XPMC -- compensate for activities other than cycling (e.g. running)

Hi – seems like stress, strain and XPMC would be warped for athletes that do more than only cycle. For instance, I stopped cycling in September to focus on a half marathon in October. While I didn’t cycle during that time, the overall impact on recovery and fitness seems skewed. I also like to alternate days between running and biking, but xert only picks up a portion of the story. What are your thoughts on how to get a total view?

Hi Peter. There are a number of ways to enable the ability to account for activities outside of cycling. What drives of the XPMC is XSS. What drives projections is XSS, together with Focus. Hence, getting estimates of XSS and Focus for non-cycling activities will provide a much better overall picture.

The simplest method is to simply allow you to enter the metrics yourself for example, on activities without power data. These could be running or cycling or any activity. This capability is currently in development and should be available after new year.

A method available today, is to use Xert Mobile with HR Derived Power during your non-cycling activities. Although not something useful retroactively, it can be used on future activities. Tune it with your min/max HR and the power data should be relatively accurate enough to provide decent XSS and Focus numbers for other activities.

Later next year, we’ll be offering the ability to determine XSS and Focus from HR data. Similar to the current test feature in XM, the algorithm will calibrate your HR response f rom other activities where you have a power meter and an HRM and use the algorithm on activities without power. This will also improve data coming from XM.

Not everything is in place yet but it’s coming.

You can have power data from running using a Stryd power meter. (https://www.stryd.com/)
Running power is a fair bit greater than for cycling so you would need to be able to have two different FTPs.

How about doing this using something similar to rTSS in Training Peaks - this is GAP (Grade-Adjusted Pace) driven. So the LTHR pace == running FTP and other paces interpolated.

I would find this useful as well - right now a good part of my harder trainings are running - but XPMC shows degraded performance, even though I was doing >100% VO2Max hill reps the other day.

Normalizing all the various metrics is the challenge. If you mix XSS from running with a Stryd and cycling with a power meter, things are not going to have much predictive value. Technically, it could be possible to calibrate the Stryd with your power meter via our HR Derived Power. That might actually be pretty good in the end.

The rTSS concept would be interesting in situations where an HRM was not used and you still needed to have some sort of overall strain estimated. I’m sure it could work quite well once your running FTP is established and if used together grades, factoring in your body weight. Still momentary surges in running might not get picked up and these are the ones that shift the strain from low to high to peak strain.

I like the idea of using rTSS. Especially for intervals, that’s going to be a better measure than HR … for the same reason that power beats HR on the bike (e.g. responsiveness, impacted by other factors).

I think the xert analytics are great for biking. My main goal right now is to build up for some competitive duathlons, and it would be awesome if the analytics could integrate well across sports.

Hi Peter. Duly noted…

Test out Xert Mobile’s Derived Power if you have an Android and see how it works for you. It is more responsive than HR and does account for effects from heat, for example. Using some form of calibrated power is going to help us determine how to slice up the XSS into low, high and peak XSS to maintain the integrity of training load values. We also need some measure of power to determine how endurance energy (i.e. glycogen stores) are being used up. This will play an increasingly important role, especially for multi sport endurance racing as well as other sports. Resorting back to generalized estimates on strain without power data will affect confidence in any interpretations.

I’m sure that Xert as it now stands could be used to analyse running power as generated by a Stryd running power meter. All the calculations such as HIE, Threshold Power, PPO, XXS, Training Load and Form would all need to be running specific. This would require a way of marking a file as either a ride or a run to keep the data separate from each other without going to the expense of running two accounts.

You may then want to be able to combine the data from both running and riding to measure Total Training Load and Total Form. It could be interesting to see if it is possible to eek out from the data any cross training effects.

Armando – I would be willing to invest in a Stryd if you want to include that in testing!

Rohan. Yes there is a bit of complexity in all this that needs to be addressed. Some values like XSS could in fact be combined, if they are derived from signatures associated with each discipline. If you now add in swimming, you can imagine that it gets a bit complicated to create a way to manage both the separation and the ability to combine them. But we are thinking along these lines…

Another approach would be to have the ability to scale power data by discipline. So if your FTP from Stryd is 1.2x that from cycling, we could scale the power data down and simply label it as running. This would save the trouble of having to manage two separate fitness signature histories, one for running and one for cycling. This of course assumes that the signatures will scale the same from running to cycling, which may not be true.

Peter, at the moment we’d have to keep your data separate as the Stryd data in your account will create discrepancies in your signature and thus to your strain and training load values. We could maintain your data in a separate account and then load in the strain from your activities to the original account.

Armando, I’m guessing that running and cycling signatures won’t look anything like each other.

Looking at my data for the last 12 months:
Cycling vs Running
Pmax 1052/487
CP 268/327
W’ 27.2/10.9

Perhaps there might be a way to transform one signature to the other. This might be another alternative. We’d have to analyze signatures from athletes that use both a power meter and a Stryd to see if there is a repeating pattern. For example, your 5 minute power is roughly equal.

I think you could have separate power signatures for each sport, but have the stress and strain interact across the signatures. Certainly fatigue from one impacts the other, and to a certain degree fitness from one helps the other (particularly base as opposed to peak). This would be useful for running, but as well for where people use different bikes (e.g. TT) where the power signature is likely quite different.

Yes. Each exercise modality would have a unique signature. Even within a sport there could be multiple signatures, as in XC-skiing. Eventually, the auto-identification of the specific modality, together with measurements of power (direct or indirect) would be the way to manage training across these sports and their influence on training loads, recovery loads and predicted results.

Armando – my stryd is in the mail, apparently. Should be delivered next week! I am excited to get some power data for running. What’s your current best view on how to stack up XSS? Certainly if I do hard running intervals it has a big (negative) impact on what I can do on the bike the next day (sometimes longer) and vice versa. I don’t think mapping the power signature is the answer though … having multiple power signatures but with stresses and strains that somehow stack across activity types would seem closer.

Hi Peter. We don’t have an easy way to manage multiple signatures. However, you can simply use the save/lock feature on a per activity basis. It’s a bit more work and you’ll need to keep track of whether a workout uses one signature vs. another. But it can be done and it should help you maintain your overall numbers more precisely.

Thanks Armando – I’ll follow-up to see how that works once I have some data.