Why are smart intervals so dumb?

I struggle with the same issue. If I fail to hit the targets for an interval, the workout demands increase and I end up quite often quitting the workout. It has been a frustration of mine with Xert for a long time now, sometimes this means that what should be manageable endurance becomes threshold and so on. I suspect I’m a similar type of athlete, in that most of my ‘fitness level’ is made up from accumulating a lot of XSS during long rides 200-400km, this then means that I get recommended 3.5 star workouts and when I start to fail an interval the next one is even harder and eventually I’m unable to finish the workout.

I’ve seen hints on here that there is going to be a change which should give more credit for and the type of endurance XSS, but there doesn’t seem to be anything concrete. I’ve pretty much resigned myself to the conclusion that unless I absolutely micro manage my freshness, workout difficulty etc Xert really doesn’t get what I do. At which point, I realise that essentially I’m pretty much writing my own training plan, so what’s the point of paying for Xert?

To demonstrate with an example, I recently rode over 400km and 7000m of ascent on a two day bikepacking trip, Sat-Sun, on the Monday Xert showed me as fresh and was advising 3.5star workouts that I’d struggle to do normally. I know that long distance riders are probably outliers when it comes to the algorithms, but there is no way that I woulddescribe myself as ‘fresh’ after that. I’ve recently stopped doing the Xert workouts and gone back to set intervals instead, as I find it less frustrating.

3 Likes

Hi Tobias,

Which workout was it - a standard Xert workout, or a coach/personal workout? Do you feel that your fitness signature is accurate to your fitness level? Only asking because we do a lot of testing with our workouts to ensure that they’re challenging yet achievable.

That is a good example of the ‘freshness’ factor not scaling particularly well. Even if you have a high TL the level of XSS you must have completed, must indicate that you cannot be fresh. I wonder what the Xert model is thinking when it decides on your freshness.

1 Like

FWIW, You can only achieve the ‘tired’ status when your high/peak systems are strained. For very long events, you’re not really applying strain on those systems (unless you’re doing attacks/sprints along the way, perhaps). The XSS from those events is placed almost entirely on the low-intensity system.

There are ways in which the XSS distribution and freshness ratings can be improved. This is a big part of what will change with an updated model.

1 Like

In my case it’s been several, Killing Me Softly was one of the worst from memory but there have been quite a few instances of what the OP describes, struggle with one interval only for the next to get harder as a result and from there on in it’s usually a downward spiral. I stopped using the EBC app and imported the workouts into Zwift for a while as then the power demands have more of a static range. I completely get that by doing so I lose the ‘Smart’ part of the workout, but like the OP I have had several failed workouts when I use the ‘Smart’ function. I do think that part of the problem is that I’m recommended workouts by difficulty based on my training load, but that load is built up of a lot of low strain, probably much more than most, so when I come to do intensity I’m always struggling with what Xata thinks I can do.

Like I said before, I’ll be interested to see how/if things change when the hinted at changes take effect, as of now I’m tending to do fixed duration intervals, but even then, workouts that I’m just getting through get described as moderate rather than how they felt, which to me was quite difficult. My power curve is very much a steady slope rather than a steep plunge!

Changes? What changes?

I think it does pretty well for a completely novel method of activity analysis.

All models are wrong, some are useful. The key is having something that can be applied to everyday rides in a useful manner.

2 Likes

Yup. It’s very easy to find extreme edge cases where the model may not work well, which is why I pointed out that having something that works for most scenarios is more practical than endlessly debating if the model works in every scenario. Having MPA front and center of the Xert model does make it easy to nitpick Xert for being incorrect, but other models have the same limitations - do power zones change after 4.5 hours of racing and 5000 kJ of work performed? Not that I’m aware.

I’m not disagreeing with you, just want people to understand that this isn’t solely an Xert issue - nobody has long term fatigue modeled perfectly. If they did, that’d be the holy grail of endurance sport! :slight_smile:

3 Likes

Killing Me Softly has intervals based on XSSR which get easier, not harder plus targets decline during the sets. Example:

KillingSoftly

(2) Xert - Workout Designer (xertonline.com)

I don’t consider this a particularly tough workout, but I like short and spiky intervals (ex, Ronnestad, Tabata). However, some folks don’t do well with this type interval depending on their signature and strengths/weaknesses.
As a test you can always switch to Slope mode for a workout and see how well you match targets under gear/cadence control. If successful, I suspect there is an issue with trainer response in ERG mode (AUTO).

I only use slope mode, cannot stand ERG mode. The issue I’ve had is, say I fail that first big interval, then the rest increase the power demands to accumulate the required XSS. After that it becomes harder and harder to reach the increasing demands, and if I can’t manage the first one, then with increased fatigue, I struggle then to do the rest. I’m not criticising Xert, I actually think it makes a lot of sense, but I have struggled to get it to align it’s recommendations with how I feel and what I can achieve.

Case in point, yesterday I rode for just over five hours on quiet lumpy terrain, it was also very windy. Here’s a picture of the intensity graph

I get that mostly it was low intensity XSS, but today’s recommendations include workouts such as Something Just Like This. Looking at that workout, I very much doubt that feeling like I do after yesterday I would get through it, and that in a nutshell is and has been my dilemma whilst using Xert, it seems to think that I’m recovering much, much faster after my long rides than I really am, again I get the low XSS issue, but maybe it’s other things as well, for example age, I recently had my 50th birthday and when I was in my twenties/thirties I could probably do a workout like the above after a ride like yesterday, but now I need more and better recovery, so combined with the large amount of low xss that is contributing to my training load coupled with age, the recommendations don’t seem to stack up well with how I actually feel. Again, absolutely not a criticism, I just thought I’d comment so that the OP knew that they wasn’t alone with this issue.

I have found that Xert seems to work quite well for me in the base phase, but I’ve always run into issues when I start Build and whilst using Continuous Improvement as when intensity is introduced that’s when I start to fail workouts.

1 Like

I can certainly appreciate that ride felt a bit harder than the XSS analysis shows. Depends what the purpose of the ride is, but it takes a lot of discipline to keep the intensity low on endurance days when riding outside, if that is the goal of the ride. This is something that I think a LOT of cyclists struggle with.

Below I have an example of a 2.5hr low-intensity ride that I recently did outside. Fortunately I live in an area where I can create a route which is relatively flat, but I did still find myself on 8+% gradient a few times. I just knew that I needed to find a low gear and slowly spin away to the top. I’m also not afraid to ride in the small ring, depending on wind speed & direction. But what rides like this allow me to do is recover from a SMART sprint workout on a Monday and be ready for a hard (192 difficulty/5-diamond) 30-30 workout by Thursday.

I accept the fact that I wont be smashing any segments or averaging 32kph on my ride. But these easy rides allow me to adequately recover so I can do those things on my hard days :slight_smile:

1 Like

What was the ending classification? (title, specificity, focus)
Example, Moderate Polar Endurance versus Difficult Mixed Rouleur Ride.

Have you tried riding to Focus outdoors?
In one respect it is similar to riding an indoor workout in slope mode in that you control how many intervals you tackle at what intensity to achieve target focus and strain score for the day.
The work allocation goal is the same even though indoor vs. outdoor charts will look vastly different.

I assume you have athlete type set to Climber (10:00 focus) at the moment.
Something Like This is a serious hard start VO2max workout. That would be very tough even though the intervals taper off and target decreases as difficulty rises.
With Continuous ATP you are in charge of intensity dosage and focus plus how much rest you need between high intensity efforts. Sometimes I take two days off in a row and that makes a big difference how I feel during the next workout or ride.

I only have standard Workouts. Here are a two examples (from the suggested list, i selected climber to actual get a bit better in that area)
Example a) “Repeated attacks 4.0” (with some manual warmup)

/activities/rgmar1ppployls9u

I did manage this workout 1 or 2 times, and it was really though

Example b) " Higher Ground - 4"

/activities/dzm0shb4lze3lc3z

Despite the description that this is “Randoneur-Style” (and I’am one!) I never managed this maybe the first block once.

Well I did a more-or-less classic FTP Test outdoors and managed a breakthrough that way, so I guess my FTP is about right. My PP seems like 100 Watt to high. So seems like the HIE is wrong?
Otherwise Indoor I usually have to force breakthroughs by doing sets of hard but short sprints like “CX Ronnstead”.

ChopStick seems to explain my Problem very well. Only difference is that I really like ERG mode and can’t stand Slope mode :slight_smile:

Here’s an example of my more extreme end of riding:
/activities/ktr120erik9cxw4y
Without stops this somewhat like ~175 Watt average which I think is pretty good considering I only had 2 days of rest from the previous 1200km ride (see my other recent post).

Sadly I can’t test anything right now because between this 600er ride and the train home somewhere my frame snaped ://////

P.s. I had to reduce the urls because of 2 links limit.

Once you reach 4 stars status you are going to get handed some tough ones. :wink:
You want to be especially fresh when tackling those workouts.
I haven’t successfully completed Repeated Attacks 4.0 but had better luck with 3.5 although the second set can generate uneven results. :grimacing:
Ditto with the Higher Ground series (2.5, 3.0, 3.5, 4.0 options).
While ideally you should be able to complete matching TL to difficulty (stars count to diamond count), you can always step down to a lower difficulty level the next time and see how that goes.
Also, lowering intensity when in AUTO mode is an option. Sometimes it only takes a few pts to make it through or skip an interval. The end result has the desired results unless the entire workout is a wash out. Sometimes that happens. Add some notes to description so you can reference them next time.

By default, the recommended list includes workouts from the Standard, Coach, and Personal folders (user-defined or imported). You can uncheck Coach and Personal to limit choices to Standard. The reason you may want to do this is Coach and imported workouts may not follow Xert’s best practice guidelines.
The vast majority of Xert curated workouts in the Standard folder have some level of scaling applied. That includes SMART intervals tied to your signature values (not %FTP) such as curvilinear based on XSSR, along with intentional breaks and scaled RIBs. You may also notice interval sets taper down as difficulty rises (shaded portion of the chart). The set or intervals get shorter, target watts decline, or some combo. That helps me a lot. If I can make it past the halfway point intact, the remainder should be easier. :thinking: :pray:

1 Like

Oh. Never realized that I thought it selected the difficulty based on the freshness. But that explains why he even suggest the Hardness Test Level 14/15/16 …

Problem seems to be that those long-distances events during the summer really inflate my stars. For Saturday even 5 now after uploading the other event.

I have no coach, and I have no idea for my own workouts. Getting some intressting workouts designed by someone who know what he was doing was the main reason for me to switch to Xert :slight_smile: I’ve used Bkool before and they basically only made me do boring 20-minute-sweetspot stuff.

Have you tried manually altering the Training Mode on your Xert app to Slope say at 2% then using that to get your cadence up and then returning the Mode to Automatic to continue the work out?

Somewhat. I usually switch from Auto to ERG-Mode with just 100 Watts to get up to speed again.
But having to adjust manually is NOT what I would consider a “SMART” Workout and that’s my OP well - lets be honest - rant.

So @Xert: Either make them actually “SMART” or rename them to something more appropriate to the real deal.

Form determines recommended intensity (low, high, peak strain).
TL (stars count) determines the range of difficulty (diamond count) recommended.
Example, at 3 stars status, 3 diamond workouts will be recommended that include both low strain (below TP) when form is rated tired and high/peak strain (above TP) when form is rated fresh.
Active recovery (low watts or rest days) will be recommended if very tired.

TL is measured by XSS over time (hours/week) that includes a variety of low, high, and peak strain (work allocation ratio).
Workout intensity varies from low strain (endurance workouts = blue/aqua range intervals) to high and peak strain (yellow/orange/red range).
When you are fresh high/peak intensity is warranted and you can select from a variety of workouts that include that level strain. Athlete type defined determines focus duration point of those workouts.
When you are tired (high/peak strain recovery required) you can select from a variety of low strain workouts.

Depending on your circumstances you can always dial things up/down manually by using Filter to select a higher or lower difficulty range (diamond count) or a different Focus (athlete type), Or change recommendations by adjusting the Freshness Feedback slider whenever predicted form does not match how you feel.
The most important factor is TL over time. You can adjust IR accordingly to accommodate periods of more or less training. However, anyone doing ultra-endurance events will need to ignore XATA at times (large surplus created) and select appropriate workouts/rides until you return to a "normal’ training load over a period of two weeks or more.
OTOH I don’t consider any type of ultra-endurance as normal. :smiley: (way beyond my league)

This is a great explanation.

1 Like

But it could be, couldn’t-it? When the cadence drops below e.g. 20 that could be taken as an indication that the interval could not be completed and the target resistance could be set to a real low value to allow restarting. Am I getting this wrong?

Regards,
Ludwig

1 Like