I listened to the podcasts with Andrea and found it interesting how the modelling was working. I completely agree with the scientific principles being used and then combining it with the modelling to create profiles for individual athletes. However, I am unsure about the quality of the data used in the model’s construction when the training plan doesn’t inherently produce the desired data.
I had a more detailed look at my power profile and I can see where see how it has fitted the critical model the critical power model and how it asymptotes to give a predicted critical power. But I believe my critical power is too high. I think my critical power model should be more like the red line below.
I know I don’t have data in the 30s - 8mins range to support this curve which is why the it is modelled as such. I think the curve is modelled well given the available data and doesn’t have the wider context of myself as an athlete being a twitchy hybrid. Given my phenotype, I assume my curve should be steeper than that of a more diesel phenotype.
I believe this has come about as I fail to see where in my training plan there is the opportunity to set these maximal single efforts, in the range which I am lacking, when I am prescribed a high quantity of intervals. I know I could do more testing, but then these efforts will expire after 6 weeks and I would need to repeat the testing. This then invalidates the whole principal of continuous monitoring and then how these zones can be fed seamlessly back into the plan.
Apologies for the long silence on this, and thank you so much for your thoughtful message. I truly appreciate the time you’ve taken to listen to the podcast and share your insights.
Honestly, everything you’ve written is spot-on, leaving little room for debate. At present, we aim to understand the “phenotype” of our athletes by analyzing the ratios within their thresholds and comparing them to common patterns. However, we don’t fit the CP model for different durations, as we need to keep it standardized for meaningful comparisons. It’s always a trade-off.
You might notice that we’ve recently updated the period during which a maximum effort is considered “valid”—it’s now six months. In the future, athletes will have the option to select their preferred period, but again, this is another trade-off.
Your point about the logic prompting athletes to perform efforts to “complete” the profile is excellent. As we mentioned in the podcast, the reason we don’t follow this approach is that there isn’t strong evidence to suggest that a more complete profile necessarily translates to better performance. The primary goal of the training plan is to keep athletes on track and to develop the characteristics we know are reliable proxies for performance.
We’ve only recently transitioned from an FTP-based approach to a power-profile approach, and there’s still much to be done to fully integrate this seamlessly into exercise routines. Establishing CP reliably requires a sufficient number of maximal efforts, which makes this methodology more sensitive to fluctuations in training load and regime. With Athletica, we’re venturing into new territory; this kind of practice isn’t commonly found in other AI-assisted training apps. Hopefully, as we continue to learn and refine, we’ll all progress together.
Thank you again for your thoughtful feedback. Please don’t hesitate to share more—your input is incredibly valuable!
