After months of testing, this week Zwift rolled out their latest version of Pack Dynamics (v4.1) to all worlds. We’ve already summarized how PD4.1 differs from PD4 in this post, but now it’s time to measure those changes and learn how they affect the everyday Zwift experience!
Below you’ll find the results of our standard 4-rider drafting tests using PD4.1 and road bikes. We will follow up with a test using TT frames in a drafting TTT scenario, but we believe the same conclusions will be drawn from both sets of tests. Let’s dig in!
History of our 4-Rider Speed Tests
Our original TTT speed test post from 2020 gave team time trial riders some very welcome guidance about how to ride their races most efficiently. Then in 2021, Zwift rolled out Pack Dynamics 3, and we ran a second test to see what, if anything, had changed. (We found speeds hadn’t changed, it was just harder to hold an efficient single-file formation due the lack of sticky draft.)
In August 2022, Zwift+WTRL announced enhanced TTT features, including the ability for TT frames to draft in TTT events. So we ran our tests using TTT frames. Then several months later in April 2023, Zwift announced the rollout of Pack Dynamics 4 game-wide. So we ran our standard tests with PD4.
This week, Zwift rolled out Pack Dynamics 4.1 gamewide. You know what that means… time for more tests!
Test Goals
We set out to answer three questions with these tests:
- Is road bike power savings in the draft with Pack Dynamics v4.1 different than the savings with Pack Dynamics v4?
- Does Pack Dynamics v4.1 alter pack speeds, and if so, how?
- Are there any other observable differences between PD4 and PD4.1?
Test Parameters and Methodology
All test riders were set to 183cm height, 75kg weight, and rode Zwift Carbon road bike frames with Zwift 32mm carbon wheels.
Tests were done in an isolated event on Watopia’s Tempus Fugit route because it’s the flattest on Zwift and has a timed section (Fuego Flats Reverse, 7.1km long) which could be used to measure the speeds of each test formation precisely.
All tests were done with four riders.
Tests and Results
Test 1: the Churn
For our first test, we put all riders at the same 300W power setting. This resulted in a rather disorganized group of riders, but crucially, there was much less churn than with Pack Dynamics 4. The difference was so remarkable, in fact, that we made a quick video to show it:
The riders would often be 3-4 abreast across the road, with no rider taking the “pull” and no riders benefiting from a significant draft.
- All riders @ 300W
- Segment time 10:36.4
- Speed: 40.19 kph
Segment time with Pack Dynamics v3 was 10:13.4, which dropped to 10:04.5 with Pack Dynamics v4. Pack Dynamics v4.1 makes a huge difference with this particular formation of riders, dropping the average speed by over 1.5 kph!
Test 2: Single File @300W
The second test had the lead rider holding 300W, with the other three riders in single file behind, holding the minimum wattage possible to stay in formation. This is what you would see in an outdoor team time trial:
- Rider 1 @ 300W, Rider 2 @ 226W, Rider 3 @ 202W, Rider 4 @ 200W
- Segment time: 10:36.3
- Speed: 40.19 kph
Notes:
- The “minimum wattages” stated for riders 2-4 on this test and other tests below should be considered approximations, as it is impossible to figure out the precise wattage required to hold formation due to Zwift’s dynamic physics engine and very small undulations in terrain, even on Fuego Flats.
- Riders received power savings of 24.7%, 32.7%, and 33.3% – lower power savings than our PD4 tests, but still significantly higher than PD3.
- As expected, the further back you are, the bigger the draft effect. But the additional draft benefit (only 0.6% more) for the 4th rider is less than we’ve seen with previous pack dynamics, which is a bit of a surprise.
- In a TTT situation with all riders taking equal pulls on the front at these wattages, each rider would average 232W. (With PD4 the average was 225W, while PD3 was 246W.)
- Crucially, test 2’s segment time perfectly matched test 1’s. That means the front rider set the speed of the pack in both tests, and there was no churn/slingshot effect unrealistically driving up the pack speed.
- It’s worth noting here that we did a solo rider test at 300W steady, because we were curious if there was any advantage to the front rider if there were riders behind. There is not. Our solo rider turned in the same time as this 4-rider group.
Test 3: Single File @400W
This test was similar to Test 2, we just bumped the front rider up to 400W, then increased the power of the riders behind accordingly.
- Rider 1 @ 400W, Rider 2 @ 308W, Rider 3 @ 268W, Rider 4 @ 258W
- Segment time: 9:33.5
- Speed: 44.53 kph
Notes:
- Riders received power savings of 23%, 33%, and 35.5% (2nd, 3rd, and 4th rider, respectively). Very similar to the power savings seen in the 300W single file test.
- Once again, the additional draft benefit (only 2.5% more) for the 4th rider is less than we’ve seen with previous pack dynamics.
- In a TTT situation with all riders taking equal pulls on the front at these wattages, each rider would average 309W. (With PD4 the average was 302W, while PD3 was 326W.)
Test 4: Hybrid
Lastly, we tested a formation that many Zwift TTT teams have used, wherein there is one designated rider in front, and the riders behind simply churn in the front rider’s draft. This reduces the hassle of trying to maintain single-file positioning while receiving some of the benefits. But how does it impact efficiency?
- Rider 1 @ 400W, Riders 2, 3, and 4 at @ 308W steady
- Segment time: 9:33.4
- Speed: 44.53 kph
Notes:
- If there wasn’t at least 1 rider holding 308W or more in the pack of 3 behind the front rider, the pack of 3 would get dropped. (This number, not coincidentally, matches the wattage of the 2nd rider in the single file line of test 3.)
- The pack of 3 did not benefit from the increased speed of the “churn” like they did with PD4. PD4 allowed three riders holding 273W to churn and sit on the front rider’s wheel, but here with PD4.1, if we required all 3 riders to hold the same power, it had to be 308W minimum to hold the wheel.
- In a TTT situation with all riders taking equal pulls on the front, each rider would average 331W. (With PD4 the average was 305W, while PD3 was 335W).
- The hybrid format was a solid option with PD4, but it is now terribly inefficient compared to single file formation in PD4.1.
Conclusions
Let’s answer the three questions we stated at the top of the page:
Is road bike power savings in the draft with Pack Dynamics v4.1 different than the savings with Pack Dyamics v4?
Absolutely! Here’s a table showing approximate power savings with PD3, PD4, and PD4.1 based on your position in a TTT group of 4 riders:
| Position | PD3 | PD4 | PD4.1 |
| 2 | 17% | 28% | 24% |
| 3 | 25% | 33% | 33% |
| 4 | 30% | 37% | 34% |
As you can see, the difference in savings between PD4 and PD4.1 is smaller than the difference between PD3 and PD4. Clearly, Zwift is dialing in their Pack Dynamics.
Does Pack Dynamics v4.1 alter pack speeds, and if so, how?
Our first test seeks to emulate the front of a hard-charging race, and that test returned a significantly slower result (1.5kph) with PD4.1. So yes, we can confidently say that PD4.1 slows overall pack speeds because it reduces the churn/slingshot effect that drove pack speeds up previously.
Are there any other observable differences between PD4 and PD4.1?
Yes. The four tests above showed marked differences from PD4 in two areas not already mentioned: the “shape” of the draft, and the potential for wasted watts.
Draft “Shape” Notes
Zwift altered the “shape” of the draft cone (the draft “shadow” behind each rider) with PD4.1. These changes could be clearly seen during our tests, in two ways:
- The cone’s “fall off” was faster. That is, riders lose the draft benefit sooner as they drift back and off the wheel of another rider. This may explain why the draft benefit for the 4th rider in our tests above was only slightly better than the 3rd rider’s.
- The cone is narrower. That means in a small group or single-file situation, left-right positioning is more crucial than ever. Steering may be beneficial here, but in a TTT without steering, teammates behind the front rider may need to make room so their friends can “slot in” and not waste watts sitting on the edge of the draft cone off to one side.
Wasted Watts Notes
What you don’t see above is something interesting we observed while wrangling our bots into single-file pacelines: PD4.1’s modified overtaking algorithm has, in some sense, created a new “sticky draft”.
In a single-file scenario like test 2 above, it’s now possible to have one rider (say, the 4th in the line) holding 30 more watts than the 3rd rider, without overtaking the 3rd rider. This wouldn’t have happened with PD4, but it definitely happens with PD4.1. It’s not the same as the old sticky draft, but the result is similar, if not more pronounced.
We assume this is due to whatever algorithms Zwift uses to reduce forward/backward movement in the pack. Especially in a single-file scenario, it now takes much more intentional ramping up of power to overtake the rider ahead, even if that rider isn’t the front rider in your group.
What does this mean for racers? It means PD4.1 makes it much easier to waste power in the pack. Similar to how the old sticky draft might have stuck you on someone’s wheel whether you’re doing 250W or 270W, you may now find yourself able to maintain a pack position whether you’re doing 300W or 340W!
So when you’re sitting in the pack, try easing off your power to see how it affects your position. Learning how to ride most efficiently with PD4.1 may take some trial-and-error practice, but you’ll probably find you can save watts by riding smarter.
Dynamic CdA Notes
While not a part of the tests above, PD4.1’s dynamic CdA changes are worth mentioning. Zwift tells us “the current setting in regular races give only a 3% benefit if your power goes above 20% or more of your last 10 sec. power.”
The goal here is to give a slight “boost” to attackers. Notably, though, dynamic CdA changes are not enabled for individual or team time trials, according to a knowledgeable source within Zwift.
What It Means for Road Racing
For road racers on Zwift (vs TT), we believe Zwift’s new pack dynamics will result in races that more closely reflect IRL racing in 3 ways:
- Pack speeds will be a bit lower;
- Breakaways will stand a better chance of staying away;
- Taking a pull on the front (and even simply moving forward/backward in the pack) will be a more purposeful action as opposed to an accident caused by the churn/slingshot effect.
Your Comments
Got comments or questions about Pack Dynamics v4.1? Share below!
Many thanks Eric for your hard work to show us these finetuned results. And a “well done” to ZHQ to deny the crazy churning at the front. But two things are surprising. 1st ZHQ said, that it will be only harder to pass a rider at the front who has no draft, but with this new style of “sticky draft”, it also seems to be harder to pass a rider who is in a good draft position. 2nd ZHQ said, that the draft effect will be higher than before. These tests show a different picture. Perhaps it’s only higher in… Read more »
Yeah, I think the larger draft effect is something you’ll see in larger groups. PD4.1 will definitely be interesting for TTT riders!
Would be very interested to note the draft effect down to pos 8 in an 8 rider TTT – we rode this morning and several riders noted it felt easier / needed less watts to ride in positions 3 and 4 than it did in positions 7 and 8 – which just seems odd, and maybe an effect of the modified draft cone.
Agree, but like Carsten opinion, advantage new PD in TTT only “with” steering. Thanks Eric.
They’re doing some sort of check to see if your nose is in the wind, aren’t they. It’s becoming more dimensional (which makes sense given the steering).
It’ll be cool and stunning to start seeing simultaneous left side and right side ‘lead-out trains’ with a real gap in the middle. Not sure if we’re there yet but I’d love to see it tested.
On the Dynamic CDA comments. I had read somewhere that the changes in CDA would not occur on a TT bike in a TT or TTT. So you might want to check into that.
This is correct – I amended the post above. Thanks Casey!
So far, I prefer the previous pack dynamics – especially when not using steering. There is so much extra and wasted effort required in the bunch now to maintain position or to move up in the pack or to overtake. I believe this needs to be tweaked further. Or, maybe consider one set of dynamics just for races and another more ‘relaxed’ set for group rides. I saw a lot more variability in my watt output just trying to do a simple robopacer ride at a steady zone 2 pace.
You are correct- there is a ton of wasted efforts with this new pd4.1-
it def needs to be tweaked further-
it appears they/Zwift just needs to make separate dynamics for races/ tts/ and group rides-
For me it’s the opposite – much prefer the new dynamics: it’s more realistic, much less churn & stickiness, and means if you have a gap the only way to get caught is for the chasers to put out more W rather than there just simply being more of them
4.1 is a big win for me
In a racing situation I can understand this somewhat. But in a casual group dynamic it is a bit ridiculous the amount of effort you need to put in now to get back with the pack. If you are not careful, it is extremely easy now to drift back in the pack and getting back to your place in the group is much more difficult now – especially in large groups – like robopacers. I find it easier now for group rides to ignore or stay away from the larger groups and either ride solo or in smaller packs where… Read more »
Respect of the pack for the rest of the riders in the pack is a whole different can of worms. Riding solo feeling easier, for some packs, sounds true to life too tbh.
Agree completely. The first thing is that RoboPacers are around 1.1 mph slower with all the other problems enumerated above. The problems with narrowing the draft cone is not eliminated with steering. I have it and it can be very difficult to stay directly behind the lead rider. I frequently go back and forth several times trying to move as little as possible and end up over correcting. Why are they screwing with group and RoboPacer riders when trying to satisfy racers?
Interesting. I’m also 183cm but my weight is 95Kg and its pretty hard work out front and I can see why. Would love it if Zwift just added some numbers or a graph on screen to represent the draft numbers. As a newbie I find it hard to stay in touch with the rider in front of me to maintain a stable position. If say 35% from above is the best draft percentage you can get, then the graph would be at 100%. Cycling on your own the graph would drop to zero. Simple fast response bar graph would help… Read more »
sauce for zwift
No thanks, already looked at it, too much additional stuff I don’t want and don’t want to run another 3rd party App either. Screen already has too much unwanted stuff on it, much of what is already there wants to be “Optional”. Once you also run the companion App, all you want on the big screen is bikes and scenery and the important stuff .I want power, heartrate, gradient and draft.
you can customize sauce to no end, and literally have only the draft figure on screen if you want. really easy and minimal cost, I can’t imagine not having it now I’ve used it for a few months. but you do you!
Some of the Sauce for Zwift features should be incorporated into Zwift as part of the standard offering – just made optional for those who don’t want or need them. In particular, the visual indicator for draft level would really be helpful now with these new pack dynamics. Also, Sauce for Zwift doesn’t work across all platforms – namely Apple TV.
Sauce for Zwift should make the additional screens ‘optional’. I for one wish there was a way to visually see how much draft your are getting in real time – especially now with these new pack dynamics. Without it, the system is too laggy for effective responses especially in racing situations.
Everything Sauce does should be built into Zwift.
I don’t think so. I have been using Zwift for about a year and its only lacking a couple of things. The auto camera change is a no brainer and I have to run a free 3rd party script and remember to start it every ride. All it does is move through the 9 camera angles automatically every 15 seconds. The display of the “Draft” that Zwift is already using in the draft dynamics is another no brainer. All these things should be in the settings to be able to turn them on and off. I would get rid of… Read more »
Thank you Eric! Do you have a sense from this testing whether the new pack dynamics will make it harder for a petite rider (I am 162cm and 59kg) to stay in the draft? I already find that hard at higher paces if the yellow beacon is significantly taller and heavier than me, and dread finding it’s now worse!
It’s def gonna be worse for you – 🙁
I haven’t tested it extensively yet but it doesn’t sound definite at all. Pack going slower and slingshotting less should make some riders more comfortable staying in. If you can hold SOME level of power you should be able to stay in if that level of power is enough for the pack you’re in. We might see a lot more smaller packs, although it’s hard to say from Eric’s tests what will happen when the road is full and beyond the first couple of riders if it plateaus or drops off steadily or abruptly. Sounds to me like being in… Read more »
The height of the rider has nothing to do with it… in fact, taller riders have to work harder than shorter riders, all other things being equal.
What you’re probably seeing, Holly, is that your W/kg is higher than the larger riders/leaders. That’s perfectly normal, and happens outdoors too. I once road 30 flat miles outside with a strong yet tiny rider on my wheel… I averaged about 3.5 W/kg, while she averaged over 4.5!
I was wondering last night if this had been rolled out (hadn’t seen a post on it) because the Coco group seemed significantly slower than “normal”. Even with a rather large group the speed on the flats was often as low as 37kph. It seemed worst in the dirt and broken pavement sections around the volcano, so I was wondering if they updated the rolling resistance on those surfaces (that have previously been the same as pavement if I’m not mistaken)
Yes, I noticed this drop in pace for the Coco Volcano circuit. 3 full laps in the pack was 36.9 kmh this week, but was 39.7 kmh for the same circuit and distance in April. The size of the group can make a bit of difference, but not that much! My 50 km zone 2 session has gained a few minutes (it was also 177 Watts average this week, cf 185 in April for the same distance)
Im curious how long the reduced cda when attacking, is it only for 10 seconds or the entire time you are holding that higher power?
My question too. I assume it might be just naturally resolved when the >=20% higher power numbers gradually slip more and more into your average of 10sec-window; thus, increasing the avg and thus you would further need to step up your current power in order to stay >=20% on top of the increased avg power. But that’s just a speculation and I’m keen to see the facts.
Also, I have no idea how much -3% of CdA is worth?!
My guess is it adjusts very quickly (say, every second) so as soon as your power drops below “20% or more of your last 10 sec. power.” your CdA bonus goes away.
thanks for the post Eric. I can definitely tell a difference in racing and in joining the robo pacers. much slower speeds. In my race my average power was way less than almost every rider, even those behind me. I came in 9th out of 25, averaging 2.8 w/k. Just about every other rider was 2.9 or higher. I thought, how is this possible, until I read about the wasted energy you mention within the pack. Looks like I may have been the most efficient!
I had a crack at a race last night and thought I would try test out the differences. Here is what I found. Warm up with pace partner.. I found it was harder to stay at the back of the bunch as I had to keep watts/wpkg closer to the pace partner than I had previously. In the race (France/Petit KOM circuit – All Grades Combined), I managed to break away off the front of the bunch early on and held a 10sec gap by just watching the wpkg of the leaders of the bunch. It was much much easier… Read more »
Great writeup, Stu!
I don’t know what else they have changed but for a 95kg rider it seems to be way better. I no longer need to pump 320W to stay out front on a 2.6Wkg C Coco ride and for the first time ever I managed to do 15Km over 30M out front of the pack instead of a few hundred meters. 256W average for the ride. Also seemed much easier to ride the wheel of a person in front of me.
It’s better, but still disadvantages larger riders compared to real life. As far as I know, Zwift still assigns higher rolling resistance to larger riders.
If Zwift displayed Watts rather than W/kg, the disadvantages would be more obvious. Displaying W/kg, incorrectly implies that riders with equal W/kg should go the same speed.
I think the heavier rider while in the pack and riding on the flat really shouldn’t be penalised. The difference in aero must be minimal. Going up a hill, sure and that’s really noticeable and you get pinged, fair enough because the heavier rider has a clear advantage on the downhill but I’m not sure it makes up for the uphill.
Many thanks Eric, guess we need to get some race practice in ahead of ZRL otherwise we’ll be dropped! Oh wait, I get dropped anyway so no change for me, I’ll just get dropped quicker I guess!
Hi Eric. I did my own group ride and race test of PD4.1 The group ride was the Asia 130 with approx 200 riders and I found I averaged approx 30 watts more for the duration of the event. That’s a massive increase over the duration of a 130 km ride (some 3 hours). All this sitting in the pack and not making any additional effort. This event is a ride, NOT a race and my goal was endurance training. So for those unclear on what ‘endurance’ means, in the most simplistic explanation is, riding at a Hear Rate in… Read more »
For me 30W more is basically nothing at 120-150 even for 3hrs, feelable at 200-230, and starts to be a real ask at 300+. If the biggest complaint about PD4.1 is that staying with the pack you want to or holding steady power consistent with that pack is too difficult (or puts you outside the zone you intend your session to be in to reach your goal), then I think that’s not really a substantive complaint about PD4.1. Personally I’m ok with it requiring adjustment in selection of group and ride levels and I’m ok with artificially fast Strava segment… Read more »
Hey Dan,
I’m trying to square your experience with mine, because when I rode (for instance) with Coco for 30 minutes today, I averaged 193W… same as ever with her. Not sure how you needed an extra 30W, unless you were pushing it at the front, dangling off the back, or wasting watts somehow.
I’ve found races to more animated with PD4.1. Breakaways definitely sticking more, and as more people figure out that that’s possible, I bet you’ll see more and more attacks happening.
Eric, I did a ride with Miguel today on Volcano Circuit. The ride was 1kmph slower than the exact same distance in June.
Absolutely spot on …
I’ve done I think four 1-1.5 hour-long races now with PD4.1. I really dug it the first time as I’m an aggressive racer and was able to successfully execute bridges to groups up the road a few times that I don’t think would’ve been successful under PD4. But I’ve since changed my mind and at this point pretty much really, REALLY don’t like PD4.1. I mostly race the Chasing Tour races and those events often break apart pretty early, so I find myself in smaller groups of 6-10 quite often. Here’s what I keep experiencing: I (and other riders) are… Read more »
Reading this about the rider in position 4 doing surprisingly worse then before this new Pack, this could explain my feeling in a TTT race with 3 other riders. I was the only one without a steer, and i felt like i had to work harder then doing leads. I’m also convinced that the 3 steering mates are riding a different line and going in 4th position, it was very hard to grab the wheel of P3 because i couldn’t steer to P3. Now i have to find out: 1. Was the new PD4.1 responsible? 2. Was (not having) the… Read more »
HATE, HATE, HATE!!!! THE NEW STICKY DRAFT – PD4.1!!! HATE IT!!! TO THE POINT OF LOOKING AT OTHER APPS…DESTROYS PR/STRAVA IMPROVEMENT! HATA OT HATE IT HATE IT!!!