Easy. If you know what your actual FTP is, and both old/new AIFTP matched / match that.
Or, ability to complete similar benchmark workouts at set FTP gets you close too (e.g. - how hard is 2x20 with 5 mins rest….)
Its not FTP so not sure how it can be marketed as such.
Mine was within 3 watts on release and could believe both numbers. 3 months later its 30 watts higher and predicted to be 43 watts higher in 26 days. C’mon, 43 watts in four months 
1 Like
Great progress @Jonathan , I am pretty sure you strength train as well, but I noticed there is only TR workouts in your calendar. Is there a reason not to have all activities sync into TR for a better picture of activities? I started removing any TSS associated with my Strength workout so that the AI would stop making my workouts shorter or easier. my level of strength training is not impacting my cycling workouts and I was getting frustrated when I psych myself up for big threshold/VO2 workout only to find I have been given an endurance ride because I did arms at the gym in the morning…. is this why you don’t include these workouts as well? thx
Sharing this post as it is relevant to much of the discussion here:
@Wilsot3 I haven’t been strength training intentionally like I should 
. Part of my daily routine is basic push ups/push ups/lunges, but I’ve only just started doing specific strength workouts last week. Need to be better!
The approach you are talking about is one way to do it, but I’d rather have the workouts there on your calendar so it knows what you are doing. If you feel like you are fine and able to do the structured work, then click on the monopoly style card on the workout that was changed and revert it. Do the work, fill out your survey honestly, and then that will give valuable info to the model.
6 Likes
I just don’t include my sets in my calendar. TR only suggests logging sets to absolute failure, which is not how most people (should) train anyways.
6 Likes
Just my 2 cents. To me FTP and AIFTP are not confusing they are just rough indicators of how you are doing. So if I’m training hard and the numbers go up then I’m doing well. When the numbers go up then the next weeks training is harder measured by RPE. After a few weeks you adapt and it goes up again. Jonathan did some rough calculations that he needed X additional watts to finish in under 7. While it’s interesting to do the calculations for fun they miss a lot on nuance, make assumptions and are just rough calculations. In the end he will train hard and his body will adapt at the rate it can and he will get the time he does.
1 Like
Convergence achieved: AI FTP, eFTP, mFTP, and OG 20-minute test are all within 1 watt 
.
4 Likes
As a CO resident I’m very interested in this as well. I adjust FTP up/down depending for significant changes in altitude, mostly to keep my PMC reasonable.
2 Likes
Why not throw a ramp test in just for fun? 
2 Likes
That’s like a planetary alignment that only happens every 600 years… Whatever efforts you are doing, it must be something that all the models can lean on.
My eFTP is 40 watts lower than my AIFTP. eFTP is pretty crude and seems to lean on shorter efforts for it’s calculation (which I seldom do maximal efforts on), while AIFTP seems to be able to interpret an FTP from multi-hour efforts (which is really the focus on my training and hard efforts). My “real” FTP sits somewhere in between, pretty much splitting the difference.
4 Likes
I have the minimum duration for eFTP set to 10 minutes, maybe that has something to do with it, you can set it up to 30 minutes, I believe.
2 Likes
Yeah, the problem with eFTP isn’t the minimum time it’s looking at, it’s that it doesn’t appear to recognize stuff over an hour. If I go do an hour at 280w, it will bump my FTP up to 280w. But if I go do 3 hours at 280w, it’s not going to estimate/raise FTP further base on a reasonable IF conversion. Just like you can estimate FTP from short efforts using a percentage, you can do the same with multi-hour efforts. And in many ways, using a longer effort is better because the anaerobic contribution becomes totally irrelevant over longer durations.
Since I rarely do short maximal efforts (and short stuff is a relative weakness for me), eFTP will dramatically underestimate my FTP. I will eventually get into longer maximal threshold stuff in the later stages of build and that will get eFTP pretty close, but still a bit short of what my FTP is. TR’s old AIFTP model would get really close, the new AIFTP overestimates for me by 15+ watts. Part of that I guess is the “normalization” at workout level 3. But I also suspect the TR dataset that AI leans on is not as established for higher volume training and is giving too much credit for longer efforts at higher IF’s.
2 Likes
@Jonathan Congrats and great work! Would you mind walking through how you arrived at 320w as a goal for your FTP? Not trying to add fuel to the FTP Wars but would be interested to hear how you’re using the new FTP model to determine bench mark goals and pacing strategies since we’ve diverged from the “FTP = 1hr Power” paradigm. Thanks!
1 Like
Pretty sure he uses best bike split to determine the normalized power he needs to hold for the duration of the race (based on target weight, crr, and cda).
Then dividing by the IF he thinks he can hold for 7hrs gives him his target Leadville FTP which can be scaled up to Reno elevation.
2 Likes
Agree to a point, but Leadville is more of a math problem than most races. There are certainly variables with wind, surface conditions, drafting, etc. that can swing finish times by a bit, but the climbs are long and predictable with a very high degree of accuracy. If I look at my strava history after riding the course 25+ times, the climbs are just simple math problems translating watts/weight into time. The first time I did it, I did a bunch of strava stalking to figure out planned power targets and split times for the key segments. They were dead on. Within seconds for many segments. The hardest part for many isn’t translating watts/weight to time, it’s estimating/adjusting the power you can do at altitude. If you’ve never raced at high altitude, you just don’t know your physiology until you do it. But once you have the experience (like Jon), that just becomes part of the math. Some bike races are math problems (particularly time trails), and that can be an effective way to approach Leadville.
3 Likes
Wait. What? Who’s living in a fantasy? I’m not doing the race. I think that’s how Jon’s approaching it based on his comments from the podcasts. Of course one must make estimates and can’t control everything, but FWIW, I think the approach I described is the most sensible and rational one.
My “fantasy” aside, how would you approach the problem?
2 Likes
I’ve used bestbikesplit for gravel races where it got my finish time down to a few seconds. These are races that are 4-6 hours long. Of course you have to have a good handle on cda and crr in order for the data to come close, but it’s pretty dang accurate.
2 Likes
Patrick nailed it.
It’s not exact, but if I am training and adapting to the point where my FTP would be 320, that would put me at my best power and should allow me to maintain the sort of power needed to go sub-7, assuming my weight is around 148lbs and all other things go well.
8 Likes
Enjoy your self! Jonathan, Chad, and Nate were by far the best of the podcast days!
2 Likes