Plan completely changes (a bit too much) when using the Adapt button

Hey - I’m being a little bit surprised by the macro level changes I’m seeing in the plan when I use the adapt forecast button.
This of course is going to be super hard to explain/prove as you can’t ‘go back’ and see what the old one was to do an old screenshot.

I’ve use the adapt forecats button more than once since I started (I removed SUndays the other week for example) and it’s generally just moved things around a little (in that case it removed Sundays per my request). but I did notice the wee red button reappear on my planners ‘adapt forecast’ button yesterday. So hit it again.

My plan was broadly - ride Z2 for 7-10 days, do one high intensity workout (which was always the Cheung Ronnetstad (?sp) one, then revert to Z2 and repeat.
As of last night, unexpectedly (as the plan had me on rest day today and Z2 until Sat 23rd, I’m suddenly on HIT today (well I did that yeserdya and the day beforfe) so that’s odd, and then HIT again in two days … then back to the Z2 for 8 days, Z2 for 10 days cycle.

Quite a change in approach inside the current week I feel.

And of course I don’t know what it was that triggered that.

So - a suggestion - if there’s an algorithm at play, we must know what the algorithm has used as it’s decision making, so perhaps, a report should go with the ‘Adapt’ mechanism to say this is what has changed because … ?


I have seen this issue as well. Days suddenly change from low intensity to high intensity or the other way around, even if you more or less did as prescribed.

Actually, I am not convinced that Xert will give me the exact same plan if I do today’s workout as prescribed and adapt the plan tomorrow.

Also, I think it matters when you press the adapt button. If you do it in the morning or just after today’s training was supposed to start.

Below is an example showing that today’s training changed from high to low intensity (and a slight change in XSS) when doing the adapt at 9:29 vs 14:11 the same day.

1 Like

First off I don’t recommend doing this on a phone but if that’s all you have to work with at least flip to landscape mode. :wink:

You can review the before/after changes by running Adapt Forecast from the Training page.
Scroll down to the plan chart, note the magenta entries (broken constraints) which may be coming up soon or in weeks further down the road.
Next zoom in on this week and the upcoming week or two.
Select Adapt Forecast and watch what changes.
Adapt Forecast will look at your upcoming week first then scan the remainder of the plan and rebalance as necessary within whatever constraints you have configured (FAI Settings and Availability).
You can hover over each day for details and watch the change in the weekly hours graph.
To flip back and forth between the prior chart and adapted one select Cancel then Adapt Forecast.
If you agree with the changes, select Save.

Thanks for sharing @hpbieker - good to know I’m not going a bit mad.
I had kinda thought that the plan largely self adapted (but within sensible limits (again an assumption based purely on my own belief that that seems sensible)).

((I also get different info on my iphone vs the web that perhaps stems from the web interface being a software algorithm suggested plan, and the phone doesn’t seemtgo reflect that I set that on the web interface oddly. But that’s probably a different post needed))

Ta - for whatever reason never noticed that

Was that message for me or for @scoobmw2 ? I normally use Xert on the computer, but took a few screen shots on my phone to figure out when the adapted forecast changed.

It would be helpful for discussions about Forecasted Training Programs if you could include the specifics of your plan (Target Event or Target Focus Power) as well as the current & target TL’s, availability, etc.

Generally speaking, the farther away your target event/goal is, the more ‘wiggle room’ there will be in your plan, since what you do today doesn’t matter. If I’m training for a big race in 120 days, then the specifics of whether I do high intensity today (or tomorrow) don’t make a material difference on race day. Therefore, there are infinitely many different programs that still get you to the same end result… the program ultimately needs to settle on one, which is what the optimizer aims to do.

However, as your target event/goal date gets closer, or the desired amount improvement is increased, there becomes less room for error - deviating from the plan could snowball into ultimately being unable to achieve the goal by the end date. It’d be nice if we could somehow calculate a ‘program rigidity’ value that indicates how much flexibility (or margin of error) is in the user’s program.