Ah, this makes sense now. I always had problems when giving 2 max efforts for 2 durations, let GC Model WBal and then see the graph go below 0 for some next races. Quick VO2max gains or just a flaw in the data/model 
. For that reason I then used FTP test or similar to model back.
To be honest, most of what Xert is pushing is pseudo-science for cyclists.
Golden Cheetah is a solid free analysis program if you ignore W’ and WBal.
If you really want to have a decent modeling program and are willing to invest many hours into learning, then WKO4 can be really good, but you really need to know how to use it properly to get anything of value out of it.
5 Likes
I used it for the free trial, but couldn’t make head nor tail of it. I had no idea why I was getting breakthough’s when I and couldn’t understand why at times I wasn’t. It was just too complicated, I I think it overestimated threshold power quite a bit compared with testing.
I’ll stick with the fitness/form/fatigue graph with strava premium.
1 Like
There you have an estimated ftp number based on your power profile records for a chosen period of time as well, which is maybe the most useful feature Xert has? I guess TR will implement this as well soon.
I tend to ignore the estimated FTP on strava also I’ve never found it to be accurate. I didn’t thibk the Xert one was accurate either going by the numbers obtained through testing.
2 Likes
MPA explains the power curve and is not derived from it. Vastly different.
5 Likes
We’ve never promoted Xert as published science but then again, FTP, TSS etc. have not been validated externally either. At least with Xert, you get to validate the predictions yourself directly. This is what MPA provides and is what has drawn many thousands of athletes to use it as part of their training, including many of those using TrainerRoad. Xert can be very complementary to a TrainerRoad user given the ability to track new concepts such as MPA, XSS and Focus Duration that are unavailable elsewhere. Just as with the other metrics that have gained utility through practical application, so have Xert’s.
8 Likes
thanks for the feedback!
Do you feel FRC (and dFRC) are any better than W’ and Wbal?
If so, why?
It depends on the scenario.
In general I think FRC does a better job of representing what you could achieve on a regular basis when compared to W’, given the design of WKO’s PD curve, assuming equally fed models between the two. My CP and W’ curves for a given season are somewhat nonsensical, whereas my WKO4 PD Curve and associated metrics track much more tightly to my performance and capabilities.
To my knowledge, no one has put forth a properly validated model for reconstitution, which makes WBal, dFRC or any other type of dynamic estimate of anaerobic work capacity too error prone to be accurate. The WKO team even say that dFRC is more or less a “beta”.
I’ve also yet to see any concrete guidance of how it can be used to create a specific type of adaption. It’s interesting, but I’m still waiting on a practical application for it.
2 Likes
Thanks Steve, I tend to agree.
It’s interesting that the GC project claims the metrics they use are based purely on “science” (there’s even a whole page on it!) but compared to what I intuitively know are my own capabilities, the markers you mentioned (CP, W’ and Wbal) do seem “off” when compared with WKO’s equivalents - using the same data set.
It also appears (to my simple understanding at least) that the WKO team are more upfront about the fragility of some of these metrics (e.g. the perpetually “beta” dFRC). In fact Tim Cusick specifically states this in his recent first demo of WKO5.
Generally speaking (back to the original topic), I think the power training world is heading away from “testing” and towards modelling for the establishment of metrics, and on the design of adaptive/predictive training plans. To me, a merger of Xert and TrainerRoad (and a massive UI and usability cleanup) makes sense.
I’d also place a bet on WKO6 introducing a workout builder based on the phenotyping and PD smarts they’re beginning to accumulate. I’m not sure how much else they can improve on the existing feature set (and how many more new metrics they could introduce)
By the way, to your knowledge, what are the key differences in the design of the PD curve between GC and WKO (and Xert)?
I’m using both right now; I like Xerts model where it assumes your FTP (which is pretty close to WKO’s mFTP) and recommends training for the day but the actual training efforts are lacking compared to TR.
I’m not a fan of training plans, nothing is worse than missing a few days and feeling your plan is ruined. I’m also not a fan of testing. We do enough efforts software should be able to predict it (within reason). Ultimately I’d like something where I can wake up, hit a button, and it spits out a workout based on my goals, fatigue, etc.
Xert is almost there but not quite. Sometimes it will tell me to do a 40 XSS and suggest a 100 (because their library is limited), so I usually use a TR workout. TR has a much better workout experience so right now I’m using both. In the winter I may go full TR with a plan if this experiment doesn’t go well.
3 Likes
IMO: The market is bifurcating into 2 distinct markets. For lack of better terms, I’ll call them Market 1 and Market 2.
Market 1 certainly seems headed in the direction you describe. It includes companies who sell to users who don’t want/have time to acquire a deep understanding of performance analytics in order to coach/self-coach and have incorporated adaptive training into their offering. Xert (as highlighted by Allen/Coggan/McGregor) is the clear early leader in offering adaptive training in this space. And as stated elsewhere in the Forum, TrainerRoad is heading in the same direction. I think a merger of the two companies is unlikely as the TR team (hinted elsewhere as well) believes it can bring additional value than what exists today.
Market 2, and specifically Training Peaks (i.e. WK05) and perhaps GC, is heading in the opposite direction. And that was very clear from Tim Cusick’s webinar today - that WK05’s objective is to provide more and better tools for the coach and self-coached athlete. I think adding adaptive training as a “feature” would alienate its customer base and be counter to all of their recent initiatives, including making modeling the anaerobic battery (dFRC) and better understanding the impact of a workout (TIS) available to its customers.
1 Like
The one area where I think Xert could do better (actually all models) is using machine learning to tune the decay rate parameters (e.g., how many days is you CTL an average of, how many days are used to calculate ATL, etc.) plus “notice” when you are doing sub-threshold (e.g., sweet spot) training blocks and not decay your calculated FTP.
It wouldn’t make sense to use the Xert calculated FTP with a TrainerRoad sweet spot plan, as Xert would be continuously lowering your FTP throughout the plan.
3 Likes
I’m taking a similar approach at the moment. I agree that Xert’s library of workouts is limited and the recommendations don’t always correspond to the “Training Advisor”. I am trying to see how Xert classifies certain efforts- like 2-3 min VO2, 5 min threshold efforts, etc. For example, When they recommend a rouleur workout I assume 6 min threshold (or just above) efforts would do the trick- but they often classify those as Sprint TT or Climber for me.
I tried the free trial of Xert, and found it to be frankly awful.
They used their modelling to significantly complicate, instead of simplify training. Nothing seemed to have any good explanation behind it. It has the germ of a good idea inside, but the execution for me is just all wrong.
A Rouleur workout would be basically intervals at your 6 minute power in Xert or similar to a VO2max workout
1 Like
The other day I did Red Lake +4, so 6min intervals at 108%. Xert classified that as “Sprint TT”.
Intervals at my 6-min power, which for me is 125% of TP, would make for a super-painful workout. Give me a decent warmup and I could hit those numbers, but I would not be able to repeat it.
Thats about right as SprintTT is classified as your 20min power so 108% dependant on your particular strengths would be correct. My 20 min power is 107% of TP.