How is progression determined, how often is the prgram updated and program continuity error


Hi there!

Couple of questions that came to mind over the past 4 weeks of usage:

1) What metrics are used by the ML algorithm to determine progression? Is it a weighted formula comparing different workout stats over time?

2) How often is the program updated by ML analysis? Is it limited to once a day and after new data has been added to save CPU cycles? How many future days are modified when the program is updated?

3) I changed my training focus after 3 weeks from half-marathon, train to maintain to 10k with an "A" race in December. This weeks' sessions didn't seem to consider last weeks' sessions. So far, I haven't run 3 days in a row. This week, however, began with a run while last week finished with 2 runs in a row, including my longest run to date. I re-arranged the schedule to have a day off, but I thought that's one of the things the algorithm would take into account.



(1) we track markers of your load and use a modified version of Banister's 1971 classic TRIMP paper.
(2) this is updated at midnight each day after which any changes are made out to 10 days ahead (and not further).
(3) first - always modify in accordance with common sense. You know better than Athletica or any other training platform. Because you've mixed plans here we have the case where we can see some inconsistencies between them and we will likely go back and adjust accordingly (thank you!). We have checked your situation and from the loading perspective the 3 days in a row run was acceptable on our end. Again however, if something doesn't feel right in your body you should always change it up to correct that, so your actions were best practice.