How is Forecast AI supposed to know which training is best for me if I’ve always trained in an unstructured and unfocused way?
In a podcast (unfortunately, I can’t remember which one), it was mentioned that Forecast AI analyzes my previous training to determine what works best for me. And this is exactly where my understanding problem lies: What assumptions does Forecast AI make for a completely new athlete without any training history?
One clarifying question: do you mean no history of structured training, or no history of rides with power data?
If it’s the former it’s not likely a problem, though you do need some max efforts in your history so that fitness signature estimates are good enough. I think Xert personalizes how well you respond to higher vs lower intensity volume based on that history, but it doesn’t need structured training to do that as it is always allocating work into low high and peak buckets anyway
If it’s the latter, the recommendation is to have at least 3 months of history (with power data), but I’m not sure that means you can’t use it without that - happy to hear from others on that. I guess it will use the defaults for training responsiveness until you build a history. You’ll also need to assume a starting training load, and do some max efforts to get a good starting signature, but otherwise I guess jt still works, but in a little more generic way
As @wescaine mentions if you have history with power data, it doesn’t matter if it’s unstructured. All Xert needs is a general idea of your historical training load with a variety of activities on file that include some max efforts under fatigue. You can verify your estimated signature by performing one of Xert’s fitness tests from the Library. Perform in Slope mode and ride to failure points on any intervals designed for that purpose.
If you have no history (or no history with power) you can set Program type to Continuous, seed your fitness signature manually, and start accumulating a variety of power data activites for three weeks or more. You can then experiment with XFAI plans (populated calender) or consider a phased progression with XATA (blank calendar).