Are the draft figures from Sauce for Zwift accurate? How is it possible to be putting out 260 watts and receiving a draft benefit of 220 watts? Just before this screenshot, it was showing 380 watts! On average for this ride, I received a benefit of 123 watts. Those are a lot of watts that do not correlate with speed! Btw, I have been checking those figures out of curiosity for a long while and they don’t make sense to me…
The way it is supposed to work is that Zwift calculates a % draft savings number, which applies only to power going into wind resistance – not acceleration or rolling resistance or climbing. For example, suppose you’re going 400 watts on flat ground @ 50 kph and weigh 80 kg with the bike, with Crr = 0.4%. 43 watts goes into rolling resistance, leaving 357 watts of wind resistance. This would correspond to CdA = 0.22. But then you go into a group, and your CdA drops by 40% due to drafting, so you’re now (for an instant) at CdA = 0.13. Suddenly you’ve freed up 0.4 x 357 = 143 watts. These free watts go into acceleration, increasing speed, until you are going at a speed associated with 400 watts with a CdA = 0.13. But this puts you on the front of the group, and you lose the draft benefit. So your CdA increases back to 0.22. You now decelerate again.
I believe the Sauce number is based on the assumption you are holding your position in the pack, not moving up or back, so is only a rough idea of how effective the draft is. It’s a first step in a calculation done by Zwift to adjust your speed. So trust it if you’re holding your position, but if you’re changing position, then the calculation would also need to know your rate of acceleration, road gradient, and rolling resistance, which Sauce may not have.
2 Likes
I think it works well as a barometer to showcase the benefits of drafting, though the word “watts” might be better left out. After reading your replay and reflecting further on the figures, I realize they aren’t very precise. The laws of conservation of energy and aerodynamics ensure that the energy savings for drafting riders cannot exceed the total energy output of the group as a whole, minus losses from rolling resistance, drivetrain inefficiencies, and other factors. Drafting savings represent a redistribution of the lead riders’ energy, not the “creation” of new energy. (In fact, there have been times when the calculated savings seemed larger than my actual power output, and I have a very consistent riding style!).
I quickly put together a model for a pack of 10 riders, with the first rider putting out 230 watts w/ weight of 165lbs. The maximum energy savings for the last rider due to drafting is around 83 watts! Either Zwift or Sauce is overestimating the figures.
Depends massively on the speed, no? The draft savings at 60kmh are (and should be) way more than double those at 30kmh.
what do you actually mean? Yes of course, it depends on speed.
At its core, speed is a function of power, influenced by a variety of other factors as shown below;
By the way, with the help of AI, I put together a decent model based on a set of assumptions drawn from my real-life and indoor cycling experiences -not needed to post here. Sauce for Zwift could have used either speed gain or watts saved—ideally, it should show both!
My point is that the ‘watts saved’ shown by Sauce for Zwift don’t always correlate with speed—they often feel overstated. This becomes especially noticeable when riding on a flat route and transitioning in and out of the draft. The cause isn’t clear to me, but it’s likely due to Zwift overestimating speed, combined with Sauce for Zwift’s rough calculations or only accounting for some of the variables involved. Anyway, I was hoping someone smarter than me might have a more mathematical explanation.
I actually mean that you said that the maximum energy saving was around 83 watts, but didn’t specify at what speed - and if you don’t specify a speed then a simple man like me is free to assume that it’s at any speed, which is clearly not correct, as your model will produce very different results in climbing vs flat vs descending scenarios.
Yeah - that’s not me. 
The sauce number is not calculated, Zwift broadcast it (privately) and Sauce intercepts it.
If you are putting out 260w and the draft benefit is 220w, you would need to be doing 480w without any draft to maintain the same speed.
The easy way to test this is when descending, when the draft figure may be super high (like 600w+)
P.S. I strongly advise against using ChatGPT to provide formulae to prove physics, unless you already know the formula and can sense check it. It has been trained to stupidity. There is a very high chance there is a fundamental basic error in there.
6 Likes
Interestingly, the model achieves saving of 40% when riding behind a pack of 10, compared to riding solo, which aligns with the findings in most literature on the subject. As someone who uses AI professionally, I don’t find that to be the issue.
Descending isn’t quite relevant, as gravity comes into play. As far as I’m concerned (and what I am trying to figure out), the difference between riding alone vs. riding in a pack on the same terrain should be reflected in draft benefits: watt savings or speed gains.
If, as you mentioned, Sauce doesn’t calculate the figures and they are instead broadcast from Zwift, then those numbers are inflated due to Zwift’s pack dynamics and overestimated draft benefits, which actually makes sense. I guess Sauce is quantifying it. Philosophically speaking, you can’t save what you never had 
Btw: when I took the screenshot, I was actually 3rd in the pack -which makes it even worse LOL!
Yes, saving isn’t the correct terminology.
You are also correct that the number won’t bare much meaning versus reality, because Zwift’s speed calcs are inflated and the pack dynamics are pretty much made up (which isn’t necessarily a bad thing, but they have been hugely tweaked over the years to try and create an enjoyable experience).
Gravity has nothing to do with it - the reason you have such a big draft number when descending is because the speed is so high - the energy needed to overcome air resistance at 80km/h is huge compared to 30km/h as air resistance increases exponentially with speed… so when that air resistance is removed (drafting) you see a high number.
1 Like
Follow an empty pace bot for a bit on a flat course and you’ll see how it works pretty quickly. Assuming roughly equivalent weight and height… you should be able to follow a pacer if your draft number and watts add up to their power.
Junctions, downhills, bad steering, some corners, uphills, height/weight changes… all make it less useful. In general, though, people can ride much less power in the draft than they think.
I’ve noticed the watt savings are always unfeasibly high. Back of the envelope calculation shows it to be approximately double the true saving for me at pretty much any speed.
As a side note I get 0w saving on the Radio Tower descent. It seems to tail off at extreme negative gradients.
Perhaps a draft gauge measuring the level of the maximum attainable draft. based on the position of the rider relative to the pack, ranging from 0% to 100%, would be more meaningful
I’ve followed Constance (we are the same weight, I’m taller) at my watts + draft value = bot watts for 40+mins so it’s accurate under some conditions.
Staying in the optimal draft at the lowest possible watts is the key skill of the game. Very few folks are good at it.
1 Like
100% on the money! that’s why it would be nice for Sauce to gauge the optimal draft -not so much watt-savings.
Hello, I just learned about Sauce and watched a video about it. I have to admit, it gives Zwift that special touch. Are there people who use it and can suggest an “interface” for me? Or do you think it’s not worth it?
@Jos_Hellkuhl Personally, Sauce has transformed my Zwift experience, adding a whole new dimension to it. After years of frustration with Zwift’s outdated HUD, I finally decided to take the plunge, install Sauce, and customize it to suit my needs—and I couldn’t be happier. To be honest, I was on the verge of canceling my annual Zwift subscription and switching to another platform, but Sauce completely changed my perspective.
I’m currently running it on Windows, but it’s also compatible with Mac and Linux, so there’s flexibility no matter your setup.
Thank you for the response, I’ve downloaded it now as well
Now I just need to figure out what to display and how. What would you recommend? @XavierC
There is a Sauce discord server that is a great resource for everything Saucey. In the Sauce settings, at the bottom is a support link and if you click that you can invite yourself to the server.
@soj_hell The possibilities are endless! I’d recommend starting by adding a few overlays based on your needs, and start tinkering from there —if you check out my screenshot, you’ll see what I use.
Here are a few things to keep in mind: Zwift primarily has four HUDs (maybe more)—freeriding, Climb Portal, AdZ, and workouts, etc… When creating your overlay, be mindful of where these HUD elements appear so they don’t overlap. You can always create different profiles in Sauce to accommodate each scenario.
Personally, since I have enough screen real estate, I use the WinSplit app to allocate 87% of the screen to Zwift and 13% to Sauce. This way, I only need one Sauce profile, and Zwift’s different HUDs don’t interfere with it. I also wasn’t a fan of Sauce’s default colors, so I tweaked them to better match Zwift’s natural color scheme.
[quote=“XC, post:8, topic:644060, username:XavierC”]Philosophically speaking, you can’t save what you never had
[/quote]
Suppose the draft benefit is 75%. You are putting out 100W. Therefore you would need 400W without the draft benefit. This means you are saving 300W.
So there’s no problem with the draft benefit being larger than current power.
And 75% draft benefit in the middle of a pack is typical. This is the result of computational fluid dynamics calculations from Eindhoven University in 2018:
