I think I’m learning that the AI really doesn’t want you to overdo yellow and especially red days, which is a cool insight, but can be a bit of a let down. I had a “no chain” day today with a 6.4 V02 workout in middle of a 2.5 hour ride, and I figured I might pop a red day after. In spite of feeling good, my prediction dropped seemingly because of the new red day. “Comply!”, says TR.
My AI predicted FTP went up, when I turned off the fatigue detection auto changing workouts. I’m of the opinion that fatigue detection is oversensitive to rides below LT1. That said, TR doesn’t track LT1 (or at least doesn’t present it to users)
They had said on a podcast recently that fatigue detection was just tracking TSS.
It’s then hooked into other systems, as part of automating the training plan. Meaning it’s a separate component from the new AI feature. It’s very easy to lose track of how the components are connected under the hood, and which ones are AI, which ones are not.
As an aside, fatigue detection will let you stack more TSS in a single day, than it will spread out over two days. This seems like an artifact of how the system was implemented, rather than a representation of real world recovery physiology, especially knowing that it’s just a a TSS freshness tracker.
For example. I have VO2 scheduled today, rest / red day tomorrow, then sweet spot intervals the next day without any fatigue warnings. If I add a 1 hour Z2 ride tomorrow, the sweet spot day turns yellow. If I instead drop that same 1h ride today ontop of my scheduled VO2 workout, the fatigue detection goes back to normal, and I can do sweet spot in 2 days time without any extra fatigue detected.
1 Like
Sure seems like it. For max FTP increase anyway!
Joe
+1
Use fatigue management for guidance right now, imho.