training that keeps generating despite my parameters.

Even though I have set a 1.5hr availability on Fridays, I keep being prescribed 1x workout a week at up to 4hr on a Friday?
image

I have reset the parameters multiple times and this is all I get. I have also tried it in various browsers throught the day, I’ve logged out and back in, nothing works. Very strange bug.