Smart trainer vs power meter accuracy

Has anyone compared smart meters power output with some other power meter (like Garmin Vector)? FTP test on my Elite Realtour gave me yesterday 4.2W/kg … which sounds too good for me, I guess a number below 4 would be more realistic.

Also my average speed is a bit higher in Zwift than outdoors - but I can’t judge this as I’m not slowed down by traffic, turns and corners etc + it’s relatively flat comparing to my usual routes.

It can vary a little but if a Smart Trainer is properly cablibrated the answer is ‘close enough’. 

And yes my average speed is higher on Zwift too, but then I think it’s realistically what I would get outside if I never had to steer, or brake or worry about ever going too fast for the situation.

1 Like

Well - the key is in “if a Smart Trainer is properly calibrated”. 

I don’t think I can verify this myself without external powermeter, so I have to trust Elite

Power is very subjective:

Powermeter to powermeter it varies.

Different powermeters get power in different ways.  You should read some of dcrainmaker reviews for the details of this.


for example the kickr when tested against others.

I have a Tacx Neo smart trainer and, while I have no basis for power comparison with anything else, my performance in Zwift seems pretty close to real life, bearing in mind the default 50% difficulty on ascents. Maybe that’s why people appear to be performing better than real life. Perhaps try setting trainer difficulty closer to 100% and see if that slows you down enough. :slight_smile:

I’m going to comment but I’m not sure what I have to offer is directly relevant to the OP’s question.But here’s my take…

I use a Kickr and have trained and raced with power meters for over ten years.

Once a spindown has been performed, the Kickr creates a similar “feel” and output as the various PMs I have at hand. Using Trainer Road I have no fear of being misled. I can and have done an hour on the Kickr and TR and then gone out and done an hour in the real world with a powertap and both efforts numbers have been really close or basically the same.

I have noticed the extra speed in Zwift too. I’ve input my real weight and have the “Trainer Difficulty” setting at max. I have given this extra speed a good long think and maybe it is accurate for Zwift, after all it is, again, not the real world.

But we are trying to create a simulation as close to the real as possible so I would like to figure out a way to get the speeds down to what I consider real world speeds. Maybe I need to add a few pounds to my weight and take trainer difficulty down a notch or two. Maybe.

Zwift seems also to over report my wattage a bit but it is difficult to say if this is true too. Indoors, I tend not to hold back. Using Zwift, I can just bury myself and just step off onto the couch. I can’t do that in the real world. In the real world I have to get home. Lol. So maybe that’s all good.

But here is the interesting part; When I compare my Zwift ride data with rides done out in the real world they are very similar.

The Zwift rides have higher peak power output numbers but overall they have about the same TSS, average watts per time periods, etc. That extra speed shows up as greater mileage. NP and IF are a bit higher due to the higher peaks but I think these are valid-ish numbers. Like I mentioned, I find it far easier to damage myself on the trainer then I do outside and NP and IF reflect this.

So am I comfortable adding Zwift data to my outdoor training data? That’s the real question and I am still thinking on it. For now, the rides I do on Zwift are just to keep some pedaling in my legs while I’m stuck indoors. These numbers aren’t all that interesting - it’s just volume. But in a few weeks the numbers will be important and hopefully by then I’ll know if I can trust the Zwift numbers to reflect the outside world enough to add then to WKO+.

1 Like

I am a relative newbie but really appreciate Alan’s comments.  I am an informatician in the health care setting and often compare large data sets across institutions, including laboratory values, histology, etc.  Even with highly trained staff, tuned instruments and reagents, the results still vary - so it is no surprise that Zwift vs reality don’t exactly match.  

The question is, can modifying settings within Zwift (increased weight, difficulty, etc.) be used to correct for the variance between approaches.  I wonder if Zwift (or anyone) has systematically investigated which settings would provide the truest indoor vs outdoor comparison?

As Alan suggested, having the trainer difficult at/near 100% and adding additional weight to the person riding could help align the results?

btw: if I’m not happy with my smart trainer power readings, is it possible to adjust it? I guess it should not allow me to increase power, but decreasing it should be allowed … ? Increasing my weight is one option, anything else?

(I don’t care so much about KOMs, I’d rather be seeing numbers similar to my outdoors experience)

Not being happy with your smart trainer power readings is an idea fraught with complications and controversy. lol.

I’ll interpret “not being happy” not as not trusting but as not producing what you feel is an accurate Zwift “experience”. I suppose you could interpret not being happy as insulted by or overly encouraged by or “WTF, dude?” but that’s not your or my situation.

Regardless of any of that your power numbers are supposed to be THE independent variable amongst the raft of dependent variables you’re using to measure and manage your performance. Having it available to manipulation takes the independent part out of it. Right? So you shouldn’t be looking to mess with your power meter’s numbers. By all means calibrate the crap out of your power meter but don’t seek ways to manipulate them.

But then how to get what you want out of your Zwift rides?

Increasing your weight in Zwift should slow you down but I don’t know how or by how much. I guess our only option is to make the changes and see what happens. My gut (see what I did there) tells me I’ll be slower to accelerate and slower to climb but see faster downhill speeds too. Will I net slower over a few laps or will I net no change? I’ve yet to give it a try but I will at some point.

I have seen some crazy power spikes too. These I have been editing out of the data in WKO+. The occasional 1100+ watts for one second stuff deserves to go. I can only imagine the weird ANT+ programming that allows that to happen and I have little hope that it will be or could be coded out so I’m fine editing them.

Right now I’ve been tweaking the Trainer Difficulty setting - you know, one thing at a time. I took it down two? notches from max today. I only noticed I was shifting more on the climbs but not moving more slowly but that needs to be validated with a few more rides - I might just be tired today, just a little.

Of course the real hope is that the Zwift programmers will offer us users a few more buttons or sliders so we could adjust our specific trainer set ups to reflect more of the real world. I would hope that’s in the works but who knows, maybe that would open too many doors to Zwift doping or some such monkey business.

I have a Tacx Vortex Smart and Powertap G3 and initially the power numbers seemed to be off significantly. After about a week, the numbers ran really close. I mostly do training mode so I watch the watts closely. I took my G3 off so I would stop wearing out the bearings and put on an old, bombproof, wired powertap and the watts are really close. Zwift drives the ERG mode perfectly and my power and heart rate graphs look good. I am convinced it is perfectly adequate for training during the off season. 

My personal experience with a Tacx smart trainer is not good. My Power2Max was reading around 240w while the Tacx PM readings were around 360w. Using Tacx on Zwift would have made me a cat A rider while according to my P2Max I am a bottom cat B rider. I decided to swap the Tacx smart trainer for a fluid based regular one and go with my P2MAX readings which are I believe the accurate ones. If it happens that Tacx was right I am going for a pro contract :slight_smile:

Zwift racing seems to be dominated by “Smart” Power meters and digital doping is very real. 

The biggest proof against ‘smart’ PM readings is the fact that some of the riders who actually win races on Smart PM don’t even sweat much. Once guy averaged over 4w/kg won a race with his maximum heart rate barely braking 150bpm while I was around 40th with an average HR of 166bpm and I am over 50.

The winner might have been Cavendish in disguise but I am quite positive he just tinted his “Smart” trainer to read high. 

I got the Power2max NG and the Tacx Flus, working with both there are Power2mas  30W more than of Tacx flux. This is happenening after 170w. Both have been calibrated… now Im not sure to which system to follow… both has the same weigh.

Any idea/ help?

Hi Oscar
The actual watts and w/kg only matters when competing virtual with others. I’ve tested a fesew Tax smart trainers and compared them with my Garmin pedals. The output does differ! Some trainers more than others.
But for personal training purposes the actual number is irrelevant as long as you allways measure with the same tool/trainer.
With proper effort and some structured training, you should see that the numbers are increasing - and this will show on the streets in the real world.

1 Like