@XCO_mtb: this would help - I guess
1 Like
You mean the high fail rate shown? I can only imagine it has to do with when you look at the workout. If you did a hard workout that day and then looked at that workout the predicted difficulty assumes you would do that today on top of your already done workout. The predicted difficulty is different if you pick a free day in the future with a rest day before so it knows you are rested.
If it’s not that, this really seems odd.
1 Like
This.
If you don’t like their number then it doesn’t matter what it’s called you’ll need to track two things anyway.
The name is irrelevant.
Definitely agree with this. I think FTP should be something that can be used outside of TR. For example - pace climbs at 90-95% IF, etc. We use this platform to get better in the real world, not just as a training target in isolation.
And, I just went and looked at the TR Workout Library some. Some of my benchmark workouts for whether my FTP is set correctly are level 6, not level 3.
Will be interesting to see what happens to their number when I start pushing the Threshold workouts hard a little later this winter.
7 Likes
You can already see that in FTP prediction. Doing Threshold workouts bumps it up so that the algo can bring your PL back down to 3. This results in an unrealistic 12W (3.3%) improvement in 28 days in my case. The trend will only stop (by design) when you start struggling at WL3 Threshold workouts and get stuck there, which will be really depressing, and will mean that your FTP is set too high.
3 Likes
That’s what I was wondering and started to suspect as soon as Nate mentioned the level 3 benchmark.
I’m doing custom workouts and a custom plan right now, so will have a block where the focus is 2 threshold a week. And, Level 6 has been a benchmark, solid “Hard” productive Threshold for me. Level 3 I’d blow right through if my FTP is set correctly. Sort of feels like TR is back to the whole time crunched approach loading more and more intensity.
Will monitor and see how it goes.
3 Likes
Same thoughts here. Probably I will need to decline FTP raises in order to progress to longer intervals. @Nate_Pearson Having an option for longer intervals / lower FTP / FTP prediction around level 5 Threshold would be helpful for us more experienced athletes.
10 Likes
Just sparked an idea. Maybe an advanced setting that that allows the user to set their target Threshold Progression Level. Default it to 3, but let users set it on their own. For example, I could choose 5, 6, 7, etc. based on training goals.
With that said, there are times of the year where I’m just not doing Threshold - I’m at a level 9 sweet spot right now, but haven’t done a real threshold workout in months, so my Threshold level is artificially depressed right now. The system thinks I’m a 3.3 threshold right now, but that’s just because I haven’t been doing them.
Seems like this is linked or similar to the zone-specific slider in plan builder, but I don’t think that impacts your TRFTP.
4 Likes
if you’re doing a custom plan with all custom workouts, does any of this new stuff really matter? Not arguing your point, just wondering if it’s worrying about something that won’t affect your training anyway?
3 Likes
Short term, not at all.
But I am a subscriber, have valued the additional check on FTP and info they provided in the past, and have been thinking about coming back to TR at some point. So yes, I’m paying attention as this does weigh into if I do.
(Right now I’m working with a coach, and I think that gives me a more tailored plan and a better overall approach to the event(s) I’m doing than I’m seeing from TR. But, at some point, I’m going to take a step back from how seriously I’m taking cycling fitness and fighting to get faster, and be a little more casual about it, prioritize other things more, etc.)
I was at Threshold level 3.8 before the update. AI FTP reduced my FTP by 10% (for understandable reasons). Now, this value is certainly false. I want to emphasize that I fed the model bad data, so I understand the result.
The question is, what it the best way to show the model that I am capable of more? So far I have tried choosing a 9.7 level Threshold workout and a 5.4 level Vo2 max workout. In the Vo2 max workout, I increased the intensity to 110%. Both have not really changed the opinion of the model so far, which is not that surprising, although I hoped that it would at least move in the right direction.
On other idea is to do the workout as prescribed and rate it easy or moderate, assuming it would be one of those two.
Basically I wonder what the best way is to move the model in the right direction.
I would select and complete the harder (or easier) workout, and answer the survey accurately. Or just do it as prescribed. That’s the data that will influence it.
2 Likes
What @Nate_Pearson and team could do is to add a modelled power-duration curve like stryd is doing. This would be a great addition for when race planning. And would also mean that the you wouldn’t need to compare the new FTP metric to FTP from other sites like zwift, strava, intervals.icu, trainingpeaks and so on. You could probably extract a somewhat reasonable guesstimate of an FTP to use on the other sites from the power-duration curve
9 Likes
Am I better posting on here a problem I’m having with the New AI version or sending an email to support about it. Thanks
This seems like a good technical solution with bad optics and messaging, regardless of its accuracy’s and merits.
In the app previously and currently you set up FTP as the primary tracker of progress in the app, workout levels I think we can agree are secondary.
You get a little confetti when it FTP increases after a month. Its huge dead center on the opening page. It’s now predicted how much it will get better.
Why would you reuse the name, knowing by your own data above approximately 70% of the people (if you use treshold 3) are going to lose FTP? You’re setting up 70% of your customers base to loose progress on the key metric you’ve taught them to track. Now all of those people start with a bad taste for this.
Call it something different and this goes away.
To all the people going “the number doesn’t matter”, then why even display it? Just have it hidden and the platform chooses the workout? It’s because people do care, because the app is designed to make you care.
10 Likes
The new FTP AI detection gave me 352W when my current FTP is around 310W. The old detection has been pretty accurate even compared to traditional tests done outdoors, but this new estimate is way off.
I hope it improves as I do more workouts.
2 Likes
@Nate_Pearson for those of us who don’t use TrainerRoad plans / workouts, do you still see the AI FTP number as accurate & useful for pacing efforts (e.g., outside climbs), setting Tempo / Sweet Spot / Threshold effort levels?
Based on prior posts, it sounds like TR did consider these options, but thought it was a bridge too far to completely abandon FTP as a concept. I assume they knew there would be fallout and confusion changing the calculation/approach, but I’m sure every path they considered had challenges. FTP means a lot of different things to a lot of people and you are moving people’s cheese when you change how it’s calculated. And that’s going to be disruptive, particularly for people who use that number for things beyond right-sizing TR workouts. If changing the way the number is calculated and displayed makes the TR system more effective, then that’s probably the most important thing from TR’s perspective. And I’d bet the vast majority of TR users are fine continuing to call it FTP because they weren’t that committed to the old number/calculation. And I’m sure the new number will be an improvement in the eyes of many, but that doesn’t mean it’s better for everyone.
For me, the new calculation took my FTP from 295 to 318. And for my definition (or any common definition) of FTP, that 318 number is a good bit too high for me (I’d say ~15 watts higher than any reasonable physiological FTP calculation). And I lean on that FTP number for a lot of things outside of setting wattages in TR workouts. But it’s not the end of the world and I’m only a few days into seeing how it will play out. Maybe as I get into my season and the system has more data, the TR FTP number will get closer to my physiological FTP. In the meantime, I’ll just make sure I use a more appropriate number for the other things I lean on FTP for (pacing, other apps, etc.). It’s a bit of a pain, but not the end of the world and I’m mostly curious to see how the change translates into prescribed workouts at this point.
My first workout with the new “FTP” number in TR was an over/under workout today that was essentially an over/over from a physiological perspective (or at best the unders were right at FTP). It was only 3 x 9’ intervals with short (1’) overs and long rests in between, but it was still very hard. I completed it, but I am wondering if that type of workout is right for me during base phase and only in my 2nd week of structured training this year. But I like trying new stuff, so I’m gonna let it play out a bit and see how it goes.
7 Likes
This is definitely one of those things, whether tat was change “ftp” or get rid of “ftp” or do anything else. There will be fall out. They needed to balance that fallout and they chose what they chose. There will be fallout, there would have been had they chose differently.
I’m amused at the beta vs release comments though. Beta thread had lots of we need info on what to expect give us info. But they couldn’t provide solid info because it was beta and they didn’t want leaks or have wrong info out if things change. Now I actually have reads posts with people saying they should not be expected to have to read about something.. Even in documentation and presentation they can’t win.
3 Likes