RGT vs Zwift: A fluid mechanics professor (and Cat. 1 racer) compares the two

We examine the verisimilitude, fairness, and in-game physics for racing on the two virtual cycling platforms.

Heading out the door? Read this article on the new Outside+ app available now on iOS devices for members! Download the app.

Zwift is the largest virtual cycling platform, with as many as 49,000 users riding in the software simultaneously. Thousands race on Zwift daily, and the platform is hosting the UCI esports world championships later this month. But the RGT Cycling platform also hosts racing events and continues to build a base of users. RGT hosted a Team USA selection race for esports worlds, and this weekend is hosting the USA Cycling esports national championships.

So which platform is better for racing? We asked Dr. Shawn Litster, who teaches fluid mechanics at Carnegie Mellon University, in Pittsburgh. Litster, a Cat. 1 road racer who twice represented Canada at the world junior championships in downhill. He regularly races in Zwift and RGT. His response? Define what “better” means.

Also read: What you need to know about the UCI cycling esports world championships

Zwift is far and away the most popular and better-funded platform. Besides the plethora of race options, Zwift has various virtual bikes, wheels, helmets, shoes, eyewear, and apparel to dress — and speed up — your avatar. Zwift’s ZRL series alone has more than 19,000 racers participating every week. So, if better is defined as a bigger community with more events and more options, then Zwift comes out on top.

But what about game performance in terms of real-world simulation?

Studying aerodynamics

Litster said both games use simple models for simulating aerodynamic drag. A taller rider will have a greater surface area, which means more drag and slower speeds at an equivalent watts/kg than a shorter rider. Neither RGT nor Zwift accounts for rider aerodynamic positioning. Each rider is given a default setting, regardless of how high or low you ride outside.

Annemiek van Vleuten on her way to winning the Olympic women's time trial.
Zwift’s aerodynamic modeling of taller riders — who have a greater calculated surface area — requires them to produce enough raw watts to overcome the greater drag associated with their greater relative height if they are to go faster than shorter riders. Just like in real-world racing. (Photo: Martin Rickett/PA Images via Getty Images)

“Where the forces they include are likely drag force, gravity, traction, and rolling resistance. Traction would be where a rider’s power comes in – calculated from input power, wheel radius [wheel-on trainer, –ed], and rotation rate. The mass would be the rider plus the virtual bike,” said Litster.

Litster said tall riders in an aero position often beat much smaller riders in the same position for the same reason this happens outside — greater raw power. If a taller rider on a flat course produces the same watts/kilogram as a shorter rider, all things being equal the taller rider will go faster.

Also read: How much power do pro cyclists produce while training?

Litster said that with respect to modeling aerodynamic drag, the favor tips to a smaller rider who can do bigger watts, since smaller riders have less surface area, and neither Zwift nor RGT accounts for riders being in aero positions on their virtual bikes. He believes that both Zwift and RGT use a model that calculates rider drag based on a model that “sees” riders as cylinders.

“Likely, [Zwift] models CdA as a non-interacting sum of separate components,” said Lister. “It’s unknown to me if Zwift scales bike CdA with height,” to accommodate a relatively larger bike, in-game, for taller riders. Lister provided equations that he suspects both Zwift and RGT use, but could only speculate about their application in the respective platforms.

CdA = CdA_frame + CdA_wheels + CdA_rider

“Usually, for complex shapes, we don’t distinguish Cd and A and we just use CdA,” he said. “The drag force would be calculated according to the equation:

Drag force = (1/2)*Cd*A*ρ*V2

Cd = drag coefficient, A = area, ρ is air density, and V is velocity

The drag power is just the force times velocity, or:

Drag force = (1/2)*Cd*A*ρ*V3

The A is projected area, which means it is the area that if you looked straight on at a rider, took a picture, made an outline of the rider, and measured the area of that outline,” clarified Litster.

In trying to understand how the game behaves explains that Zwift may be using some conventions, to standardize the game behavior and performance across a wide range of riders’ inputs for height and weight, as well as in-game bike and wheel selection.

How can you go faster in virtual cycling?

Raw watts — and timing a winning sprint — are both important on flat courses in Zwift and RGT.

“I assume Zwift is treating Cd as a constant for a given bike/wheel in CdA and scales it with height and weight based on how those would scale with area. I doubt they consider any velocity effects on Cd. So there is likely a CdA scaling factor for equipment versus some reference and then they scale versus a CdA reference for area,” Litster said.

Both Zwift and RGT both require riders’ heights to be input into the game to determine aero drag (and penalize accordingly). Litster indicated that both of these platforms are not doing anything “terribly innovative or outlandish to model drag coefficients,” and handicap riders based on their dimensions. So which platform comes out on top with regards to better simulation of aero drag? It’s a toss-up. Through the fall of 2021, RGT seemed to approximate the real thing with better accuracy, he said, but with the multiple updates Zwift has pushed at the end of 2021 and beginning of 2022, both games are equal in simulating drag.

“I think that as long as you’re putting in your correct weight and height, you’re going to get a pretty realistic number,” said Litster. “If you’re somebody who is a flatlander or time trialist or criterium rider: hit the gym if that’s going to give you an increase in your power output.”

“A little bit of increase in CdA is going to pay dividends in the watts per kilogram reduction that you’d need to do if that mass is contributing to your power,” he said. So, greater body weight can be an advantage on flat courses if the greater weight results in greater power output, otherwise, greater weight is being modeled in-game as greater surface area, which means additional drag for riders to overcome.

This is a position echoed by VeloNews training columnist Zach Nehr, who won the Pan American qualification race for the 2022 UCI esports world cycling championships and will be representing the United States at the UCI event in late February 2022. Nehr said it’s still pretty simple: more power means more speed.

Drafting: Which platform does it more realistically?

The CMU physics professor said he thought Zwift and RGT use simple models for simulating aerodynamic drag and drafting.

“Roughly treating a human as a bunch of cylinders, you can get an idea of how CdA should roughly scale with height and weight.

If we have the following for a cylinder:

Cylinder volume = π*(D/2)2*L
Cylinder area = L*D

Height and weight can be calculated as:

Area = 2*sqrt(V*L/π)

“So, riders’ area scales with the square root of weight and height, treating weight and volume as equivalent, and not distinguishing density differences,” Litster explained.

A visualization of how Zwift models drafting. Darker red offers the most draft, while the yellow and blue indicate where some draft is still being created by a rider’s “wake.” (Illustration: Zwift/GCN)

“Zwift appears to use a triangular-shaped wake — approximately 30 degrees — when modeling group drag, which decays over roughly 10m,” explained the CMU professor. ”This likely starts at around 30 percent reduced area drag and decays from there. Then, those individual wakes seem to be superimposed (linearly added) for many riders, which increases drag reduction and broadens the draft zone.”

This means that — even more so than in real-life riding — the advantage to riding in a large group in Zwift will mean overpowering those riding in a smaller group. If you’ve ever tried to solo off the front of a group in Zwift or RGT, you’ve experienced this.  So which is better at modeling group drag-reducing dynamics? Like modeling individual drag, Zwift and RGT are about equal, he said.

Climbing routes vs flat routes: which game does it better?

When climbing in virtual cycling, watts-per-kilo is the all-important metric, and both Zwift and RGT use a relatively straightforward calculation, Litster said.

“When we are going uphill, we are increasing our potential energy:

PE = mass*gravity_constant*vertical_height

Likewise, the force of gravity is modeled as:

Fg = mass*gravity_constant

The power is rate of potential energy increase, or force times velocity:

Power = mass*gravity_constant*vertical_velocity

Zwift allows you to hide the heads-up display.
What’s the secret to climbing faster in both Zwift and RGT — and outside? More power with less mass.

Litster said rider height influences speed more in Zwift than in RGT, even while climbing, but all things being equal, a rider’s power-to-weight ratio is the biggest determinant of speed. A taller rider who generates a higher w/kg than a shorter rider with the same mass will climb more quickly.

So which virtual cycling game comes out on top with regards to modeling climbing? This is another tie.

“There is very little uncertainty in this modeling compared with aerodynamic drag estimates and rolling resistance estimates,” Litster noted. “You could assume that climbing in Zwift and RGT is going to be more realistic than riding on flats in terms of virtual performance compared to real-world performance — at least in terms of power output conversion to bike speed.”

When asked about how it models aerodynamic drag and drafting, Zwift had not responded to VeloNews’ request for confirmation of Litster’s hypotheses.

Dr. Stefanie Sydlik is a colleague of Litster’s at Carnegie Mellon who races for Canyon eSports and will be representing the United States at the 2022 UCI esports cycling world championships in the women’s race. Dr. Sydlik said the differences in RGT and Zwift are noticeable, including that RGT does not appear to scale aero drag by rider size. On the other hand, RGT tries to simulate real-world cornering with deceleration and acceleration into and out from corners in the game. Zwift has no such feature.

“RGT factors in breaking into corners and positioning, but no drag scaled by size,” said Sydlik. “The Zwift algorithm approximates aerodynamics with weight and height — which roughly works — and may favor someone lanky like me more than in real life.”

Gamification vs real-world riding

Zwift also has things called Power-Ups — tokens riders pick up while riding and can deploy to temporarily reduce weight or drag, or increase the benefit of a draft.

RGT, on the other hand, tries to faithfully recreate dynamics like slowing down for corners, which Zwift does not.

Zwift’s commitment to fairness for high-level racing is unmatched by RGT. For premier races in Zwift, riders must provide videos verifying their height and weight, uses secondary power meters, and submit outdoor field testing.

While RGT has a weight verification process — and requires riders to submit to a fairness protocol in order to compete at the USA Cycling national championship event — it doesn’t have outdoor verification requirements.

Verdict: Zwift is still on top

So which platform is better for racing? RGT’s power-requirement adjustments for in-game braking and cornering are the biggest differences between the two virtual platforms. But otherwise the modeling between the two is quite similar, for riding flats, climbs, descents, and in groups. Further, Zwift hosts thousands of events, with deep customization menus, superior graphics, and in-game chatting. In short, Zwift has several advantages.

An American in France

What’s it like to be an American cyclist living in France? Watch to get professional road cyclist Joe Dombrowski’s view.