My simplified private take is that it’s the maximum hard effort you can sustain for an extended period of time without the lactate bursting through your ears. 
1 Like
And also an FTP test, or hour power, is still subject to variables any given day/ session. AI FTP at least takes into account at least 4 weeks of work.
1 Like
This is a very interesting point that I’ve not seen made before. 
It certainly has the potential to smooth out the effects of the baddest of the bad days and the goodest of the good ones.
1 Like
Tested 330W; AI 336 W
Previous tested FTP 323W
1 Like
For me at the beginning of the year it was within 1watt of both my Garmins estimated FTP and a 8min test @90%
Garmin = 256
8Min = 258
AI FTP= 257
That was with a right only crank PM
Now that I’m running dual PM I can see that my right leg was /is consistently 10-12% lower (55/45 LR Balance) probably due to hip operation.
Now that I’ve recently changed power-meters ( reading 13% higher) , I’m giving TR some time to self correct.
But at that point in time AI FTP was certainly incredibly representative of a 8min test for me
3 Likes
I really don’t know, but I just don’t see dismissing it as not valid on the basis of a test is really right because of the human variables.
1 Like
AI: 270 - which I thought was too high, so tested
Ramp-test: 270
Impressed
3 Likes
better do a 20m test too to triple check 
My second ramp test (Feb 4) had me at around 175, and then I did an AI test only one workout later and it gave me 188 and I’ve stuck with that for the second block and it has felt right, at least with threshhold workouts, which have been “hard” but doable. I did the AI test again today and it estimated 197, a nice surprise but I decided to stay at my current ftp since my workouts have been getting harder, and Chicoma -5 tomorrow looks brutal, way harder than any SS workout I’ve done.
1 Like
Don’t forget that when you accept a higher aiFTP your progression levels will likely drop. And then AI will shift future workouts. So what you are/were looking at probably will change.
1 Like
I tried to do that by hand once by looking at AI FTP’s suggestion and then doing a ramp test. But AI FTP’s number got into my head and I couldn’t finish the ramp test mentally, I pulled the plug after 15 minutes (i. e. this was not a “ramp test overestimates FTP” problem).
I’d love to have a blind test feature where I could compare it after the fact, but I can see that this opens a whole storage room full of cans of worms.
1 Like
After long term usage of this AI FTP, I’m pleasantly surprised at its accuracy. For many reasons, my FTP over the winter and after a few months of base and rebuilding, about 2 weeks ago, TR AI FTP estimated my FTP @ 224.
2 days ago, I did an 1 hour effort, seated, in the drops, 2.4 km flat road, with circles on the ends so I had to slow down and my NP was 218 and I still had a bit left in the tank so for sure, another +6 watts on average was definitely doable.
I was surprise this was even more accurate than intervals.icu’s eFTP which for me, used to be pretty accurate. Kudos to the programmers at TR!
Now only if they can starting counting my outdoor rides…
3 Likes
Power was 216 and NP was 218. Intervals.icu had me at 200 about 3 weeks ago and have been consistently lower than TR by about 20 watts over the winter training. I have been doing my indoor workouts based on TR FTP and the level of exertion seemed to be correct.
I did polarized with various levels of endurance from 50-70% and VO2 Max. nearly all sessions were 90 min with a few VO2 at 60 minutes. Anywhere from 4 - 7 hours per week. No max efforts but a few VO2 sessions were very hard. I also did a lot of walks, nearly daily and a few eMTB rides tossed in.
1 Like
FTP ≠ hour power, so your FTP is likely higher than 224 W. 200 W is way too low if you can do 8 % more than that and had some breathing room left.
Oh, I thought FTP was hour power… So if you do a ramp or 20 min FTP or any FTP test, then the calculate result should be what a person should able to output for 1 hour.
I have used intervals.icu for a long time now and it used to be pretty close to my ramp test but only in the winter was coming in like 20 watts low…
Well, now that it’s outdoor riding, which TR AI does not take into consideration, and I won’t bother with a ramp test so I’ll use 224 as a guidance and ride by RPE.
1 Like
Defining FTP is hotly debated around here, many people do say FTP = 1hr power. Others say FTP is really the power you can hold from somewhere between 30-70mins, others says its really max lactate steady state. Calling FTP 1hr power is reasonably close, if not 100% true.
If you’re not doing max efforts, intervals.icu won’t give you an accurate estimate of FTP because it bases its estimate on your max power over a certain duration.
TR AI does take outdoor riding into consideration when determining your FTP, but doesn’t currently adjust your progression levels based on outside rides.
I think you’re on the right track with using 224 as your FTP and keeping track of your RPE while riding at that intensity as well.
3 Likes
A major flaw in all of these discussions is that FTP is, by definition, operationally defined. The fact that you get slightly different numbers using different testing protocols is expected, and not proof that any given method is superior to another.
The real crux of the matter (IMHO) is “does any given testing method give me a number that is useful in guiding my training or conducting my races”? If the answer is yes, it doesn’t matter which method you used.
Races are won by the combination of fitness and strategy, not by the number you assign to an arbitrary concept.
An analogy: We talk about intelligence, and we all know basically what it is, but when tested in various ways, we get different results, and when we try to define it, it gets increasingly slippery and hard to pin down. So where it matters, we rely on an operational definition - i.e. a score on a specific test - and we understand that other measures may not give the same result. We also understand that the score on that test is only one of many factors that will determine any other factor we might want to measure - academic success, business success, problem solving, writing ability, etc.
FWIW, the AI FTP detection in conjunction with the TR program has consistently and reliably given me workouts that are a challenge to complete and are difficult but achievable. The fact that my FTP might be 10 watts higher or lower tested some other way is irrelevant if my training is going well.
7 Likes
Thanks BenB,
Interesting about the various ways to define FTP. I get the 1 HR power and the LT but to just pick a point in time between 30-70 min seems quite arbitrary.
Also what do you mean by “max effort”. Do you mean one hour all out?.
In the past year, for indoors TR, I’ve been sticking with Polarized training. Lots of endurance and some VO2 rides, both at 90 to 120 min sessions and I rarely ever do any other zone rides.
When I’m outdoors it is mostly 90 to 120 minutes of upper end Z2 based on RPE and I’ll do the last 15 to 30 minutes in VO2 to anaerobic power. And I’ll also do 1 hour sessions of just VO2.
Once each summer I’ll do an an all-out 1 hour session but my main target has been max distance in one hour so I put a lot of effort into being aero. I think I can probably have a slightly better hour power, FTP, if I wasn’t in the drops the whole time. Hard to say, maybe I’ll compare next time. Currently I’m training nose breathing and even that has an affect on performance.
I don’t race and at my age I’m just happy to set simple personal goals.
No. Only a minority of people can hold their FTP for an hour.
FTP is the power at lactate threshold measured in a field test. Field test means that instead of having to measure lactate levels while riding, you do some other test that lets you infer the power at lactate threshold.
The reason people started calling it hour power was that in Coggan’s initial research (= the guy who came up with the concept of FTP), his demographic were highly trained road cyclists who would do TTs. For this select group, there was a strong correlation between the average power they could hold for one hour and the FTP.
I don’t think so. Holding your FTP for close to 1 hour requires special training and/or aptitude. Someone posted a study recently, even among professional cyclists, the average time they were able to hold their FTP was around 50 minutes if memory serves, and only a minority of pros were able to hold it for longer than an hour. The other categories (such as ambitious amateurs) tested worse on average.
(If you’d like, I can look for a link to the study. I just have to head out very soon.)
This is super important: max power ≠ max speed in many circumstances. Many cyclists forget that being fast = high speed, not high power.
3 Likes