We don’t give the data to anyone outside of TR, so there’s not a need to anonymize it. We don’t store left/right power for TR rides, but we do for rides from Strava/Garmin. If left/right power becomes the key to unlocking the next evolution of cycling training I will eat a TR hat 
.
This was just an example. As a data nerd, I think it might be useful to save the “raw” data — just in case something interesting shows up. Left/right power data was the first thing that came to mind.
And you needn’t use left/right power data to get rid of a power imbalance. But to speculate, perhaps you could detect when someone gets close to his or her limit, because you notice a tendency in the data for the left-right balance to increase. (On my dumb trainer, I know I get close to my limit if I am getting more and more sloppy with my power targets.) So this additional data might be useful one day, although not necessarily in the way you initially thought it would.
We are totally interesting in save lots and lots of data. As you said we never know what will impact it (although I suspect L/R won’t).
In general I’m against editing any workout live based on performance. We have too many rides where people “sort” them selves out for one interval, refocus, then nail the rest.
I believe that if someone knows that the workout will get easier if they start to fail, they will fail more often to make it easier.
The key is to make it so people get challenged appropriately in each and every workout and you get that failure rate really super low.
I am probably saying too much!
You can’t answer something as short as tl;dr here, but if the forum rules allowed it, that’s what I’d answer to @Alex_Fastfitnesstips. Instead, I’ll have to switch to French:
Ce que l’on conçoit bien s’énonce clairement, Et les mots pour le dire arrivent aisément.
Because going hard is one of the most important things to stay fast? That and maintaining muscle mass (weight lifting). And latecomers like me might also benefit from a lot more volume to increase vo2max via adaptations in stroke volume and mitochondria.
Thought I might have read something about aging and RPE but couldn’t find it.
What you are talking about is implementing a more sophisticated controller. Something like:
- The trainer is set to erg mode, with the target wattage being the mid-point of the range
- Once you “stabilize” at the mid-point, a slope is calculated such that if you kept your cadence at the average (e.g., 90 rpm) then you be doing the mid-point power
- The trainer is set to slope mode using the above calculated slope
- If you move outside the target power range, then the slope is increased / decreased to get you back towards the middle of the power range
- At the end of the interval, the trainer is set back to erg mode at the recovery power level
The above would feel somewhat like riding on the road where the gradient is constantly changing. If done poorly, this would feel awful, and you would be constantly “chasing” the power number. This could be done, but would take a lot of work to “tune” it, and it might only be possible with certain trainers. That is, if the trainer lag to changes in slope are too great, then this system could become “unstable” and you would be constantly hitting the power range limits.
Any reason that TR doesn’t capture L/R for TR rides? I’m not saying its the next best thing, but would be great if TR captured everything that is in the FIT file spec, so it would be there if people wanted to look at it.
A simple answer that comes to mind is that older men (btw: just men and not woman?)
- are not as likely for inflated FTP due to ego
- have spent more seasons with TR so that the portion of those with FTP < 75% MAP are more likely to know that fact and adjust FTP down
And as a result hit VO2max better.
Nice insights @Nate_Pearson! Always nice to read teasers on the forum or podcast!
I reckon you could use the data in a myriad of ways such as subtly tweaking your “FTP” or giving the users fitness metrics by which they can track their progress. Plus, there are all these machine learning approaches that identify unexpected markers.
Given your not-so-subtle hints, it seems you are already actively working on it And I am glad to do my small share to fund that endeavor.
Correct, I would think you would need a scaling erg to get it to work at a somewhat acceptable way but I was just spit-balling ideas. But I am no expert at control theory so I am sure theres some significant issues I am missing. Anyways, I am perfectly fine with ERG and TBH don’t see another “better” system to use in a pure performance standpoint. If I want higher wattage I just increase the %, conversely if I want a lower wattage I lower %. Pretty simple. I am happy with the current systems and would not really use anything else. I use resistance when I just want to spin the legs.
The only reason is because of higher priority device work. We will eventually capture it and push it to Garmin/Strava.
Hope I’ve done enough to earn your last-minute visit down to Melbourne… 
And, shameless plug, I’ve got 11 years experience in IT strategy with the Department of Health down here in Victoria. And I’m a self-taught (but as-yet unqualified) coach.
If you’re ever looking for some part time consulting, beta testing, UX feedback etc drop me a line. I love this stuff
Wish I can find the PDC with multiple users used by Dr. Coggan to illustrate the huge deviation in shorter duration power outputs. The assumption that 120% of FTP is power at VO2 Max for everyone is just not true but work in general as a train prescription. Same goes with >150% of FTP for anaerobic capacity. If this is not sufficient, then use the appropriate estimate (test) to narrow the training to target those energy systems, e.g. MAP for VO2 Max and CP & W’ for anaerobic or learn to understand your own PDC. To gripe that FTP doesn’t does reflect your physiology capabilities above FTP is a bit silly. FWIW, W’ bal model is still work in progress for almost a century since initially postilation (Hill, 1925). Haven’t see anything that demonstrates 4DP & Xert does a better job than Dr. Skiba, PhysFarm. The free Garmin app held my interest for about a week and it’s an interesting post ride analysis with Golden Cheetah (now available in WKO5).
The question is also what you use as a point of comparison. The point of TrainerRoad as I understand it is to make structured training, which previously necessitated a coach, accessible to a wide audience. Compared with a good coach, TrainerRoad will probably be less effective as it has to be more generic at this point. The question is, though, less effective by how much?
Of course, ideally, you’d want a computer-generated custom plan that takes your power profile into account. There are lots of new technologies available, not least all the various things that are subsumed under the moniker “machine learning”. You can be sure TR has been working on that for a few years, but getting something out that works is quite hard I imagine. For example, if the software overestimates your VO2max power workouts can become impossible for you. Also, perhaps the deviations from the defaults could be very small, so if you are an experienced rider, you can manually tweak the difficulty of a workout by a few percentage points.
However, there are so many other factors such as sex and age. For example, you could give older athletes more rest days when planning the season with the Training Plan Builder. TR user data will be a treasure trove.
And given that development time is a finite resource, it makes sense if the company decides to re-allocate its resources elsewhere. IMHO beefing up TrainerRoad to the point where its apps rival the functionality of TrainingPeaks would be a better use of time in my estimation.
Let me pay more. TR keeps me fit and fast!! Plus when I sleigh hills like Santa I yell “Trainerroad b*tches” as I pass my mates. Well worth the money.
I’m guessing people complaining lack the understanding ARPU, LTV, CAC, magic number or all that other jazz. Afflicted or affiliated? I’d submit afflicted with a lack of intelligence.
For those not in the SaaS world there are lots of good reads:
Benchmarking Hubspot's S-1 - How 7 Key SaaS Metrics Stack Up by @ttunguz
ARPU: How to Calculate and Interpret Average Revenue Per User
Decoding SaaS Metrics – Average Revenue Per User (ARPU) - Chargebee Compass
First and foremost, you all are running a business not a nonprofit. As long as you are providing value, I’ll pay. Most others will as well. Of course, the price is elastic: figuring out what to charge is up to you. So far it has been stellar.
You are also providing jobs. Nothing wrong with having healthy margins. Especially to have a little cash with the economy is not as healthy.
That being said, plan builder alone is worth the price of the subscription. If you only knew how many different ways I have attempted to do this on my own. I just scrapped every stupid excel sheet. Man. Hey. That plan builder is bomb. What a time saver.
In summary, forget the haters. Merry Christmas! TR b…
I initially tried Sufferfest because of word-of-mouth and price. It worked in the beginning but I eventually fell into over-training. I was following their template plans but they don’t have the structure (base/build/peak) that other platforms have. I enjoyed the individual workouts but overall it was flogging me to death (which is a badge of honour apparently).
TR podcasts sounded sensible and intelligent so, after trying coaching for a while, which wasn’t working for me either, I took the plunge with TR.
Training balance is a lot more sensible now and I’m seeing improvements, although it’s early days. I like the way plans are organised so it’ll be interesting to see how this works for the long term.
And a lot of 5 star ratings!!!
I’ve always thought that was a little mean
That would be great and would avoid the inconvenience of having to stop a bit and twiddle with the intensity (for those of us using a laptop off to the side). I don’t need it often, but when I do, I really don’t want to break my stride by stopping to tweak something (and not get it right, have to raise it a bit again, and so on).
Am I right in thinking that provided you continue your subscription you pay the rate you started on when you renew rather than the increased rate?
thanks
I’m not getting involved with this one, but not having (very good) training plans, doesn’t make Zwift a “not training app”, you can get your training plans from multiple sources and do them in zwift, create your own (even better if you do it in trainingpeaks), create workouts without need for a seperate app, if you have a couch who uploads training workouts to Trainingpeaks (like some of sufferfest endurance workouts) then they download to the Zwift trainingpeaks folder and appear automatically
What makes a training app, is more complex than does it have a training plan (that may work for you)