Zwift’s May update included two new start pens and corresponding routes which weren’t mentioned in the release notes. They’re located at the top of Zwift’s two biggest climbs – one on Alpe du Zwift, the other on Ventop:




Start pens are used for events, of course. And right on schedule, two new event-only routes were also added to the game along with the start pens:
Each of these routes begins in their respective mountaintop pens, then has a short lead-in to the KOM banner before making their big descents. The routes don’t end where the KOM start line is in the other direction, either. Alpe du Zwift Downhill ends around 200m past the start line, while Ventop Downhill ends ~1.7km after the start line, which makes sense since the Ventop KOM start is located well into the climb.
Clearly, Zwift wants to hold some downhill events. But why? I have one theory. But first, let’s look at why downhills are currently a bit goofy in Zwift.
Don’t Try This Outside

Zwift generally gets good marks for realistic avatar animations and game art, especially when compared to other cycling games. But one place Zwift has always been very unrealistic is in its cornering speeds.
Apart from the u-turn on London’s Classique, there’s no other place in Zwift we’re aware of where any sort of automatic corner braking is applied. And that’s what needs to happen if Zwift wants realistic cornering speeds, since there is no universal braking interface that would allow us to use our physical bike brakes on Zwift.
For whatever reason, Zwift hasn’t prioritized corner braking in its development goals. Clearly it’s doable (see London). And it’s been requested by the community for years – when Crit City was launched I wrote how adding auto-braking to the course would make it even more dynamic, and make breakaways more viable if done well (read more here).
Zwift is also known for being a bit fast in terms of overall average ride speeds, which I wrote about this at length in “Are Zwift’s Tarmac Speeds Realistic?“. While auto-braking wouldn’t fix the speed issue entirely, it would slow average speeds a bit. And that would be a good thing.
But as far as I know, Zwift hasn’t done any work in auto-braking. Yet.
Full Speed Ahead… On the Brakes?
And that brings me to my theory explaining Zwift’s addition of downhill event start pens and routes.
What if Zwift is planning to add/test auto-braking on the game’s biggest descents? The only way to really test the realism of this auto-braking would be repeated descents. And the easy way to do that is to start at the top. Bingo!
Years ago, rebellious racers held some underground events which were ironically located atop Alpe du Zwift. The goal was to get to the bottom first, which meant riders would reverse weight dope, ride TT bikes, and anything else they could do to go fast.
The races were fun, but of course lacked realism.
Imagine, though, a downhill race where your bike slowed in each corner. You would have to sprint out of the corner back up to top speed. Wash, rinse, repeat 21 times on the Alpe. That would be a fun and fresh challenge!
Smart Brakes
Going deeper down this rabbit hole, if Zwift is testing auto-braking with the intent of implementing it all over the game, I have one request: make it work like an IRL race. Make it smart.
IRL (and this is especially noticeable in crit races with tight corners), a single rider can take a turn faster than a group of riders, because they can pick the ideal line for maximum speed.
Replicating this in game would require several things to happen:
- Auto-braking would have to be increased for larger groups. Or put another way, single riders/small groups would have to be allowed to take turns faster than larger groups.
- Single riders (or riders who are in a single-file line) would need to take a different line visually in tight corners compared to larger groups.
It may not be possible to easily replicate outdoor turn dynamics such as #2 on my list. But one idea I had years ago on Crit City was to base the auto braking on how much draft you’re currently receiving. That is, let someone who is on the front not receiving any draft take the turn faster than those who are sitting in and drafting. This is basically what happens outdoors, and the math would be easy on Zwift’s side.
In fact, if the riders receiving the biggest draft benefit were slowed the most in corners, this might nicely replicate the IRL “concertina effect” where the pack compresses into corners then stretches out as riders sprint back up to speed.
If auto-braking was done this way, you might have a turn where (for example) a single breakaway rider can go through at 45 kph, while riders sitting in the peloton are slowed to 40 kph. That’s fun stuff, because now there’s a real incentive to attack and get into a breakaway!
Bump it up another level to super smart auto-braking by letting drafting riders go through turns at a higher speed if there are no riders to their right or left. Again, this is what you see IRL – when you aren’t boxed in, you get to take the fastest line.
Concluding Thoughts
With Wahoo recently acquiring RGT Cycling, Zwift is surely looking at RGT’s feature set and prioritizing the implementation of any strong features RGT has that Zwift lacks.
Auto-braking is one such feature.
Assuming Zwift is in fact testing auto-braking on its big descents, here’s my order of auto-braking dreams:
- Fun to test, slightly useful: Auto-braking in the corners on Alpe du Zwift and Ventop
- Baseline realism goal: Auto-braking on all Zwift roads
- Race-friendly realism: Draft-influenced smart auto-braking to encourage attacks/breakaways
- Game-changing, best-in-class feature: Super smart auto-braking and rider routing to reward small, well-organized breakaway groups and team time trials
Your Thoughts
Do you want to see auto-braking in Zwift? Got any great ideas for how it should behave? Share below!
Autobraking would be a game changer. Been calling for it for years. I think the maths is easy enough to do, determine a max speed around each bend depending on the ‘lane’ you are in. A lone or front rider would automatically get the best lane, and this of course works with steering too.
I really hope that this is what these pens are for, can completely shake up racing on the platform.
I would love the auto-braking. Hope it happens!
I have a question about the top of ADZ. The big statue on the top used to pop up on the screen, but doesn’t any more. Why is that?
That sounds awesome if implemented the way you describe. The next step after that would be to take advantage of steering and let you steer into a sharper line on your own, but again, autobraking would be stronger the steeper the line you took. Work that in with drafting effects on autobraking as above, and there might be a reason to use steering in the game. On a downhill, maybe you’d want to take a wider angle than most others so you wouldn’t need to scrub off as much speed as those chasing you if you want to get away.… Read more »
Agreed Paul and that auto braking AI would have to react pretty fast because if you are picking your line thru the corner and you go from being in the draft to our of the draft as you hit the corner then do you also get the quick “speed boost” like Eric mentioned in his simple math method above? I mean, if that was the case, talk about a paradigm change for immersion…now you are paying attention to your effort (watts, HR, inclines) and your position in draft becomes even more important and while you are concentrating on position, you… Read more »
I add a button that raises and lowers the supposed brake. And the curb turn with the sterzo
Zwift did not have to create these start pens to test auto-breaking. I can hear the sub-2 w/kg group ride leaders “Okay group we’re starting easy out of the gate, blob up at 80 km/hr” 😂
I thought it was some sort of gift to you, to help you test gear downhill. At what point can we expect frames and wheels ranking charts for downhills ? 🙃
Ha!
Downhill performance, in my experience, is basically determined by aero performance. So the fastest bikes on flats are fastest on descents.
Specifics for how much aero affects downhill would be useful when calculating what bike to use for individual efforts on courses with up and downhills (eg surrey hills). We can accurately compare on climbs and flats, but without a number to put to gain/loss on downhill we are flying blind. Depending on the course, aero bike, climber, all rounders, aero TT bike, climber TT bike are all potentially in the mix and everything in between until we can determine which one is fastest overall.
If you have the least braking if you’re out of the pack and by yourself, I wonder if someone with a big sprint (and the ability to hit high numbers over and over again) just says, “Screw it, if I’m drafting, I’m losing. I’m going with my heaviest and most aero bike.” and chooses the Canyon Speedmax with the DT Swiss Disks. If s/he hammers the first sprint right from the gun to get clear and hits big numbers out of every turn, does the pack see them again? My main concern about this type of race is that folks… Read more »
Maybe they were developing corner physics, braking and steering for the new Zwift bike and were going to promote it with these new routes
I need to start eating more
I can’t wait for the “Look Mom No Brakes” power up.
Absolutely!
If Zwift does some sort breaking and other physics adjustments closer to irl then RGT and other apps loses one of it’s few big adavantages on zwift as zwift as most everything else: plenty of roads/paths, communities of riders and racers, surface/air resisantance factors, etc. For RGT to compete it woud need to add alot more riding options first off, same as the other apps. What Zwift has to do is fixes, minor additions, and refining some sections of the program better. While others apps have alot more work to do to come close to Zwift, then fixes to bugs… Read more »
Auto-braking would be awesome. Having to sprint out of corners on descents where you can supertuck the whole time would make descents more dynamic. Without braking, racing on Fan Flats can be frustrating. You can lose many metres off the pack if you “randomly happen” to be forced to the outside of the hairpin before the sprint. If the inside line riders had to slow down more, that would even out the random road placement lottery for that spring segment. Last time ZRL raced there, I wound up focussing more on how to make sure my avatar stayed on the… Read more »
I have to sprint after every hairpin ??? … my toilet-break-option is lost 🙈🤣
It WILL be interesting to see if you can still coast down the Alpe, even with auto-braking. I would assume so!
Does side-to-side positioning really matter in a race without steering? I thought we all went through the corner travelling same distance and our displays just showed us in different spots.
Hey Paul, I notice in big group races, when I’m on the outside of a turn I definitely lose position in the bunch, and vice versa. It doesn’t put me out of the bunch but I can go from the front to the back pretty quick. Also, race distance isn’t always exactly the same for all participants.
@Cameron Allan Yeah, I notice that difference as well (it always seems like I’m on the outside of every turn), but, when I started Zwifting (back with the original pack dynamics and before steering was an option), I was told that everyone saw that, but we all followed the same line through the center of the turn regardless of what the screen showed our avatars doing. That may have changed through several different versions of pack dynamics (and the introduction of steering), but it was always a common question from new races back then, “why am I always on the… Read more »
Relieved to have my 100kph badge already. Not sure if this will reduce chance in future!
Bologna was the easiest place (for me) to get that badge and I’m not sure there are enough turns on that to slow you down much.
This could be nuts on some point races where the sprints happen just after a big turn.
big fan of making it more realistic.
This would be a killer for anyone doing an Everesting on ADZ. Right now, there is a great recovery between efforts.
Similarly for the PRL routes: I know I’ve seen people saying they actually changed bibs on the way down the 8th hill.
But I’d take the exchange.
My guess is you’ll still be able to coast down the Alpe with autobraking – it’ll just take longer!
Does Zwift really know where my bike is In relation to other bikes and if so does it match where I see it is? If not, how do I accurately pick lanes for cornering?
It… sort of knows. Rider locations are a lot more consistent across different clients than they used to be, thanks to pack dynamics 2.0 and how they’re calculating positioning, etc.
But it’s not perfect.
The only way to really “choose a lane” if you don’t have steering enabled (and very few do!) is to not be in a group (no riders to either side of you). Then the game routes you on a better line for sharp turns.
Not my area of intense interest BUT re: autobraking (and you covered this and a LOT of nurances) … I wouldn’t want a change where JUST having more horsepower is win win where in real life skill matters. The “everyone goes fast through corners” thing doesn’t bother me I can imagine crit sprinting monsters losing out there on zwift vs. real life. My MO is carve the turn fast with least slowing requiring LESS sprinting out of irl crit turn say and being in the right place (further up but not far enough to be semi forced to put my… Read more »
Another area I would LOVE auto-braking is when I’m locked onto a rider in the draft. I want to be able to follow someone’s wheel and for Zwift to auto-brake (or ignore my power) to keep me behind that rider. I think there should be a button or control to indicate I no longer want to lock onto that rider.
Couldn’t they just set up a dev environment where they could place a Zwifter wherever they want? Seems strange that they’d need to set up a pen to start at a specific spot. Plus, free ride routes don’t start in pens, just at specific points on the road; couldn’t they just do that? Maybe they’ve got some fun new game for descents, like, steer and collect all the rings, like those Mario 64 slides. Right now, those big descents are pretty boring… you just leave it running to collect the XP points. (Would appreciate an auto-descend feature, with the option… Read more »
It does seem a bit strange to think that the though process inside zwift went something like “we need to implement more realistic cornering physics” -> “the only way to test this is to do downhill races, please create some pens atop of the largest hills in game”.
The pens are too well-designed, in my opinion, to be just a “we want to test something real quick” sort of solution. I think they’re planning on doing actual downhill events once autobraking is perfected.
Surely. It’s smart. It will be something heavier riders can have an advantage in, so it keeps that customer base happy for little effort.
If they are too well designed for “just want to test something” then isn’t that a point against your “we want downhill for testing auto-braking” theory?
Auto-braking is another thing that will make existing PRs on routes impossible to reach in the future (the first one being making gravel/dirt slow you down).
Yeah, this is a very unfortunate side-effect of any change to the physics. We would lose years worth of data for comparing all of our older segment efforts. At least in London the braking on the hairpin has been there since day one, so all the data is consistent, but slowing down the gravel wreaked lots of segments in Watopia, and braking would have similar effects.
Yup. But I’m ok with that, because what’s the alternative – Zwift physics/routes never changing?
Although I DO wish Strava had a “reset all of your Zwift PR times” sort of option!
That’s the one thing I appreciate about Zwift only keeping your 30-day PRs.
I would actually prefer Zwift keeping the physics the same forever, there hasn’t been a change that I prefer yet. Clearly you have a different opinion.
Reseting your own PRs doesn’t solve the issue of placement on the leaderboard, before you could get a decent percent ranking there, but after a slowdown no longer. Unless everyone else resets their PRs.
Theoretically, they could develop an algorithm that would take previous power output and courses, run those numbers against the new physics/braking algorithm and correct those times. Highly unlikely they’d ever do it though.
Strava doesn’t have all the info necessary to do that (and it’s probably not saved anywhere on zwift’s side either): they also need to know the amount of drafting at every instant.
If they make it a route, this will be the easiest route badge ever!
Truth.
As someone who is not yet at level 50, I’m not going to complain about it, if so.
How about some Track riding like,Wahoo or RGT or that new app from Chris Froome,where you can ride on indoor cycling tracks
i guess steering could be used too to to scratch precious seconds
It would be really cool if you could steer, and if you picked the right line, auto-braking slowed you the minimum amount. Sort of like what happens on Repack Ridge currently.
Imagine if they implemented this (soon-ish) AND results-based rankings for categorization, and, when all the folks who leave for the summer come back and ask, “what’s changed, did we get a couple new km of roads?”, you get to say that racing dynamics have significantly changed for the better.
I don’t know what they have planned, but I’m excited to try it out.
Given this Wahoo Fitness’ Debt Ratings Downgraded | SGB Media Online (sgbonline.com)
“Wahoo has a good record of successful new product launches and it expects significant sales and earnings contribution from several new products and refreshes in late 2022.”
I would guess high odds a new kickr bike is coming and units are out there being tested under NDA right now. Something that really gives RGT a new racing featureset on the horizon and zwift trying to respond?
Zwift is not that fast to react to anything.
Finally ! Make those heavier riders in each cat pay accelerating out of every corner, they won enough already. To win, now they’ll have to work for it !
Can the new downhill routes be selected for events? I didn’t see any events on the alpe downhill yet.
No, they’re not selectable by us common folk for use yet!
Felt a bit travel sick descending the alp with all these hairpins and the scenery zooming past and rotating. Unfortunately auto braking will make my abysmal ave speed even worse 🚵♀️🚵♂️
Sounds like a great idea in theory. However in practice a major change like this would not be made quickly. It would also make coding new roads more difficult so that they took even longer to be released AND expect bugs galore if this ever happened.
Dropper posts allowing you better aero descents!
Pack [braking] dynamics through a corner vary….the front riders ride like the breakaways….[simplifying] the ones in the middle of the pack ease off slightly [for various reasons from line-of-sight and safety] and the rear of the pack depending on the severity of the turn may actually hit the brakes….hence the accordion effect. so in short the front can maintain the speed of a solo rider but those in the back [wheel suckas will be whip lashed]…..Always safe and easier to be very near the front.
Do you think it’s possible you could get an entire pen to agree to cross the start line and not pedal the entire way down and see who wins?
Resetting my weekly goal to 1,000 miles a week now that I can do most of them going downhill!
Did the finishing gate on Alpe du Zwift get moved back recently?
The overall route seems to be the same length, but it looks like there is more road to cover on the flat between the top of the climb and the gate since the last update. Maybe I was just more tired!
Auto braking will be awful RGT has it and it sucks. With no control for the individual rider no skill like real life. It will be just frustrating. Like believe me you don’t want them to add wind ether. Zwift is fun not real.
We could also add stop signs and red lights and cars buzzing you and drivers yelling and potholes to add to the realism. And then if you’re really unlucky you get hit by a (virtual) car and you’re zwift account is disabled for a 6 months.
COOOOOL!!! Finally we all can catch up w/ Tim Searle’s Distance… Log in… Chose Road to Sky reverse… watch your Avatar doing 12.44kms… repeat… repeat… repeat…
how will it actually be applied?
Rider skill aside…irl corners generally have maximum speed they can be pased through.
If below that max speed i would hope no braking is applied.
This may help the lightweights get left behind in zwift….
Irl my heavy friends have to brake much harder than me….
I hope this wont be another tax on light riders….
Much as I think “accurate” auto-braking would be a really good improvement my main worry is that Zwift still isn’t accurate enough on rider positions so the braking would end up being incorrectly applied – even now after “Drafting 3.0” we still see riders at the front of a bunch in ‘drafting’ sit-up position despite having no-one ahead of them
If they are going down this route, maybe they should make the downhills full scale, meaning stop derating the descent by 50% as they are currently doing. Or what about an event mode only setting where they could make the down-hills full sized.
Downhills are full scale in terms of their speed effects in game. They’re just halved in terms of what’s sent to your trainer, so you don’t spin out as easily.
Not sure if there’s much advantage to sending the full descent gradient to your trainer. But maybe I’m missing something?
Oh, thanks for the clarification. I was translating the trainer effect into a speed effect, but obviously I didn’t think about it because the downhill speeds can be very fast. Regardless looking forward to what changes may be coming.
In a race, sure go for it. When I’m free-riding, nah it’d just be annoying.
Two months later and nothing else on this that I’ve heard?
Nothing I’ve heard, either.
I’d suggest Zwift add to every user a small bright red or maybe orange light titled “Auto-Braking ON”, i.e very near the Speed display. John Conant<