Still waiting on an ultra light rider update

then i will give you a cautionary tale because the first and last time i ever called someone out on zwift was a 47kg adult. it turned out he was actually anorexic. his power numbers were undeniably BS but it really doesn’t matter. there are a million other people on zwift at every weight with BS power numbers too. i still feel like ■■■■ about it two years later

i should have known better too because when i was in my early 20s i weighed 48kg myself


Don’t young riders have quite high heart rates?

Some of the light young riders at 55kg can and do put out very big watts relative to their size. They started riding very young on fixed gear bikes and racing and just got better over time.

As i said I was at 50kg in my 20s, and my biggest son is 17yo, 174cm, 53kg with a 267w ftp now (e ftp on intervals). Still a B though (245w for zwiftpower).
And at one time he was 43 kg with 217w ftp if I remember well.
So this kind of numbers can be legit.

And irl a teenager with these numbers is very fast uphill : on small ascents (600m of d+) he was way faster than me at this age. On the ventoux or the galibier though, I was faster (he couldn’t start slow, so he burnt too many matches too fast).

So maybe the physics are not wrong.

Well when I was a 52kg junior I had climbing strength and acceleration to match senior/cat 1/2s, and beat most in climbs; and could hang with senior/cat 1/2 race packs and compete in bunch sprints for lower placings… up to around 2 hours max. But the weakness of lightweight juniors is inescapable - their absolute physical strength and endurance is that of a child’s/young teenager’s. And the lighter the more of an issue it is when racing seniors at senior/cat 1/2 race lengths.

Your son at 43kg is probably at the limit for having the physical strength to hang around and compete on suitable courses/race lengths. 37kg -15lbs- lighter than your kid (which for top flight juniors would be 0lbs fat and 15lbs of muscle mass)… 37kg is simply too light, not enough muscle to have any sort of sustained strength/endurance versus senior riders. 37kg/81.5lbs is simply incapable IRL of competing against cat 1/2s on any course, nevermind competing with pro riders.

1 Like

That’s my informal observation as well. I think heavier riders climb better than they might in the real world and lighter riders are strong on flats and maybe even sprints. I have no proof but I too think it’s meant to keep riders together and enhance the social aspect of the platform

1 Like

I agree… this was a few years ago, and his endurance has totally changed (as his weight…) now!

1 Like

I have looked into some research on how CdA can be estimated, and both Cd and A are based on weight. For A(rea), height is also a factor. Some of the main research papers are using a small number of partisipants who are adults with weight in the 60-80 kg range. If the formulas from these studies are used, I’m not surprised if CdA for weights outside this range are highly inaccurate.

EDIT: This is maybe why RGT has equal CdA for all riders (uncertain about this)
EDIT2: This may also be the reason why I can’t find any ZwiftInsider speed tests with riders outside this range (60-83 kg)

It would be great to have an ERG-based test on Zwift, using a perfectly flat course with drafting effects disabled, to benchmark speed versus weight/height/power. Could make it a kind of speed/power ramp test?

Be nice to know where the model starts to fall apart (though Zwift probably know this already from modelling at the extremes?)

1 Like

It has been discussed before on the forum that tall and heavy riders are unfarily punished in terms of CdA on Zwift, here for example: CdA dependency on height issue also a user mentiond that he had looked at this and is planning to publish a small article in another thread Pack Dynamics Test Events (December 2022) - #1020 by Ulf_Quentin

My assumption (we know about assumptions) is the base formula was created and verified using pro and Cat 1 -2 racing cyclists that have a lot of data

My guess is that when they looked at this data to determine how height and weight affects CdA they made the mistake not to isolate each variable. That is they just looked at how weight affected CdA by looking at the power and weight of each rider and they did not take into account that heavier riders are usually also taller. Then they did the same for power vs height to determine how a riders height affects CdA. This means that the resulting model will be off and punish taller and/or heavier riders with too high CdA compared to smaller riders. This is just a guess but it would explain why they got the model so wrong even though they looked at real world data. Zwift should really look into this since they are losing users to RGT because they have better “physics” modelling @James_Zwift @DavidP

This means that the resulting model will be off and punish taller and/or heavier riders with too high CdA compared to smaller riders

And it is true that taller/heavier riders are “punished” in real life cycling because of that correct? How do you determine how much is too much and what is the boundary?
Because from your suggestions we would need to apply some magic transformations for the extreme ends of the scale.

How often is this a problem in racing events compared to all the other problems?

Yes, taller and heavier riders have a higher CdA than smaller riders. What I am saying is that in the model that zwift uses the difference is too big. I am not suggesting anything about doing something specifically for the very ends of the spectrum of rider sizes, I am saying that the whole model is wrong. It is likely most apparent for the smallest riders.

I don’t have the time to look up the available data to support this claim right now but I could do that tomorrow if you are interested. @DavidP

As for how often this is an issue in racing events I would say pretty much all the time except when going up a steep climb. One of RGTs main selling points is that they have better physics modelling than you guys and judging by how much they are growing I would say that probably it is working.

Interesting :thinking: In another forum thread related with pack dynamics there are some users claiming that the Zwift model (for solo riders) is super accurate.

1 Like

I have not read these comments but it depends on what exactly they mean is accurate. Probably they mean that given a specific CdA and power the resulting speed is accurate which it might be. What I am saying is wrong is how CdA is depending on height and weight.

My theory is that the CdA formula falls apart (is less accurate) outside the “normal” range for weight and height.
It is not a big problem unless people find some WTF cases like the one above. If it works too differently to IRL it becomes a problem. I’m also a bit surprised that very light riders can hang on and do very well in B cat crits together with 90+ kg riders pushing high watts.

That is for a single rider in the “normal” weight range. It is also compared to an online calculator that is probably also using a similar formula to calculate CdA. (FAQ "Speed and Power Calculator for different bicycles")

1 Like

I thought I read somewhere that RGT doesn’t use rider height as an input at all? Or if they do, it’s maybe (?) based on a table of average heights for a given weight?

I would suggest ignoring the issue of larger CdA magnitudes for bigger riders completely makes RGT’s physics modelling worse than Zwifts. At the moment their drafting model (as opposed to base physics model) may be better in terms of not having the artificially high pack speeds, but hopefully PD 4 will eventually address this issue (I use both RGT and Zwift and I find RGT gives an unfair advantage to large high power riders on flat to rolling circuits anyway due to just ignoring the large rider CdA issue).

Just did a Google search:

Q&A: RGT Cycling’s James Vickers on virtual cycling experience 30. mar. 2022 — Our approach to CdA, the coefficient of drag, is very similar to Zwift’s approach. We made the decision not to include rider height. We’re not

It is a paid article so I can’t read it, but it confirms that RGT does not use height to calculate CdA.

This would be my assumption as well. So they use a simpler model with just one parameter, that does not mean it is less accurate compared to real world data. To be honest I don’t know much about how RGT model this but that does not matter. My point was that RGT is advertising that their physics modelling is better and it seems what they are doing is working for them. So in my mind it should be highly prioritized for zwift to make sure their models are accurate.

I agree that it’s unknown what RGT or Zwift for that matter, are doing under the hood. However, I’d leave open the possibility that RGT’s approach may be more popular, while perhaps being less accurate. IOW, there are a lot more average riders of average height therefore NOT complaining about the physics, while at the same time there’d be much fewer edge case riders who would be complaining. Inverse situation at Zwift though.

If RGT has ERG based custom workout capability (?) this should be simple to test differences by doing various fixed wattage speed tests in one platform vs the other, while in Zwift changing height between each workout.

From my experience, I would not call the child physics issue extremely common.

It’s reasonably frequent though. 1 in 10 races maybe?

Pack speeds are obviously, a much higher priority. However, that doesn’t make this a non issue. It needs to be addressed at some point.

A 37kg rider nearly beating one of the best Zwift racers on Earth is a clear example of how broken it is. This isn’t a 37kg rider competing against cat 3/4 riders. This is a 37kg rider competing against a world class professional athlete.

Which is totally impossible.

The fact that it is possible on Zwift is the issue.

Personally, I wish Zwift took more care of their customer base.

I’d like to see an automated system that flagged performances such as this and immediately contacted the rider to address the validity. In this case I suspect the rider hasn’t updated their correct current weight, or they have an optimistic power meter. Or it’s Remco’s younger brother :joy:

Any rider exceeding 6w/kg for 20mins indoors is exceptional. Doing this at 37kgs is not something I’ve ever seen.

I’d love to see performance testing of the physics of sub 45kg riders on Zwift. I’d be very surprised if the performance reflects reality.

Couldn’t a simple table search of all Zwift racers weights and ‘hypothetical’ performances find these issues in a matter of seconds?

Set a logical upper threshold, then require verification?