I think Suf should consider real customisation depending on riders’ performance based on the each individual’s HR, power target performance and his/her ability to keep up with the program. Some algorithms can be created I supposed based on wealth of data you guys have. I believe the current customisation is pretty much static in a sense that it only looks at the initial 4DP profile without any dynamics in between. The only customisation I see is the target adjustment for each 4DP parameter.
Where to begin? This could be quite long but I’ll keep it short but ultimately, I disagree. Why?
Firstly, HR is too variable. If you are, say, doing a long interval at, say, 250W and are feeling good and your HR slowly rises and stabilises at, say 160bpm. If you were to do the exact same power and interval after having had some caffeine, that may become, say 162bpm…judging from your post you would be expecting SUF to say “hang on, you’re working harder, therefore we best knock it back” when in reality that is not true. Your hydration levels, the time of day, stress etc all impact HR and don’t necessarily impact how you feel in relation to your HR (eg you can have a higher HR but feel great, likewise some days your HR is a lot lower and feel great).
Secondly, we then have HR variability, which is pretty complex and some devices can monitor this but is often best in the morning, some devices measure as you’re riding (eg Garmin) which can provide “real-time” info on how you’re actually feeling, but whilst it does seem FAIRLY good, it is not gospel.
Based upon the above, and a few other reasons I won’t go into, I am unfortunately going to disagree with you. It wouldn’t be reliable enough, and also this is why SUF has pretty regular testing if following a training plan.
I appreciate what @fairushariri is saying. The metric updates are a little static.
In zwift if you do a race faster than you have ever done before it will update your FTP.
It would be good if in SUF some videos could be used to test specific metrics. I appreciate HR is variable but this is a technique SUF uses in its HM test to estimate your FTP. So even though it’s not perfect it can still serve a purpose.
I agree on HR being too unpredictable as a solid metric to base any changes on. But I would have thought power could be used to tweak 4DP values dynamically. I haven’t used it myself, but Xert uses power to dynamically adjust FTP without any formal testing and appears to be well regarded. Zwift Power also estimates your FTP from your ride data and in my case their current estimate is within a couple of watts of my last 4DP test.
I still think it’s worth doing the formal testing, but would be cool if SUF flagged up any anomalies in your tested numbers during the course of a training plan. For example if you underperformed in one of the metrics during the 4DP test, it may become pretty obvious when analysing data in subsequent workouts, especially if using level mode. It’s harder in ERG mode for sure as SUF would then need to analyse HR data along with power to make any recommendations.
You’re right @Peteski. My HRM context may be slightly taken out of proportion.
For instance, Power and cadence target could be the main parameter that can be leveraged. Even if the HRM isn’t accurate and subject to environmental and other condition, there must be some weightage that can be applied to it. My key message is for Suf to consider real customisation rather than a one-time target adjustment over the next 12 weeks of a program, in the case of all purpose road plan.
Not a programmer myself but there must be some algorithm that can be figured out.
The biggest problem with this, though, is that Sufferfest is an exercise system, not an entertainment system.
It matters little in Zwift if it adjusts your FTP day-to-day, especially if you’re not using a training plan.
The biggest problem is that performance is variable.
You can even out some of these factors by creating a test that means even on your freshest day you will be fatigued through the effort and the balance, hence Full Frontal.
However, if you were changing data on the fly and the entire system is based upon the performance metrics that the system expects you to be able to achieve, the only way to do this sensibly is by a measurement over time.
What you absolutely wouldn’t want is to have a day where you feel really fresh, so you outperform your targets, so the system ups them for the next day, but you can’t maintain those even though you try, introducing a lot of additional fatigue into your system. On day three you get slightly reduced metrics, but you’re over-fatigued so you absolutely crash and now the system has all of your metrics below where you started purely because you had a fresh day and the system pushed you into exhaustion.
The way to eliminate this is that you don’t make a change based upon a single entry, or even a couple, because they may not be reflective of much of anything other than what you’ve eaten and how well you slept. You track effort over time and record a trend, making adjustments based upon this…
Okay, so why not do that?
Because this would still involve records over a course of weeks, not days, to come up with a system that didn’t break more plans than it helped.
Half of the plans are only four weeks long anyway. The twelve week plans have testing every six weeks, so by the time the system can guess at the changes it needs to make, you’re pretty much due to be measured for it anyway.
They already offer a chart of how to modify workouts if you’re feeling fresher or more fatigued than normal, so you can make small tweaks for a week or two on the run-in to a test that are “safe” then get your numbers from testing.
I understand where you’re coming from, but personally the benefit of something like Sufferfest is that it provides near-professional level training to home users. The only way to achieve something like you’re asking, without the strong potential to do more harm than good, would be for someone to be strapped up to (better than standard consumer level) sensors and start recording their HR and other variability factors outside of their exercise too, so the system would also then be able to judge actual fatigue levels etc.
I just don’t see the return in value relative to the cost (both from a company or consumer perspective).
Has anyone had any experience using Xert? Since they are employing exactly the kind of dynamic metric analysis being discussed here. Personally I’m happy with the static approach, but it does make the tests ultra-critical for each following training block. Having to re-test 4DP is no joke!
just registered as it looks promising. And always open to trying out something new and evolving!
This could be one unpopular comment - I guess the bigger question here is, if xert can do it why can’t suf? Both apps (in fact all training app/program) are using the same parameter aren’t they…
I don’t think it’s that easy to do. Suf will have to hire a couple of specialist data scientists (if they don’t already have any) to build the kind of models that can make it work successfully. Also, data science is as much of an art as it is a science. Not saying that it’s impossible, but it will be something that would require a lot of research and development before they would have something viable (at least I don’t think it’s as simple as using an off-the-shelf model).
But who knows, maybe it’s something they’re already looking at, but keeping it on the hush for now.
Another training platform that adjusts training plans dynamically as you progress is Bikevo.
Again I have no experience of Bikevo or any affiliation whatsoever. But along with Xert it shows that “smart training” algorithms are a thing and already out in the market. I have no doubt that the SUF guys are already looking into this kind of development.
Right now I appreciate the relative simplicity of SUF training plans and I don’t think my metrics are changing quickly enough to necessarily require a more dynamic approach. But it would be nice to see some form of dynamic feedback on where my metrics are heading during 12 week blocks. Of course I can get this kind of insight from the likes of Training Peaks, intervals.icu etc, but would be nice to have it wrapped up into a single package from SUF.
Hi, my 2 cents here. There are few applications/platforms that right now are using dynamic approach to adjust power/speed/FTP or whatever, you name it. This is, in my opinion is new approach in sport data science. For example, TP has WKO application to update/adjust current fitness level of the athletes. If you are runner you might have heard about STRYD, there is ‘auto update’ function as well.
From my experience, this is done to replace ‘pure test workout’, so athletes don’t have to do it every 3-4 weeks.
Another aspect is that not all amateur athletes are willing to take test repeatedly.I have few athletes who are fearful of taking either FTP test or 5k run test, unless I would totally surprise them with that.Let’s admit, run/swim/bike those tests are not a walk in the park. All training plans have tests to validate fitness form of the athlete, and it has to be done in order to adjust training plan and it’s goal/target.
Which approach is better? I don’t know, and I don’t think anyone does. As I remember correctly, there was a SF article, where possibility of adjusting current level in the middle of the training block/plan had been discussed. If you feel that workouts are not that challenging anymore, go ahead and increase your FTP level by 5-10 points, try it out, you don’t need platform to trace it.
I too am not particularly bothered by metrics being dynamically changed, I like the idea that some videos could switch to level and be used to test a certain metric, but then again we have Half Monty which adjusts FTP/MAP which are certainly the two most important metrics in the training sessions.