So AI FTP ended up just about as wrong as I expected after a few weeks of 40k TT specialty and doing an hour effort yesterday largely trusting AI FTP detection + AT for a few weeks.
This is not a surprise as we know that TR ramp or AI FTP will over-inflate over real FTP for anyone with higher relative anaerobic power - but in a 40k plan, I think it really throws things off the rails.
Context was the 40k low volume specialty plan spread over a little more time, but also with weekly 10 mile TTs + 1 group ride mixed in. (which is why I stretched it out a bit) So lots of maximal 20-some min efforts, plus some shorter hard 1-5 min efforts and sprints out of group rides. TR AI FTP has landed more or less consistent with the generic 95% of 20 min max power a few weeks ago even though I haven’t hit that since.
On the good side, I thought the workouts that ramped to exhaustion over ~1h15mins (Ragged, etc.) were an interesting different kind of stimulus and mental test not found elsewhere in TR plans. And while doing these, I had my FTP auto-detected at 285 - which I knew was too high for 1 hour, but I thought, let’s see how it goes and trust the system. I largely got through threshold workouts going up to PL ~5-6 with this rating them hard to very hard, so I thought OK, maybe that isn’t the worst thing - doing the end of those long ramps plus some 3x15s, etc. a bit over real threshold could be good to maybe help raise threshold and/or make an hour at real threshold not feel so bad.
However, I had first run of the Dunderberg sustained 60-70 min series yesterday - and it immediately exposed how ridiculous that AI FTP was for me. I even pre-emptively dropped back to an “Achievable” one at PL 6 rather than the one at 6.4 it had on deck for me, gave it all I had, and was on the low side to slightly under the low target of all of the 10 min intervals. I did avoid easing up much below 200W on the 1 min recoveries, so I didn’t mind being slightly low to the interval targets - and I wanted to feel out what I could do for an hour - but net was still only 1 hr @ 262W and I had to rate it “all out - intense.” Pretty much exactly what I expected that I could do and 23W below the TR FTP.
So, we know this story well - TR badly skews their “FTP” upwards for people with anaerobic strength based on short tests / efforts - be it ramp tests or AI FTP - and then we are told to just “trust AT” to fix it all. But I’m even less convinced about that approach for a long power plan that is all about TTE - and it seems the TR system is not setup to manage this.
After that ride, adaptive training took out all the remaining iterations of the Dunderberg near-hour effort series towards the end of specialty and has replaced them with more of the long ramp type efforts at lower PLs that are a completely different kind of workout.
This really doesn’t seem right to me - if the plan was designed to transition to such sustained efforts near the end, it seems like in those last two weeks, it should acknowledge “OK, let’s scale down those long efforts closer to your real FTP, but still make sure you get used to working continuously at this FTP.” Otherwise, what was the purpose of putting that series of workouts in the plan? So I’ll likely have to replace all of these with something like the “Unicorn” series.
Meanwhile, ran a new FTP detection and it says go up from 285->287 So sure, that will work for lots of stuff in the TR ecosystem - but it will be badly wrong for steady state work - and AT isn’t even serving up that work now, despite being a 40k plan.
To a large degree, I think most people doing TTs know/expect this kind of problem in TR (see the gigantic super extended sweet spot thread…) but thought I’d share my results.
It also leaves the question for me - should I keep doing the 3x15-20 type of stuff at the significantly inflated FTP? I can usually get through them - but am I burning myself out doing that since they are in fact way over my real threshold. I don’t know what the right answer is, but I’m pretty sure it’s not “just trust in AI FTP / AT…”