My FTP was updated by AI as 239 two weeks ago. I rode a Zwift race today and averaged 255 watts over my best 20 minutes. I thought maybe TrainerRoad would automatically up my FTP based on that to 255*0.95 = 242 but it didn’t. Should I manually update or should I just wait until my next AI detection (which I was planning to hold off to coincide with end of this phase)?
Personally I would consider that reassuring that the results are so similar, and leave it alone.
AI FTP can be updated every 28 days and takes into account all of your rides, and rest, in that time - it doesn’t adjust after each ride the way eg Garmin might.
7 Likes
Leave it as it is. Imo. Thats within the error of most power meters, there really is no difference in the numbers.
5 Likes
3 watts won’t make or break your training. 3 watts is borderline margin of error for an ftp of 240 with 1% accuracy on a good dual sided power meter
4 Likes
Did you do the full warm-up protocol for the 20 min test before the Zwift race
Because if not you most likely overperformed the 20 min “test”. If you had done the warm-up and the 20 min on Zwift was only 249 W would you lower your FTP in TR with 3 W?
I am sure most of us would say that 3 W does not change anything for the training and keep the higher FTP 
I think the other parts of the protocol are important as mentioned to get a more accurate result. Personally I averaged 293W for a 10 mile TT (21:47) on the road bike last year (95% of that would be 278W). I also did a 25 mile TT on the road bike (56:47) in the same month averaging 271W so a bit lower. Both were with my Favero pedal power-meters. I used the latter for my FTP as I sure as hell didn’t do a flat out 5 min effort before the 10 mile TT…so I think that was a more realistic value. I wouldn’t go anywhere near the ramp test for FTP testing…I muscled that up to around a 300W value in the past and that sure as hell isn’t my FTP as I’m not within shouting distance of 5W/kg 
Your FTP is not necessarily 95% of your 20 minute performance. I’d leave it as is and let AIFTP handle it.
Now, if you went out and did 40+ minutes at a value above your current FTP, I’d consider matching it.
4 Likes
Coggan addressed this here on the forum, and if memory serves he didn’t necessarily agree with the warmup protocol.
He definitely didn’t agree with .95 * 20 minutes, that’s Hunter Allen according to him.
See how long you can hold that number again, if it’s over 35-40minutes then maybe it’s your ftp.
I know the idea is FTP is supposed to be FTP but I’m not convinced one outdoor performance should lead to you instantly matching the FTP you’re training indoors to.
With a 241w FTP last year, I went out and did 261w for 46 minutes.
I set my FTP to 250w and I have to say that indoor threshold workouts became really challenging. If I’d set it to 260w at the time, I’d have failed them, no question.
That could be an indoors / outdoors thing. Some people have issues with that (I personally don’t, but for some it’s a thing). The OP’s question didn’t address or ask about that.
And that’s also assuming same bike and same power meter too which can also lead to differences.
1 Like
Definitely an indoor / outdoor thing. That’s just me in winter mode assuming everyone is training indoors atm 
1 Like
Thanks for all the responses! To be clear, this was an indoor ride with the same power source as my TR training rides (Kickr) so it is not a difference of power meters. But no I did not do the full testing protocol warm up before the race. I will just leave my FTP alone for now and wait for the next AI update. I do feel like the AI FTP is the most accurate or at least the most consistent version to base the training on. I did the ramp test in November and it put me at 260. I failed some PL1 threshold and SS rides and it adjusted my FTP down.
1 Like
I thought about saying that.
More like 92% (debatable I know) unless you are elite where closer to 95% might be about right.
Probably best to forget the 20 minute test is a thing.
1 Like
FTP is not best 20min power.
It’s ~90-97% of 20min power, after you’ve performed an anaerobic depletion effort (~5min or longer). With a fair amount of imprecision in that estimate.
So your best 20min power here is actually about in keeping with your FTP being what AIFTP says it is.
If you really think 255w is your FTP, then go see if you can do 2x20min @ 255w at a 6-7/10 RPE, or hold 255w for at least 40min. If you can, then sure, that might be your FTP.
For my money:
You’re using Trainer Road, and setting the FTP in Trainer Road, to use in Trainer Road to do Trainer Road workouts. If Trainer Road didn’t update it, then leave it alone? 
If you want an FTP you can use elsewhere, or for pacing? Then yeah futz with the data, do other tests, blah blah blah. Your Trainer Road selected AI FTP is probably good for using within Trainer Road.
You’ve bought the algorithm, might as well use it.
Who are what is this in response to?
Not the OP?
They didnt think it was 255w…
That was in response to me not having reading comprehension before I had a cup of coffee…
3 Likes
Funny.
I thought it was unlike you. Coffee 
The other bits make perfect sense, so I was questioning my own sanity. 
1 Like
Bottom line, ftp of 239 and 242 are the same. End of discussion
1 Like