I have the impression that the AI-estimated FTP is too high in my case.
For example, it estimated my FTP at around 280 W. However, in a race slightly over 1 hour, I was riding in Zone 4 with an average power of 214 W (NP 228 W).
Is it possible that TrainerRoad tends to overestimate FTP in some cases, or that the estimation is simply not well aligned with real-world race data?
Would you rate your race as All Out?
TR’s AIFTP is not meant to replicate however you measured FTP in the past. It is a number they use to assign workouts where you will rate a level 3 Threshold as “Hard”. I’m not defending that, simply explaining why the numbers may be drastically different (mine is too).
Have you done any sweetspot or threshold workouts recently?
If so, what sort of wattages/interval lengths/total time in zone? How was your RPE?
Good point. I would say 90-95%, not all out. That explains the AI-real world difference partially.
True, I also really like the suggested sessions based on the AI FTP and self-assessment. They are hard, but rarely too hard. What I would love to see is a closer match with my real-world FTP. That would allow for better race pacing, like with Best Bike Split power plans.
A while ago, I did Goddard -4. Sweet Spot with 6-minute intervals at 234–243 watts. That felt somewhere between moderate and hard."
Yeah, unfortunately that workout is pretty useless in terms of trying to assess where FTP lies.
What kind of race? MTB Gravel Road, or Crit?
Zone 4 HR or Z4 power?
Why/how were you racing in zone 4? I can see this happening in a crit or a long gravel/road race with a pack, but in an hour race, something is up. Either, you werent racing, or you were with such weak competition you were able to draft and conserve a lot, which leads me to believe your FTP is far higher than your race data.
FWIW, my MTB races Average around 210, Normalize 245-260, HR would be 176 and Z 5 but if I were to do a Z2 workout, the average would actually be higher but HR may average 137 and middle Z2.
Welcome to the forum! 
I’d be happy to take a closer look at your recent training and let you know my thoughts. Can I share some of those details here in this thread?
If your workouts aren’t feeling harder than they should, that’s a good sign. When was your last threshold workout, what level was it, and how did it feel?
Hello Greeninaustin,
It was in HR Zone 4, which for me is around 152–155 bpm. The race was a team triathlon over the Olympic distance. I only did the cycling leg, riding a road bike with clip-on aero bars; my teammates did the swimming and running. The bike course was 40 km long and somewhat hilly. Drafting was not allowed.
Hello Eddie,
Thanks for your message. Yes, of course, you can share my training information here.
On June 10, I did the Whaleback session, and it was very hard for me.
Based on all of the hard workouts you’ve completed since the beginning of May, I’d say that your FTP has been in a place where your training has been pretty productive.
Those workouts you’ve rated hard / very hard (Whaleback, Border, Denali, Nagar, etc.) are supposed to be pretty difficult. 4 and 5-minute suprathreshold intervals are always challenging for me. 
Your training plan isn’t super typical, which could have some influence here, too. You’re only doing two hard, structured workouts a month, and typically riding twice a week, which might be making it hard to get a good grip on your fitness.
Also, there are some recent outdoor rides that don’t have RPE surveys attached. I’d make sure that you’re filling those out after each ride. They’re a really important part of analyzing your fitness, especially when those unstructured outside rides make up ~50% of your riding volume.
One more thing that I noticed is that you regularly have days where you log around 20 short cycling activities. What’s going on with these? Are they actually separate rides? 
Let me know your thoughts here!
Hi Eddie,
thanks for the deep dive and checking out my career!
To clear up the mystery about those days with 20+ short activities: those were actually aero testing sessions on the road! I was tweaking my position and trying out some equipment optimizations, which is why there are so many short files.
Regarding the RPE surveys: I actually went back and rated those aero test rides as “Moderate”. Interestingly, doing that caused my AI FTP to drop from 280 to 276 right after.
For upcoming race pacing, I’m actually thinking it might be a good idea to do a classic FTP test soon. Just to cross-check the numbers and compare the AI FTP with a real-world test result to see where I truly stand.
That being said, if the AI FTP turns out to be pretty spot on, it just proves how incredibly efficient the TR program is, especially considering my volume and the fact that I only do a couple of hard structured sessions a month.
No problem!
Yeah, pacing can be tricky. Depending on the lead-up to your race, your power curve might give you enough info on what you’ll be able to do on race day. You’ve raced recently, so that data might be really useful for you. 
20-minute tests can add useful info, but keep in mind that you’re making assumptions there about 60-minute power, which isn’t always perfectly accurate, so taking all of that in with a grain of salt and trusting your gut with RPE can make a big difference. 