Pack Dynamics 4 Release [April 2023]

I’m not sure what’s going on, but my experience is that pack speeds haven’t really changed in comparison to solo riders.

Large pack at 3w/kg - Zero humans doing anything over 3.5w/kg. Certainly nobody on the front doing sustained 4.5w/kg+

Solo rider at 3w/kg - WAAAAAAAAAAAAAAYYYYYYY slower.

That is the issue.

Could someone not simply run an experiment to demonstrate the issue?

Is it churn? Is it lack of collisions? Is it one of a million other things?

I have no idea, but it’s still fundamentally broken. In my experience.

Zwift Insider did run a test. It’s the churn. A pack of riders at steady state is faster in PD4 than in PD3 because of 1.5draft. The PD4 algorithm only slows a group if people let off the power when they hit the front.


Well, I think the situation with the most recent ZI article is in the corner case where everyone is pushing the exact same watts so there’s absolutely no autobraking at all.

I ‘think’ this is the article you’re referring to correct? If so, this was the conclusion:

What It All Means

For road racers on Zwift (vs TT), Zwift’s new pack dynamics should result in races that more closely reflect IRL racing.

Double-draft is enabled for all races (since it’s built into PD4), so the draft is delivering more power savings than before.

Additionally, with autobraking and other changes accidental churn is decreased on the front of packs, keeping pack speeds a bit lower and forcing riders to be more intentional if they want to take a turn on the front.

These changes should lead to more breakaway possibilities, as smart breakaway-loving riders can get more rest in the draft, and work together to stay away from the slightly slowed pack once a breakaway is established.

That’s precisely the crux of the issue: PD4 does not actually require that riders at the front of the blob work harder than before to chase down a break, as PD4 can only do its job under very specific circumstances.

As the ZwiftInsider tests show, if riders just continue with the same steady churning behavior at the front that many are used to from PD3, speeds are actually faster in PD4, as auto-braking is not activated under those conditions, while draft has been increased.

Ironically, it’s primarily experienced riders that actually know to back off before hitting the front who are affected by the auto-braking, while the mindlessly churning masses are affected to a much lesser degree.

In other words, PD4 has been designed to work most effectively in those cases and for those riders, where it’s needed the least.

And it’s been designed in such a way that it’s quite easy to work-around and avoid its effects by consciously changing your style of pedaling to be less surgy and more steady-state, which is not necessarily a harder effort depending on rider physiology and preference. If we’re to judge PD4 by its game mechanic merits, it appears to incentivise an arguably less interesting and skilful way of riding.

To add insult to injury, the increased draft in PD4 makes it harder to establish breaks in the first place, as the power difference required to snap the elastic is greater than in PD3, which means you have an easier time closing down attacks out of the pack.


[quote=“Johannes [FUSION COLOQUICK], post:366, topic:605018, username:Johannes_K”]
In other words, PD4 has been designed to work most effectively in those cases and for those riders, where it’s needed the least.

The Truth, the Whole Truth, and Nothing but the Truth

1 Like

So, in light of what Zwift can easily tweek (or are willing to tweek), what could be done to improve this?

Would it be possible to reduce the draft cone (shape and/or intensity) for the first few rows of riders?

Adjusting rider CdA values? Surely this would be a quick fix to bring down the speed of all riders by x%?

For reducing descending pack speeds, wouldn’t the proposed “corner braking” (currently being worked on) change speed and dynamics significantly?

1 Like

I guess it was already said and explained, and I can confirm, the so called “churn” is the responsible for the higher speeds of the packs, and with PD4, like the ZI tests demonstrated because of increased draft if riders keep a constant pace the speed will increase.

So, in light of what Zwift can easily tweek (or are willing to tweek), what could be done to improve this?

I’m currently exploring with what Dejan previously suggested, of implementing a more aggressive type of churn prevention, by only allowing to overtake to the front (someone not drafting), if you overcome the power to maintain the same speed if not drafting. So it won’t take you away the drafting but if you want to go to the front of the group you need to put out more power (net power, not absolute w or w/kg) than the person in front.

Adjusting rider CdA values? Surely this would be a quick fix to bring down the speed of all riders by x%?

Also currently exploring with that. The current PD4 system was prepared to configure increased CdA values if we want to. But that will bring the problem of obsolete records everywhere. And while for you and me it may sound totally fine, trust me, for a lot of folks out there, it’s not…

So I can confirm I’m working on a possible set of PD 4.1 improvements, only testable on racing events, at least initially. ETA unknown.


I LIKE :joy:


So, the ultimate aim would be to roll-out these PD4 changes universally as the new default?

Too early to say

This is how my “force advancement on the outside” suggestion is implemented in a system where you can’t actually force people to the outside. Might work out well or might not (similar to ST elimination) but worth trying. Might be too aggressive. But worth trrying. It’s ultimately a video game and most-fun and sim-like realism don’t always line up.

I’ve offered to Gerrie to formally field all complaints about ‘obsolete records’ so don’t worry, that’s covered.

1 Like

Eric’s conclusion is irrelevant. All it takes is for individuals to hold relatively steady power and the pack churns and goes faster than before. Doesn’t have to be the exact same watts.

1 Like

So lightweight riders like me won’t ever be allowed to go in front of the first rider of the pack??? I am 56 kg, when I do 300w it is a big effort, allowing me to ride irl at more than 40 km/h, but when the first guy is one of those huge 80kg beasts, 300w is nothing for them… I will almost never have more net power than the others.

No, you didn’t understood what I meant by net power. Net power is the power required for you to accelerate depending on the opposing forces. For smaller/lightweight riders, that have lower CdA values, the required net power will be obviously lower. So don’t worry, you don’t need to put out more power than 80kg riders :slight_smile:


I think David is saying if the pack is going 40kph, you’d have to put out however much power you would require to go 40.1kph or faster.


Thanks! I am relieved !

RIP ur inbox


I’d love someone to explain the math to me.

I’d assumed that given drag force is squared in relation to speed, that universally having such fast moving riders in game, further reduces the chances of breakaways etc. Given that an average Zwift race pack is already doing say 45kph, a solo rider would need to do crazy watts to stay away.

By increasing everyone’s cda to a more realistic number, would that really make any difference? Would it not just mean everything happened at a slightly slower speed? There’d be a tiny percentage advantage, as rolling and mechanical resistance would be a greater percentage of every riders power at lower speeds.

I’m 100% for lowering the average cda to better simulate average riding conditions IRL. Although as David said. There’s going to be a lot of people who freak out about records etc. I think these folk are taking this whole ‘records’ thing way too seriously, but that’s a sperate discussion.

I think in the long run, accurately modelled physics will be a very positive step. It’s going to happen eventually, so it’s definitely better to make these large changes earlier than later. Any delay, will only further entrench folks into perpetual stagnation.

There will always be dissent with change, even if it’s for obvious improvement long term.

These legacy physics decisions were probably not given the proper priorities when they were initially decided. I imagine having everyone super aero, riding in the most aero position was a cool idea at the time. However, the cumulative result makes it very hard to deliver dynamic racing.

The crux of the issue is pack speed. Somehow requiring only the riders on the front of a pack to match or exceed the breakaway riders effort has to be the target. This is incredibly complex without collisions. As currently we hold say 3w/kg, slowly drift to the front, slowly drift back, repeat repeat. Meanwhile we travel at warp speed.

I believe the target actually needs to be very accurate to real life. Not because it’s better etc. Because, the real life physics work. At 45kph you KNOW you’re on the front of the bunch, you’re doing 400w!!! Your first thought is normally, get me off the front…

That’s not remotely how Zwift feels. There’s close to zero perception of effort increase when you lazily churn to the front of a blob.

I feel that whatever changes, the single guiding principal needs to be to deliver that IRL experience. It’s those physics that make real bike racing so dynamic.

Without powerful draft, we’re all just doing a TT into a sprint.

We need to be able to hide, so we can attack. We also need anybody willing to match our attack on the front of a group to actually have to match it.

Deliver that, then we’re talking fun times :grinning:

Thanks again for keeping at it.


The formula for drag force is: F = 0.5 * Cd * A * p * V^2

Given that the fore first factors are static, we see that the force to ride at a given speed is bound by the speed squared.

NOTE: Ignoring all other factors but the speed in the equation.
Riding at 40 km/h (11.11 m/s) and riding at 50 km/h (13.89 m/s), the drag values will be 123.46 and 192.90. That means the drag makes it 1.56 times harder to ride at 50 km/h than at 40 km/h.

If we could lower the speed of the group from 50 km/h to 40 km/h, the riders would have more energy to accelerate and brake away. Increasing Cd and/or A, the drag force will be increased in a linear way, making it harder to ride at all speeds at the front. Increasing Cd or A may also increase the rider’s ability to stay in a draft, reducing the total force required to ride at a given speed.
So, it will be harder to ride at the front at a given speed, but easier to be in the draft since the area of the riders in front will be bigger.

That said, I hope that the calculation of CdA is reviewed and that the real flaw in the estimation is found, and that they are not just multiplying CdA by a factor for all riders.


Interesting. What are the major anomolies in Zwift’s CdA estimations? Is it the boundary cases (very small, very tall riders etc)?