Power meter accuracy comparison - Down the rabbit hole

Yep. I would say that the lesson is that any particular individual power meter can be off, even ones from models or brands that have a good reputation, so it’s good to have a way to check. I think dynamic tests are better but they’re also a much bigger PITA so I don’t do them very often: typically, when I get a new power meter, and then I just monitor the readings over time and if they ever seem wonky I’ll go to the trouble to do another dynamic test.

1 Like

Except TR wouldn’t be able to do this analysis. All they would be able to determine for people using power match is how their trainer’s reported power compares to their power meter’s reported power. Not the accuracy of each against a “gold” standard of “reality”, but the comparison between 2 sources of power, each of which is of unknown accuracy.

Over a large enough data set you come to a conclusion of which devices most likely will measure close to another device, and which combination (smart trainer and power meter) of devices will most likely closely track each other.

My trainer is linked to my pedals, the trainer is linked to TR so TR thinks the power is coming from the trainer. Another reason TR analysis would be imperfect.

Yup, that’s what I did, too.

Initially, my Quarq and my Elite Suito had very close power numbers (within a few W). After a year or so, I checked again (even though I used power numbers from my Quarq for training) and there was a clear offset and difference in slope. I thought about what I could do to check who is right. In the end, I just decided to “believe” my Quarq and repressed thoughts of needing to know. I don’t think my wife would be happy to hear that I need to spend €€€ on power meter pedals to check which of the other two sources of power data is right (or closer to the third power meter … :wink:).

1 Like

Yes, but you could see whether there is a mismatch and over a large sample size, you might be able to draw conclusions.

E. g. if a statistically significant portion of athletes that own an Elite Suito (to pick one trainer) and another power meter of various makes and models see a deviation between the two reported power numbers over roughly the same time period, then this would suggest that Elite Suitos have a drift over time and that their reported power figures are inaccurate.

You could also reverse that question, i. e. fix the power meter on the bike (Quarq, Assiomas, etc.) and look at whether patterns emerge if you compare the on-bike power data across various trainers.

However, I doubt TR has the time to crawl down that enormously deep rabbit hole.

With a big enough sample size, there are definitely conclusions to be drawn. Especially if you know that certain trainers are generally quite trustworthy (e.g. Tacx NEO series, Elite Justo, newer gen KICKR’s, etc…). Those units that can’t be easily (if at all) be dorked up by consumers and bad calibrations.

To that end, Zwift’s anti-cheating folks (the same guy behind Indievelo) was very clear about them doing this type of big-data exercises around trainer models, even if they aren’t applying the learnings to the general public, it helps them sniff out weird pro results.

The problem both TR & Zwift have, is simply keeping the trainer companies happy relationship-wise. Though, I’ve always found that funny, because if you say the Tacx Bushido is crap and to buy XYZ new unit instead, then people will go buy that new unit - which may very well be a Tacx NEO or whatever.


This margin for error is why I prefer the accuracy of training with heart rate.

Rabbit hole #2


Because heart rate sensors always agree … :wink:

(I’ve been comparing numbers between my Polar H10 heart rate strap and my Wahoo Rival since I got the Rival …)

I feel like I have no clue about the business side of things, but I would assume the most popular software people couple their trainers to is Zwift. Amongst the high-end, I would assume TR is well-represented, too. Shouldn’t the incentive be the other way around?

These days that’s probably true. In the case of TR, it probably doesn’t matter as much. But in the case of Zwift, Zwift depends on these companies to produce units for people to ride. To a certain extent, they don’t want to enter a bloodbath there. That said, half the time the trainer companies are upset at Zwift anyway…