Beta Forecast AI - Super interested

Whoa that is trippy!

If I leave periodization alone, and setting mine at roughly 120-days out from a race and a “reasonable” xss estimate I get something that looks way smoother than you.

If I turn periodization off - it is still way smoother than yours.

@xertedbrain this is now fixed, thank you.

Another question: Why is the daily availability picker only available after you set up a training plan? If I were to choose a new plan now for an event that is much larger than what I am trying to deal with now it would tell me it is impossible because there’s not enough time. What I would like to be able to do is set up the daily parameters before it tries to generate anything so it takes into account the available schedule, not recent history and not what is set there now.


Yep, I have exactly the same issue, I posted about it above.


@ridgerider2 and @xertedbrain

Following the new AI schedule and advice, being recommended mostly 60xss on the low days, and ~85 on the HIIT days. Any reason not to stretch for 90 or 100 per hour on those days, I have been able to complete those in the past, and with the low xss days in between shouldn’t be too much of a strain.

Also, my plan goes until March, should I attempt BT rides every so often?

This should be sorted - can you verify on your end. If you’re still having troubles, send a note to support team. Cheers!

Forecast AI is similar to a XATA phased progression using a TED but not limited to a 120-day period. Focus Duration and Specificity for an event are also configured by running the Setup Guide wizard.

One HUGE difference is the forecast plan populates the Planner to satisfy those who are lost without a plan to view. That includes a number of Xert trial users who couldn’t accept the premise XATA runs in background to guide you along the way. No need to plot more than a week at a time, if at all. Those of us who accepted that principle learned it works better than a pre-populated calendar plan. Others prefer to see where they are going.
Forecast AI gives you the best of both worlds. The calendar gets populated but you still control day-to-day selections and changes. IOW a flex plan (dry-erase board) that you can modify as desired and adapt forward as needed. Or recast at any time.
Another benefit is the XPMC chart displays the results of the forecast. It’s easy to compare your past progressions to the forecast.
Other features such as a defined weekly schedule are highlighted in the announcement post.

So yes, a game changer for those who prefer a calendar view over day-to-day guidelines.
And for those who like to experiment. :slight_smile: YMMV


The forecast “plan” remains flexible so you can decide differently as you like.
Same as using the FF slider to recommend lower or higher intensity workouts or switching things up by using Filter.

The signature calc has been updated to work with Optimal Decay as the preferred default.
You can always toss in a BT workout when feeling especially fresh and interested to confirm where you’re at whether a BT occurs or not.

We’d have to do some deep forensic analysis of the AI algorithm but in principle it’s selecting patterns and volumes that best meet the intended fitness growth, constrained by polarization but following a periodization path and preventing low intensity training training when red status and preventing high intensity when yellow status, matching all three systems as it does all this. So with all that, hard to explain exactly why. I would think that if you have days of inactivity and days of high volume in your recent past, this cycle is likely what’s needed to get the most out of your training.

If a day is missed, or done way above/below suggestion, must I click on reforcast the next day, or will it automatically adapt?

1 Like

I’m having a frustrating time generating a forecasted AI. Nearly every one results in “A training plan that reaches your target could not be found. Your target is not achievable.” The ones that do not are probably plans I wouldn’t pursue, ie I’m not going to do a specific plan for recreational events or group rides (nor do I have plan to train for CX, Track, grits, etc).

I’ve been putting in a date FAR into the future as well to give it (myself) ample time. Even hypothetical events I input that are pretty easily done by myself (and I’m pretty competitive in the second group composed of ‘group just behind the young and dumb fast guys’), that result in a targeted decrease in some of my signature parameters say my target is not achievable. Those forecasts ran will result in an increase in my High and Peak, but nothing that seems close to unattainable - especially given the long time period. Maybe the system has determined my entire physical potential is maxed with no room for improvement? Some examples attached.


And historically, those targets shouldn’t be an issue or unattainable.

1 Like

I ran into this yesterday while exploring - Forcast Beta AI will not populate a plan if I had other events/workouts or anything already pre-planned or placed on the planner. I had to erase all items off my planner and then it was working perfect after that.

I dont know if that is your issue, but take a look at your planner and make sure you dont have anything already actively on it for the future.

This may be a bug they need to fix, or may not even know about yet

Yep - fixed for me as well!

Definitely a bug and thanks for catching it. It should work perfectly with existing activities (even other races) in the calendar.

You can also try setting it up as a a goal instead of event. You won’t get a taper week planned but the training would be virtually the same, i.e. based on the final targets.

@xertedbrain ?

You need to click Adapt if you want the plan to adjust. Note, if your target is still far away, it’ll likely be ok to just keep following the existing plan, especially if you’ve already blocked your calendar for the training.

I tested it now and it’s working. Fixed very quickly. Thank you.

What is “Periodization Level” just notice that (new?) slider?