Not surprising that people get burned out on sweetspot-heavy training plans if they’re actually riding at threshold.
Ooft!
2 Likes
Yeah when I looked at her workouts, I was like, “Well no kidding you can’t do these.” Usually she would get through one of them after a couple of days off, and then the rest of the week would be a complete mess. It was kind of unbelievable, Zwift had her at 220W! She tested 20 minutes and came in just under 190.
She just did 3x17 at 180-185 this morning as a threshold set and nailed it. Her consistency is WAY up now, she can get through her workouts and her whole attitude towards the bike seems to be super positive now.
6 Likes
Interesting approach, thanks for sharing.
I think that might be the case for new athletes especially, although it doesn’t just hold true for the ramp test. E. g. if I were to switch to the 20-minute test, I might undertest at first simply because I’m not used to the effort. However, for more experienced athletes, I am not convinced that this is true: they should have a good handle on their preferred FTP test format.
I can see both sides here. Compared to my peak last season, my FTP changed by only 26 W, and changes in fitness can be subtle. Still, I like having a handle of FTP (together with other performance metrics) to better gauge where I am and to anticipate next season what kind of gains I can expect. I’m definitely a bit more of a data-centric person, so this approach need not be the right one. A fixation of FTP could be akin to a fixation on weight, i. e. ultimately harmful to the athlete.
I’m wondering why TR chose the Ramp test as opposed to the Kolie Moore 20 minute test, or the 2x 8minute test (Stephen Seiler?)
I’m going to try both of these and compare. Making sure I get well fuelled up and caffeinated before I start.
1 Like
TR cares less about your “true” FTP/LT2/MLSS/whatever and more about setting training zones that will progress your training without killing you. In their data, the ramp had the best post-test outcomes.
(They’ve talked about it in podcasts before, I specifically remember Jonathan emphasizing how much of a data-driven recommendation it was, but I don’t remember which podcasts, sorry!)
3 Likes
I’d just keep going back to more experienced athletes probably don’t need to test for FTP specifically. I think this is the nuance where you and I go our separate ways. I don’t test people specifically for FTP. I derive FTP from other tests and workouts.
100%. The reality is FTP doesn’t move very much over the course of a season. As you said, 26W in a season. You probably could’ve tested twice all season and been fine.
Now, you kind of have to know the progression here: take your offseason break off the bike, test your FTP or do some kind of effort to see where you are. After about two months of riding, FTP is probably up.
After a big VO2max block, FTP probably changes once you recover from it. Otherwise, you can usually tell when FTP has moved during threshold workouts, and that can be a trigger where you go and test to validate what you’re feeling.
So following that… you can “test” or look for your FTP to change 3-4 times in a season, and that’s about it. Meanwhile, this forum is littered with posts from people lamenting that their monthly ramp test didn’t show an FTP change and “what am I doing wrong??”
Here’s my progression this season:
July post-injury: 255W.
September after initial base: 280W
December post-VO2max block: 295W
It literally doesn’t matter if I was training with a 255W FTP instead of a 265W FTP for a few weeks there because I was extending TiZ anyway. So was I extending TiZ at 90% or 88%? Who cares? It just doesn’t matter that much at all.
6 Likes
I never said I don’t test them. I don’t test them specifically to find their FTP. I want their whole power profile, not just one number. They test more often than monthly FTP tests, but at different durations.
Any mathematical formulaic estimation isn’t accurate. A 20-minute test is more likely to give you a better FTP result because you’re at least sustaining power longer than a minute and then multiplying by some factor.
You assume incorrectly. A ramp test the way they execute it isn’t a great measure of MAP, for one. And the 75% “rule” isn’t a rule at all. It’s a guesstimate that kinda works for some segment of the population. TR picked the ramp test because it was easier and gave what they think is a reasonable estimation of a number about which they can set your training zones. Whatever that number is probably isn’t actually your FTP.
1 Like
Their assertion was that the ramp test was the best of the options they had at the time (ramp, 8-minute, 20-minute), in terms of both test execution (i.e. can/will people actually do the test) and future training outcomes. I don’t remember any discussion of bell curves when TR was talking about why they chose it. (The development of the ramp test as a tool for FTP estimation absolutely had lots of analysis like that around it, and lots of argument over whether 75% is the best default when the original data showed 72-77%, and what kind of cyclist it even works for, etc., but that’s a whole other conversation.)
If you’re thinking within one standard deviation being a majority, that’s 68% of people being within on sigma of “bang on FTP”.
TR has no interest in correctly estimating your “bang on FTP”. In their best-case scenario, they set up training zones that work for you, and you get workouts that push your capability in a sustainable way. They’re not trying to give you a wattage you can hold for 40 minutes, so they’re working with a much larger margin for error. Workout levels gives them even more wiggle room.
Anyway this topic has kind of been litigated to death in other threads, I don’t want to get too far into the “does FTP even matter” debates again 
1 Like
Because it is the easiest to explain and do: ride until you can’t anymore. You don’t need to have any idea of your target FTP and pacing ability is also not part of the equation.
The KM, 20-minute and 8-minute tests require you to have at least a rough idea what wattage to shoot for. And you need to know how to pace yourself for a given duration.
I’m fairly certain TR has internal data on how well each of the tests (sans KM’s protocols, of course) works, and I am certain they would have reverted back if TR had found that the ramp test worked less well than the 20-minute test. They are a very data-driven company. I think they have talked about it in the past on their podcast, but I don’t remember hearing numbers (for obvious reasons).
If you get used to them, I reckon either will work for you. In all cases you should validate the test result with additional workouts. I mainly stick to the ramp test, because it has worked very well for me and it is not a part of my training I see a need to play with. However, I encourage you to try things out rather than discuss them in the abstract.
You mentioned earlier that you have your athletes do something akin to the 4D power test (best power efforts at various durations, including 20 minutes). It sounds to me as if you are testing them, too, just not for only their FTP, and less often.
If all that I checked for during ramp tests were my MAP and FTP, yes. Funnily enough, I found another very important use case for ramp tests thanks to you — checking my fatigue. I find the heart rate in the last minute or two of a ramp test is an indicator of fatigue. This is really something I am thankful to you for: you explained to me that my low heart rate might be a sign of fatigue (as opposed to fitness).
I could do that on my outdoor rides, but my life currently permits one outdoor ride per week (apart from commutes), and most of those are endurance rides. (The biggest challenge in my training is to marry training with family life and work.) So regular ramp tests give me an opportunity to keep an eye on my fatigue, too.
I think TR has asked a different question when they introduced the ramp test: does it work better than the 20-minute test for the majority of athletes? It seems to me that TR’s data answered that in the affirmative. As the data is proprietary, we have to believe in TR’s claims. And now they want to replace the ramp test with something better still.
One aspect that doesn’t seem discussed a lot is whether the ramp test performs worse, statistically speaking, than e. g. the 20-minute FTP test. We have had some discussion about this before, but I don’t see any obvious reason why the 20-minute test is more accurate. If it is taken “correctly”, i. e. the athlete has figured out how to pace properly and gets close to its physiological limits, then the FTP-to-MAP ratio will still play a role. Put another way, is the “-5 %” rule from the 20-minute test any more accurate than the “FTP = 75 % MAP” test? I don’t know, but I haven’t seen clear evidence either way (apart from anecdotal evidence).
The only slight indication is that TR has stuck to the FTP = 75 % MAP rule throughout. I assume that TR has checked this as well, and e. g. compared completion rates of certain benchmark workouts depending on whether the FTP test was performed with a ramp test or a 20-minute test.
For people who are new to the sport, you might argue that because they are not as good at pacing, they will likely undertest, which is safer than overtesting. That’s an argument, although I don’t know for how long it is valid. Another equivocal factor is pride: I think a lot of people who overtest, especially if they overtest slightly (meaning their FTP is < 75 % MAP), are too proud to adjust their FTP downwards. If these people had a coach, e. g. you, then the coach could override them. But since TR is a self-training tool, you have to rely on the athlete coming to this conclusion themselves.
FWIW I did not find AI FTP to consistently work well for me. Its FTP estimate after a 4-week hiatus (a vacation in Europe) was completely useless (way too low, it would have made “threshold” into easy sweet spot and sweet spot into easy tempo). Although I have to agree that my body’s reaction was very unusual and unexpected even to me. During a season (with more recent ride data), it was spot on. Since I really like taking ramp tests, I have decided to simply go back to that.
I do agree, though, that this is a good development for the majority of riders. I could see a future where you can ask athletes to do only 2, 3 ramp tests per year and do the rest via AI FTP.
Never said I don’t test. My athletes probably test more frequently than TR users, but I’m interested in their entire power profile, not just FTP. What I’ve said is I don’t test specifically for FTP, and I don’t use these FTP test protocols to generate FTP estimates. If I want to know their threshold, I see what they can actually do.
TR makes their decisions based on what their athletes like and will do, as long as they think it’s “good enough”. Their decisions aren’t based on what is most accurate. They want to retain subscribers.
This is why something that grossly overestimates FTP is “dangerous” for self-coached athletes, and why people like me can still get clients even though TR is aiming to make us obsolete. 
I have no experience with AI FTPD. Seems neat, but again… I wanna know what I (and my athletes) can actually do.
Well said, your eyes are open. Essentially they’ve built a system that is intended to keep you training with them. Nothing wrong with that, I own two businesses and I get it, too. I recommend TR for a lot of people who don’t want a full time coach.
2 Likes
I think that’s pretty cool. Perhaps I should look into adding that next season. Right now I am struggling for the second season to integrate strength training into my training schedule, and I want to focus on that first.
Customer retention is important, but I think in their mind customer retention is a consequence of consistency and athletes not burning out. As far as I can tell, the incentives of TR and its customers are fairly well-aligned.
From what I can tell their decisions are guided by the assumption that consistency will be more important in the long run than things like “optimal” workout selection. This has led to decisions like replacing a weekend endurance workout with an easy sweet spot workout in some of their base plans. I don’t think they are guided by “good enough”, but by kaizen, many small, incremental improvements over years to get a substantially better product.
So I don’t think that TR’s reasoning is flawed, it is just weighing some of the training fundamentals differently than other people would — which is completely fair. I don’t always agree with the outcome either (e. g. I always replace the Sunday sweet spot workout with an endurance ride), but I find their approach logically coherent and consistent.
My problem with TR’s data-centric approach is that they are simply not collecting data that is likely very relevant to training, because they don’t know what to do with it now. And what you don’t collect, you can’t analyze, it is a catch-22.
One thing the TR team has responded to is the utility of doing analogs of the “4D power test” (essentially what you do with your athletes), where you probe best powers at various durations. In my mind this shows another limitation of automated approaches: how do you select the right goals for an athlete? How do you distinguish between weaknesses and limiters? Some people might have the expectation that TR (or a coach) should focus on everything (I always picture Bruce Almighty selecting all prayer emails, replying to all and responding with “Yes.”). This is definitely something where the human element of a coach can be really beneficial.
Another one is left-right balance, I’d really love to have a record of that, and I think for people coming back from injury or so, they’d appreciate having the data available to them. A third and very obvious one is sleep data: TR currently does not integrate with sleep trackers and sucks in HRV data. I’d love for AT to suggest easier workouts to athletes when they have slept e. g. 1 hour less than normal. Competitors are eating TR’s lunch, albeit slowly.
Not sure that this is the goal of TR. They said that they want to replace coaches when it comes to the task of selecting workouts, which is a small sliver of what coaches do.
In my mind, coaches help athletes figure out what goals they should pursue, decide on a volume/intensity that is suitable for the athlete, find out ways to find suitable metrics and help interpret the results for the athletes. They support athletes mentally, they take constraints into account that you cannot possibly put into software (I reckon you have hard the “Don’t get me divorced!” from one of your athletes). They educate athletes (I have learnt a lot from your posts and messages in the past 
).
This is v1.0 of AI FTP, and it is likely an improvement, especially for people who hate to test. (Quite a few people viscerally hate the ramp test, because of how it makes them feel.) So overall it is likely a plus for them. I reckon that the best combination for the largest number of people is a combination of infrequent tests and AI FTP. In the beginning of a season or after a longer training pause (>= 3 weeks?) I’d re-test to give AI FTP/eFTP/whatever-you-call-it a solid baseline.
Correct, although I’d say it is less about wants, but just what people can afford. If people had the choice between TR and hiring a quality coach for $20/month, the overwhelming number of people would pick the coach. I’d hire the coach, no question about it. Thing is that most coaching services start at around 10x of what TR costs. And this was and is TR’s business pitch: get 75 % of what you can get with a coach for 1/10th the money.
But this is also an achilles heel: even if a training plan/idea works for, say 90 % of athletes, perhaps you are part of the 10 % that it isn’t suited for.
I think you’re right on a lot of this: TR’s decisions are based on getting maximum consistency. That’s a good training principle, but the reality is consistency is also how they retain customers. I have zero doubt that is the driving factor behind their decision making, as it should be for any successful business.
I think it’s a mistake to assume TR makes all their training plan decisions and such on the basis of optimizing training. So saying, “well, TR says a ramp test is good” doesn’t prove anything with respect to training. That’s all I’m trying to say here. Same with AI FTPD. That TR tells you it’s good doesn’t mean it’s optimal. Is it good enough? That’s up to the user.
I’ve said this before, and I always caveat it with the fact that I like TR even though I don’t use it myself any longer and haven’t for quite a while: TR’s company motto is “Get Faster.” It is not “Get as fast as you can be.” Without question, most people who are out there noodling around or doing exclusively group rides or whatever will get faster with TR, but that doesn’t mean TR is remotely close to optimizing training.
I think a pretty decent use case for a high level athlete who knows enough about training themselves could be to use Train Now, picking 60min-90min interval workouts twice per week and supplementing those workouts with additional 1-2hrs of volume at .55-.6IF. Then riding 8-12 more hours of low- to mid-zone 2. Do blocks of “climbing” and “attacking” style workouts. Rotate those throughout the year (you might do six blocks of “climbing” and two blocks of “attacking” as an example) based on some kind of periodization, and it’d be a decent plan. The key would be making sure the workouts you decide to do fit with the intent of your training block. And the drawback would be you wouldn’t get an effective VO2max stimulus so you’d probably want to do a dedicated block of that outside the TR spectrum.
The side caveat to that would be having an accurate FTP to start with… actually this would be something I’d be interested in trying myself to see if it’s something viable that I could recommend to people who don’t want to pay for full service coaching. I just have a hard time recommending 3x interval sessions per week to most people who can ride more than 6 hours a week.
2 Likes
Yep, this week I’ve totalled 2.5 hours of intensity sessions and 10 hours of Z1/2. The intensity sessions being two sessions, 72 hours apart. No way will I put a third intensity session into the week.
1 Like
@kurt.braeckel
I agree with your points, so let me just add one or two aspects.
We also shouldn’t forget who the bulk of TR’s customers are and what fitness profile they have, and the people who will likely run into issues with TR’s approach.
If you joint TR and you start at, say, 2.5 W/kg and increase your specific FTP to 3.5 W/kg (derived fro the specific MAP). Now (specific) FTP/MAP isn’t the be-all-end-all metric, but for beginners, it will correlate so strongly with all other metrics that it is a good proxy for the net improvement in fitness. If you increase your specific FTP by 1 W/kg, you will be a lot fitter and faster in every conceivable situation.
Basically, in my mind the main benefit is that you are following a training plan with structure — any training plan will likely work if intensity and volume are doable for the athlete. IMHO it won’t matter too much whether you go for polarized or sweet spot base, the outcomes won’t be the same, but with both you’ll see increases in fitness. The key to success for TR (both, monetarily, but also in terms of its mission statement) is to get these athletes hooked on structured training, especially structured training TR-style.
In my experience, the next step in fitness comes when you address sleep, nutrition off-the-bike and nutrition on-the-bike, and get better at gauging how much volume and intensity you can handle and should pick for your training plans.
The higher you go, the fewer customers meet these criteria, and the more you are an outlier in terms of TR’s statistics. You need more and more individualized metrics that take into account how you want to express your performance. This is another potential pitfall of a data-centric approach, although one that TR is trying to address with its ML algorithms that aim to personalize and adapt your training plan.
I’d also add one piece of caution with regards to coaches (such as yourself) making conclusions about TR: I strongly suspect that there is a huge self-selection bias amongst people who hire coaches. They might be much more willing to do “boring things” like an endurance workout on the weekend rather than a fun sweet spot workout. (“They eat more vegetables and less meat.”) I reckon for this crowd what is physiologically optimal can be much closer to what is optimal for this athlete given the circumstances. TR with its different user base (on average) might need to make different decisions here. (I’m sure you are plenty aware of this, I just thought it deserved mention.)
IMHO you could even think of a third scenario where a coach works with an athlete who bases its training on a TR training plan. If I were to hire a coach, I don’t think I’d need to completely revamp my training and whether I use TR’s vast workout library or have a coach recreate it in e. g. TP, it wouldn’t matter much. The sad thing is that TR doesn’t have any good tools in that respect, i. e. it doesn’t have a pro/coach tier as TP does.
I think the value of a coach would lie in other areas:
- Decide on goals at the beginning of a season and how to reach them.
- Find metrics that are relevant for the athlete and monitor them across the season.
- Monitor londg-term fatigue and adjust intensity and volume accordingly. That one is huge, because it is much easier to do this from an outside perspective.
- Ensure consistency, and give encouragement and motivation.
- Monitor nutrition if needed.
- Pick up on smaller things such as “your average cadence is a bit on the low side, time for more cadence drills”.
2 Likes
Those are your classifications of boring and fun for different types of rides / workouts. You have to realise your fun is another’s boring and your boring is another’s fun.
This is effectively me. I have a competitive athletics background of over 15 years, am a sports scientist, and in a previous life was a coach. My FTP, as determined by AI FTP, is 354 W (=4.7 W/kg). I followed TR plans to a T from Dec 2020 to July 2021 and again from Dec 2021 to Apr 2022. This off season, I’ve instead been doing a bunch of endurance/tempo work with 1-2 sweetspot sessions per week. I’ve been progressing my TiZ myself, but I use Train Now to get roughly the workout I want or I pick through the library for the right workout in the progression. I’ll next be moving into a VO2 block where I’ll be doing the same thing. Then a threshold block. What it comes down to is I don’t want to think that hard about my specific workouts. I used to write all of my own programming. I’m lazy.
Regarding testing vs. AI: I haven’t tested in a year. I’ve used AI FTP a few times. But I feel like @kurt.braeckel you wouldn’t disagree with its use. It gives me a number to base my workouts on. I then go validate it with my workouts, particularly sweetspot:
I view my FTP more as a training number than necessarily a representation of what I can do for 40-70 minutes on a given day. My ability to ride at threshold for a sustained period is going to depend a lot on the type of work I’ve been doing leading up to that time. Eg, I could happily ride at 90% of threshold for a loooooong time but my TTE at 100% may comparatively be shorter. Similarly, I could be ripping 5x5 at 110-115% but have worse subthreshold TTE if I haven’t been working it. I also have the issue of that I kind of suck at riding at threshold, so I have to guess my threshold based on sub/supra-threshold work rather than going out and testing it straight up.
3 Likes
Yes, I know, I said exactly that in my post. My whole point is that the taste of the average TR athlete compared to the average coached athlete is different. Personally, I enjoy endurance rides, especially outdoors.
If it’s working for you, I’m all for it. I’ve repeated my anecdotes about athletes for whom ramp tests/(improperly done) FTP tests/made up numbers… didn’t work to “catastrophic” effect to their training. As @OreoCookie and I have gone back and forth about, I really don’t care HOW you get those training numbers… I only care that they’re accurate/valid, and I’ve seen way too many cases where ramp testing doesn’t give us that, partly because athletes just want to see a bigger number and are willing to blindly follow that bigger number straight into oblivion.
FWIW, last year when AI FTPD was in Beta and I was still uploading my rides to TR, AI FTPD was within 2 watts of my mFTP from WKO5. I thought it was pretty good. That said, I understand what goes into mFTP, so I’m more inclined to trust it than a “black box” like AI FTPD. I haven’t looked at AI FTPD lately… maybe I should.
Above threshold is where I really think TR’s workouts really fall apart. For someone like you, I would want to see a much more advanced VO2max block, back-to-back days of max efforts and max breathing, lots of Z2 volume, etc. I just don’t think doing TR-style “VO2max” zone workouts are going to be all that effective, but let us know how it goes!
1 Like
Just to let you know I am really appreciating the dialogue back and forth between you @OreoCookie and @kurt.braeckel
3 Likes