Нes, it’s fun! And mentally I don’t know why but I can hold on longer and better. Plus, 40+ microintervals are less scary than with the traditional method 
2 Likes
Having read a lot of studies and having listened to a lot of scientific podcasts on training adaptation my personal conclusion for a middle aged man in lycra is: everything falls back to basic principles. Consistency, volume, progressive overload and recovery seem to be by far the most reliable predictors of long term performance. Xert with Magic Buckets is not a coach, but a useful tool to help with that, BECAUSE it does not care about how you achieve these basic principles. It just helps to build your preferred signature.
Specificity is for specific preparation. You can still train to increase time to exhaustion for a specific effort by getting used to the pain that comes with a given power output. Use it to utilize it. Use your signature to prepare for real performance. If your event is about reaching the limit for a 20min effort, do a lot of 8min intervals@20min power in the weeks before the event. If a single 4 min climb will decide the race, go for a bunch of 30s-3min intervals@4min power in preparation. If you expect some climbs with 20min+ at SweetSpot, go for some SS-Intervals. Remember: this is the icing not the cake.
No study on VO2-max / High Intensity - Intervention did last longer than 12-16 weeks for a reason. The more effective it was in the short term, the more participants dropped from the study. A lot of “survivors bias” (only those who survived will count for the result) to find in those.
Even one of the most interesting recent studies on time above 90%VO2max (https://onlinelibrary.wiley.com/doi/10.1002/ejsc.12202) does hint to the fact that the high responders did ride significantly more volume.
So the most important thing is: keep your training regime in a way that you want to hop on your bike tomorrow or at least the day after tomorrow. Not too boring, not too taxing in the long way, keep it interesting. It may be more effective to do what you hate the most more regularly (“5x6 hard start” in my case), but if I as an hobby athlete would start hating my own rides, I would sooner or later stop doing them at all. Cycling is not my job, it is my pleasure.
6 Likes
I agree that cycling should remain enjoyable, especially for amateurs.
But I think there is another side to this: for many middle-aged riders, meaningful gains are already harder to get, and progress itself can be a major part of the motivation. So if training time is limited, it is reasonable to care about making the work as effective as possible.
So maybe it depends on the rider. For some, the most sustainable training is simply the most enjoyable. For others, enjoyment comes partly from knowing that the training is actually moving the needle.
I think this is also where the product expectation matters. Ideally, XMB would give the rider confidence that the system is not only helping them complete strain targets, but also guiding them toward the best workout they can reasonably execute today.
That kind of confidence is very valuable — and some platforms, like TrainerRoad, make this promise very explicitly. I’m not fully convinced that the current interval prompts in XMB communicate the same thing yet. They may be very useful as adaptive strain-management suggestions, but they do not always feel like a clear “best workout prescription” for the stated training goal.
EDIT: Just to be clear: I’m not criticizing Xert for not being TrainerRoad. Quite the opposite — as far as I know, TR does not have anything comparable for adaptive outdoor training, and Xert may be the only platform doing this kind of thing at all.
3 Likes
These look pretty cool and effective! I’ll give them a try soon.
There is no such thing as a “best workout for today”, even Trainerroad randomizes their workouts. As far as I heard in their podcast, they even have a button for randomizing to another workout if you don’t like what you see. They don’t mind which one you do, they mind about intensity, progressive overload, consistency and recovery needs. Structured workout patterns are needed to compare and measure them (in science as in training plans), not because your body “needs” exactly 4x240s @ 350W.
As far as I understand, TR AI will recognize patterns out of graphical images of workout files (power lines) and compare them to other users who also executed the same workout. “If A did X and B did X and Y, A should be able to do Y too.” This is an interesting approach indeed, but it does always need a limited amount of preplanned workouts to be executed and compared. TR AI will chose a proper intensity for the given workout pattern upfront while Xert will just tell you how much strain you got from whatever you did - depending on your actual form while riding.
I agree that there is no single “best workout” in an absolute sense.
But “there is no one perfect workout” is not the same as “any execution pattern is equally appropriate for any goal”. If XMB repeatedly nudges a rider toward very short high-power efforts, that still creates a specific training bias.
My concern is simply whether that bias is always clear to the user.
1 Like
Isn’t it the case that XMB creates intervals precisely at such power level, that the yellow and red buckets are filled at the same rate, at the same time? That would explain the absence of longer, moderate intervals above threshold.
I can imagine a prompt system which suggests both longer, easier intervals to fill the yellow bucket, and harder, shorter intervals to fill the red bucket.
XMB is designed to fill all three buckets in the most time effective manner. So if you follow the guidelines perfectly all three buckets will complete at the same time in the shortest amount of time at the current challenge level.
I guess you know it, but it’s not possible to only fill the yellow or the red bucket. All efforts above TP will have a mix of yellow (high) and red (peak).
1 Like
Yes, I read the article, but I don’t think it changes the basic point. Even if every effort above TP fills both yellow and red to some degree, efforts in the yellow range will presumably fill mostly the yellow bucket, while efforts in the red range will fill mostly the red bucket.
So it still seems possible to design workouts with two types of intervals: longer/moderate “yellow” intervals and shorter/harder “red” intervals, instead of always solving it with one interval type that tries to fill everything at once.
The blog post is also very abstract to me. I understand that the model allocates work across low/high/peak, but I’m still not sure what practical training conclusion I’m supposed to take from it as a user.
Based on this discussion, I’d like to ask whether it’s enough to simply follow the intervals recommended by Magic Bucket, or if I should add something else as well.
For an upcoming event of about 110 km with 2,000 m of elevation gain, I selected the FD Roller setting (Competitive Grand Fondo, Hilly). I chose this FD based on one of my older rides with similar parameters.
Magic Bucket is recommending short intervals, for example 25 seconds on and 1 minute rest. Is it enough to follow what XMB recommends, or should I also include longer intervals? When I look at other training plans, they often include sessions like 3×20 minutes and similar efforts.
I’m not very experienced with training, so I’d prefer a simple “set it and follow it” approach. 
When I started using XERT last autumn, I was training mainly indoors and wasn’t using XMB. Instead, I was choosing SMART workouts, where the approach felt quite different. Short micro-intervals like these were not really part of those workouts.
1 Like
This is exactly the kind of situation I’m trying to understand.
For a 110 km / 2000 m climbing event, I would also intuitively expect at least some longer sustained work: tempo, sweet spot, threshold, longer climbs, fatigue resistance, etc. If XMB mostly recommends 25 s / 1 min micro-intervals, it may be consistent with bucket optimization, but it is not obvious how that maps to the demands of the event.
That is where I think the communication gap is. Is XMB saying: “this is the fastest way to fill your current bucket deficits”, or is it saying: “this is the best training stimulus for your 110 km hilly gran fondo”?
Those are not necessarily the same thing.
I understand that above-TP efforts fill both yellow and red to some degree, but it still seems useful to distinguish workout styles: longer yellow-biased intervals for sustained work, shorter red-biased intervals for high-intensity/anaerobic work, and endurance work for the low bucket.
As a user, I need to understand what practical training conclusion I’m supposed to draw from the XMB recommendation.
I’m currently training for a Gran Fondo 120 km with 1,600 m of elevation gain. I do various types of workouts: VO2 max( micro intervals 30\15 or long 5x5 ) (there will be short climbs where I’ll need to hold quite high power), long intervals at threshold (8-10 minute climbs), and sweet spot (20-60min) . Need to improve everything, maybe with a focus on weaknesses
My HIE (29+) and LTP around 220 , my weakness long efforts
I’ve been thinking about this discussion a bit more and wanted to throw out another thought…
Here’s a question for everyone: If we know your 4-6 minute power, do you actually need to ride at 4-6 minute power to improve it? Or can you improve it by training above and below it with XMB?
I was thinking back to one of the ZRL races from this past winter that included two ~6-minute climbs up Box Hill (MPA chart below). I never did any dedicated 6-minute intervals all winter.
Instead, I mostly raced on Zwift and used XMB on my high-intensity days. Before the race, I looked at my signature-derived power curve and knew that, starting fresh, I probably couldn’t sustain much more than ~330 W for 6 minutes (assuming my MPA was fully recovered at the start of the climb). So I paced the first climb around 315 W and the second around 310 W, stuck with the lead group over both climbs, and finished with the front group in a sprint finish (admittedly near the back since I’m not a strong sprinter)
Would some dedicated 6-minute intervals beforehand have helped? Maybe.
But I also think this demonstrates something interesting: If your Fitness Signature accurately reflects your capabilities (at least above threshold), you already have a pretty good idea of what you can sustain… even if you haven’t recently trained that exact duration.
Historically, I suspect that athletes & coaches would discover that number by having you do 6-minute intervals to discover it through trial and error. Perhaps this is where you’re coming from? But Xert predicts your performance from your fitness signature. Plus, if you start your ~6 min interval with more fatigue than you thought, the race plane/estimated effort is likely out the window anyways.
What do you all think? Is there still unique value in matching interval duration exactly to race demands, or is accumulating strain on your 3 systems & knowing what you’re already capable of enough?
I think we’ve come full circle! Ha! This is sorta what we used to do. Years ago, Xert did something very similar… we’d recommend a Focus Power for the day (the part of the power curve to target) along with the desired XSS. But many users found that too abstract, especially those coming from FTP zones or platforms that simply prescribe workouts.
I think this actually gets to the heart of Xert’s philosophy (and what makes us so unique!)
There isn’t one magical workout that makes you fitter. A workout is simply a means of creating the training strain (e.g. XSS) needed to drive adaptation. For example, 30/30s can be a fantastic workout… so can 5×5s. They’re different workout structures, but they can both produce a similar training stimulus if they accumulate similar amounts of Low, High, and Peak XSS. So from a big picture perspective, we don’t care about whether you did 10 min intervals at 105% FTP, or many repeats of 20s intervals with recoveries under fatigue. At the end of the day, Xert cares did you accumulate strain on your 3 systems in the intended amounts?
I can appreciate the idea of offering separate “yellow” and “red” interval recommendations. My hesitation is that every additional option adds another layer of complexity to a feature whose biggest strength is its simplicity: ride your bike, fill the buckets, and you’ve accomplished the day’s training!
Many people already find Xert fairly complex. Ironically, Magic Buckets has been one of the features that’s made the platform more approachable and, perhaps just as importantly, more enjoyable, especially for athletes who simply want to build fitness without following a rigid workout every day.
2 Likes
I use longer intervals for getting used to the power I will have to push in the event, not the duration. So if I have to go 340W for 6 min at my event I would do some intervals at 340W for 3-4 min in training sessions two weeks before. So I can accumulate more time at target power and learn to feel the pain without dropping my MPA too much (wich is easier in an event than in training anyways). My idea is to improve my central fatigue on specific powers as I know from my signature, it’s not an issue of peripheral fatigue if I cannot do them yet. It is a lot easier to do what I am used to.
1 Like
Just my opinion - probably this power can be improved with XMB training. But does the existing XMB approach provide confidence to the athlete that they can hit this power target for the specified duration? I think it doesn’t. Also, at least for me, it feels important to actually do those longer intervals because the way I feel muscles engaging at vo2max power (eg. 380w for 5 minutes) is very different from on-off style intervals (eg 450w-200w 30/30s). I mean, the efforts feels different on the muscular level. I get that the strain of both might be similar.
Also, today XMB showed me a suggestion to do 47 intervals of 9 seconds about 470w. There’s no way I’m going to do something like this as it looks pretty crazy and insane 
. For context, I had to hit around 12 HIE.
just my 2 cents.
1 Like
Physiologically I don’t know, but intuitively it feels there will be a performance difference (and I believe benefit) from doing longer intervals (which draw MPA down further and are more difficult) even if (modeled) strain on ‘systems’ is the same (and so predicted fitness signature improvement is the same). That may (or may not) be mental rather than physiological, but that doesn’t mean it’s less real.
I know if I’ve done 5 minute repeats at eg 380w I’m much more confident to press up a 5 minute hill at that power than if I haven’t… I know how much it will hurt and I know how recovery will feel etc… if I’ve only done 30/15s at 420w (made up) I just don’t think I’d have the same confidence, even if my power curve says I can do 400w when fresh (the curve is a max effort when fresh… I’m interested in why I can repeat so with some fatigue
Similarly if I’ve done 20min repeated efforts just sub-threshold, I have a decent idea what I can pace a long climb or other effort… and how hard it will feel / how well I can recover… if I’m only ever riding low intensity (sub LTP) or micro intervals I just don’t have the same mentality…
Maybe Xert should also periodise difficulty to make effforts more goal specific closer to an event?
Of course, YMMV
5 Likes
For me 9 seconds is too short to even ramp up to the target power. The interval is over before it properly starts.
On Sunday I decided to go outside and use XMB for my high intensity day. What I did what was I used to do many years ago when I just had a wired speed sensor. I call it natural fartlek and consists on putting the effort by feel in the most efficient way to complete a course in the least time possible. In a hilly route you push uphill. The higher the gradient the higher the power and recover in he descends. This is in fact how you should manage a time trial race.
I noticed that the initial XMB prescription on intervals power and duration and recovery targets was adjusting to the duration and power of the efforts. The difference between now and the past is that the XMB field gives me now a measure to control the stress of the workout, which is great for avoiding overtraining.
What I suggest to people who train for a race is, for the high intensity days, to find a route either outside or virtual (Zwift, rouvy, etc 100% Trainer Difficulty) where the elevation to distance ratio is similar to the race course and ride that route by feel to complete it in the fastest time possible, but using the XMB to control and limit the stress of the workout. Over time you can repeat the same course and gauge progress. This is more valuable than a fitness metric derived from non specific workouts
Obviously if you’re racing a multi hour event you don’t have to simulate the race in your workout but you can do several high intensity partial race simulation days in a row to work on durability near the target day (but with sufficient time to recover). This is for example what ultra trail runners do (the weekend shock).
The more I learn about training, and the more I know professional athletes usually do, the less convinced I am about the real usefulness of training based on fixed intervals and physiological targets.
2 Likes