Zwift Bans Two Riders for Manipulating Race Power Data

Zwift Bans Two Riders for Manipulating Race Power Data

Zwift has just announced sanctions against two riders, giving each a six month ban from Zwift Cycling Esports events after it was determined they fabricated or modified their racing data, “bringing the sport into disrepute.”

The notice of sanctions can be found here, which links to detailed PDFs explaining the Zwift Performance Verification Board’s (ZPVB) decision in both cases:

Violation Details

Section 2.5.1 of the Zwift Cycling Esports rules state that:

  • “The smart trainer or smart bike must be the primary power data source…” and
  • “Riders must record a second source of power data

So two sources of power data are required, the power numbers sent to Zwift have to come from the riders’ smart trainer/smart bike, and not their second power source (typically a pedal or crank-based power meter). But in both cases, the ZPVB determined that riders supplied intentionally-modified power data and/or log files.

According to the detailed document for Berger, her supplied secondary power fit file showed a variance of “approx. 0%” when compared to her primary power file. This isn’t what you typically see when you record power from two different sources – but it is what you’d see if you record power from the same source, but in two places. So ZPVB judged that Berger likely connected both Zwift and her bike computer to the same power source.

And that could have been an honest mistake. But the document details how, over the course of the next days, Berger supplied a Zwift log.txt file from Berger’s Zwift device which didn’t match the corresponding file saved on Zwift’s cloud. Namely, the device list on Berger’s supplied log file had been modified to indicate that she had paired a smart trainer as her power source, instead of her Stages power meter as indicated in Zwift’s cloud version.

The document details other reasons for ZPVB’s findings in Berger’s case, but to summarize they judged that she:

  • Paired her Stages power meter instead of her smart trainer as the power source in Zwift
  • Recorded only the power from this power meter as both her primary and secondary power file
  • Attempted to modify the log.txt file to show that she had paired her smart trainer as the power source in Zwift

According to the detailed sanction document for Duncombe, she originally provided a backup power file which did not include her race, but only the first 90 seconds of her warmup. Again – this could be a perfectly honest mistake. But she later uploaded a different secondary power file to ZwiftPower, and this file showed values approximately 1% lower than the values from her trainer sent to Zwift.

An acceptable margin, except that this new secondary power file appeared to have been manipulated to look like it came from her Garmin, when in fact it appears to have come from Zwift itself.

The document details other reasons for ZPVB’s findings in Berger’s case, but to summarize they judged that she:

  • Did not properly record a secondary source of power from her power meter for the race
  • Attempted to supply a modified Zwift power file in the place of this secondary power file

What Are They Banned From?

Both riders will still be able to use Zwift whenever they’d like, and even race in “unofficial” Zwift races. They are banned (for 6 months) from taking part in official Zwift Cycling Esports events, which are typically invitation-only events for top-tier teams/riders.

Berger’s suspension has been back-dated to the point when the edited file was submitted to Zwift, and therefore runs from 18 August 2020 to 17 February 2021.

Duncombe’s suspension has been back-dated to the point when her edited file was submitted to Zwift, and therefore runs from 20 September 2020 to 19 March 2021.

Conclusion

Both of these cases give us a look into the sort of analysis Zwift’s Performance Verification Board does for its top-tier races. It’s no small job! But based on the evidence presented in the documents, it appears the sanctions are warranted.

As news of the sanctions has filtered into the Zwift racing community, one common response has been (as one Zwifter put it), “That’s all good, but I wish Zwift would put this kind of effort into policing community races.” And that’s a fair response, given how rampant sandbagging is in the category D, C, and B races happening every day.

While we should never expect Zwift to put manpower into manually checking power files for everyday community races, they could get rid of many of the sandbagging problems by implementing just a few controls in-game, such as forcing riders to join a particular category based on their historic power numbers. Perhaps opening up the (still beta) anti-sandbagging controls so race organizers can use them in events would also be a good step. Hopefully we’ll see progress here soon!

About Zwift’s Performance Verification Board

You can find the details of Zwift’s Cycling Esports Rules here, which explains that the Zwift Performance Verification Board is “responsible for ensuring the integrity of results of Zwift Cycling Esports events and series.” As such, it has “the power to update the results of events, and apply further sanctions to riders.”

The board is made up of a chair appointed by the head of Zwift Cycling Esports, a legal representative from Zwift, and a technical representative from Zwift. Zwift doesn’t disclose everyone who occupies these positions, but the sanction detail documents are signed by the chair, Dr. George Gilbert.

Your Comments

Any thoughts or comments? Share below!

About The Author

Eric Schlange

Eric runs Zwift Insider in his spare time when he isn't on the bike or managing various business interests. He lives in Northern California with his beautiful wife, two kids and dog. Follow on Strava

Subscribe
Notify of
guest
54 Comments
oldest
newest most voted
Inline Feedbacks
View all comments
Gerrie Delport
Gerrie Delport (@gerriedelport)
6 days ago

It would be nice if they do something for lower categories as well (most revenue come from the lower categories). Like putting people in the correct pen at the start. So at least we can pretend to race.

Chris
Chris
5 days ago
Reply to  Gerrie Delport

Automatically putting people in the correct pen is a good idea, not only because it eliminates a lot of sandbagging. Many of the lower-power riders are also the newest members of the Zwift community, and therefore the most likely to misunderstand which pen they ought to choose for themselves. If you’re a D rider just getting into shape — or even a D rider who likes to ride slow and has been doing so IRL for years — choosing the wrong pen is disheartening, especially when the correct pen isn’t actually labeled D, as is the case in some races.

Rogerio melo
Rogerio melo (@neoprotheus)
6 days ago

all race winners should have double measurement validation in addition to being subject to a video weighing request.

Rob
Rob
6 days ago
Reply to  Rogerio melo

Although it would make results fairer I think that would put a cost barrier on people who want to race on Zwift. Not everyone can afford two power sources. No one is winning anything other than some glory in community based races (as far as I know).

For community races and series Zwift should be able to automatically place a racer in a category based on their historical power numbers (say previous 3 to 6 months. They have all the data.

Joel
Joel
5 days ago
Reply to  Rob

I agree.

Spending $600-$1200 for a trainer, then having to spend another $450-$1000 for another power sensor to race should be at the Pro or at least Semi-Pro levels. Let alone figuring out how to record from both sets of data. Does that require 2 computers as well? Sure I could figure it out, but ask most people to do that and they will say “forget-about-it”-NYC accent.

Benjamin
Benjamin (@benjamin_pitt)
5 days ago
Reply to  Rogerio melo

Are you talking about the premier races, or simply the regular races in all cats??? It would be silly to implement such restrictions in the “click and go” regular races on zwift. And the top level invite premier races on zwift already require such steps you mentioned already.

Dirk
Dirk
6 days ago

There’s plenty of racers averaging 400+ watts in races and you would think they would go after those ones…

Jim Haysom
Jim Haysom (@haysom)
6 days ago

Interesting to read that the smart trainer/bike needs to be primary power source, and pedals/cranks/etc need to be secondary source!

Erwin
Erwin (@erwinripmeester)
6 days ago
Reply to  Jim Haysom

Yes indeed. I have bought duo Faverio Pedals to have more honest power. They roughly read 10Watts lower then my Kickr Core, but I use the pedals for Zwift races as they should be more reliable. Strange that the rules state you must use smart trainer data.

Jim Haysom
Jim Haysom (@haysom)
6 days ago
Reply to  Erwin

I’ve Assiomo DUOs and they read just over my KICKR 4.0 for races this year. Been using them as my primary power source. Simply as that’s the power I’m putting out when on the road outdoors and smashing the hills.

Adrian Amos
Adrian Amos (@ahamos)
6 days ago
Reply to  Eric Schlange

As an “old” with a wheel-on trainer, I stopped using the built-in power measurements many moons ago and went with the Stages. I’ve submitted a few power analyses to ZP, but I’m really starting to feel the disparity of being on the wrong side of the technical divide. There are already enough barriers to succeeding with online racing–why add more? Why should I not be able to race support for my son on my old Kurt Kinetic while he gets the fancier Kickr Snap? I have power, and frankly my PM is more accurate. And it stays inside on the… Read more »

Jim Haysom
Jim Haysom (@haysom)
6 days ago
Reply to  Eric Schlange

I would imagine there’s a fair amount of Zwift riders who don’t own a head unit to be able to dual record, let alone have a second power source! BTW, I’m open to any sponsor enquiries so I can upgrade my Garmin Edge 520 🙂

Jim Haysom
Jim Haysom (@haysom)
6 days ago
Reply to  Eric Schlange

Yeah, that’s a fair point – hadn’t realised things had changed. Easier to benchmark against smart trainer/bike. You’d expect that serious racers are going to riding off a handful of top end direct drive trainers or bikes, but lots of variance if you were to measure against pedals, cranks, spiders, etc. I ride/race with Favero Assioma DUO pedals. They actually read about 1.5-3.5% higher than my KICKR 4.0 across 20 mins (both regularly calibrated) having just checked the 14 dual power races on my ZwiftPower account. This was all out efforts across Cat A and B races. Now riding on… Read more »

Shaun Harris
Shaun Harris (@vgrntbeauxner)
6 days ago
Reply to  Jim Haysom

I feel like there needs to be more conversation around this. If I record Zwift races with the Kickr and all other IRL riding / racing on the PM it will ruin the consistency I strived so hard to achieve by having one of the most accurate PMs on all of my bikes.

MarkE
MarkE
6 days ago
Reply to  Shaun Harris

Dual record and have zwift get power from the Kikr and record the PM data for training consistency.

Shaun Harris
Shaun Harris (@vgrntbeauxner)
6 days ago
Reply to  MarkE

For the many Zwift events I do (lets say 5-10 /week), it just seems like a ridiculous amount of work. Is that really what everyone else is doing? I’ve been a Trainerroad user for years and have always used dedicated pm regardless of if I’m inside or out. This seems to kinda just throw a wrench in the works.

Pierre
Pierre
5 days ago
Reply to  Eric Schlange

Actually what seems strange to me is that many user of Assomia Duo have similar power value with their direct drive HT. I just made a new dual power recording on Zwift Power and their is still a 4% differnce between the Assomia and the Tacx Neo2T (I use other PMs and the values are consistant). This 4% difference is quite logic between the power applied on the pedals and the resulting power in the rear wheel because of chain frictions. Many “well known” Zwifter obtain exactly the same value between the Tacx N2T and the Assomia…

Shaun Harris
Shaun Harris (@vgrntbeauxner)
5 days ago
Reply to  Eric Schlange

Cheers Eric, I guess for the time being I’ll just chill on the whole scene and maybe look to replace my kickr down the road. Right now I’m racing A in ZRL (not PL), and I’ll just plan on it not being a problem until it is. I imagine my dual recordings may raise some eyebrows, however.

Jim Haysom
Jim Haysom (@haysom)
6 days ago
Reply to  Shaun Harris

Yeah, just replied to another comment – similar, I ride with Assioma DUO power pedals on my bike, and tend to use these as primary source for rides/races as that’s the power I’m looking at on my Garmin when I’m outside. I do tend to sometimes go with my KICKR power for ERG workouts, as the power bars are much smoother – not sure why that is. Also, sometimes I’m lazy if it’s a social ride and connect to whatever pops up. Haha.

Shaun Harris
Shaun Harris (@vgrntbeauxner)
6 days ago
Reply to  Jim Haysom

ERG has power smoothing which is why kickr output looks like that for ERG workouts (unless you turn smoothing off).

Jim Haysom
Jim Haysom (@haysom)
6 days ago
Reply to  Shaun Harris

Weird, as I’ve done FTP test today, using my pedals for power/cadence and KICKR for controllable. Both a 5 min & 20 min max efforts on “free ride” this morning and just now. However, both the warmup and cool down blocks in ERG mode during workout were all jagged, power wasn’t smoothed on screen or in graphs, plus resistance didn’t seem smooth either. Don’t seem to get that when pairing KICKR for power.

Will B
Will B
5 days ago
Reply to  Shaun Harris

The reason I see for doing this is because many power meters can be manipulated to show way higher output then you actually are doing. You can change the slope on Quarq, Garmin Vectors, P2Max just to name a few. I can’t see any way possible to change the slope on my Kickr. Other smart trainers are probably same way.

Basti
Basti
6 days ago

IT would be so easy to let the Racer enter only the races which belong to their category……zwift wake Up. IT IS time now.

Scott DeLeeuw
Scott DeLeeuw (@sdeleeuw)
6 days ago
Reply to  Basti

I was hoping some of the professional sport improvements would make its way down to us peons, but it seems like they are operating almost as completely separate entities. Your suggestion seems so simple and easy, Zwift has our power curve data historically from the beginning of time, it would be so easy to place people in correct categories. Even if you went just by the last month of data it would be better than now.

SA Spyratoss
SA Spyratoss
6 days ago
Reply to  Scott DeLeeuw

Yep, or just cross reference your category in zwiftpower, which now Zwift took over and force entry into that category, period.

Tristan
Tristan
6 days ago
Reply to  Basti

Agreed! Also, what about accessing all our Strava data and combining it? I could see riders that have gone through this ‘certification’ getting a badge or something to show that they are compliant. In the future, machine learning will automate the above process and automatically put us un the correct category. I could also see ML being used as a proxy for the Zwift Sanctioning Board’ by applying learning from previous cheats and continuously getting better at it. While we’re on innovation, I also think that we will all get body scans at some point with cheap 3D scanners to… Read more »

HanSolo
HanSolo
5 days ago
Reply to  Basti

Was just in a “C” race and the top ten were all averaging 4.0 w/kg. Yay, I won a race by cheating…again! /sarc

Aoi Niigaki
Aoi Niigaki
6 days ago

It was interesting reading the judgements. Both riders claimed that they didn’t have the knowledge to edit the files and judging by the amateurish way they faked the files I’d say that was true. What I don’t understand is why did they do it? You get DQed from the race for not having dual power sources. Fine, mistakes happen, you move on, no-one actually cares about these race results anyway. But then they appealed the decision, clumsily faked the data and that is why they were suspended. I just don’t understand why you’d go through so much effort and risk… Read more »

james
james
6 days ago

The manager….. why no mention of the teams they were riding for? SARIS THE PROS CLOSET and VISION. The fact the manager of Berger tried to cover it up and failed miserably should speak volumes for the integrity of the team. I will remind you, she was riding for Saris The Pros Closet, supposedly one of the ‘best’ e sports teams. No wonder why the rest of the women’s team are now struggling to get a top 20. Let’s not forget the joke of Megan Rathwell as well. Same team, same obscene cheating. STPC = Laughing Stock.

Vision eRacing
Vision eRacing
5 days ago
Reply to  james

Just to make it clear, Lizi was not Vision when this happened, but she’d moved over to Heino.

james
james
5 days ago
Reply to  Eric Schlange

Megan Rathwell wasn’t… still racing for them after that obvious DQ in April. Only removed from the team after the manager tried to cover with fake data… if it smells like bull….then well it probably is….

Dave H
Dave H
6 days ago

How very sad

Mirco
Mirco (@mircolup)
6 days ago

epic fail

Scott DeLeeuw
Scott DeLeeuw (@sdeleeuw)
6 days ago

Reading that just makes me think of how Zwift could do it differently. Recording two sources and then uploading the second source to Zwiftpower after seems so manual and open to cheating. You would think Zwift would pair both power sources during the ride and record the second just as a verification. I suppose a person could still cheat with an ant+ emulator of some kind. Makes you wonder if these were honest mistakes they tried to cover up with file manipulations or if they were cheating and trying to cover it up with file manipulations?

Adrian Amos
Adrian Amos (@ahamos)
6 days ago
Reply to  Scott DeLeeuw

Completely agree that if they’re going to REQUIRE multiple power sources, Zwift should bear the onus of supporting that directly.

MarkE
MarkE
6 days ago
Reply to  Adrian Amos

Part of the new UX, I’m sure. So we’ll see it in 2025, but only after adding Patagonia.

bolanbiker
bolanbiker (@bolanbiker)
3 days ago
Reply to  MarkE

If Patagonia has cool roads to ride, I am good with that! LOL (appreciate the sarcasm)

naan
naan
6 days ago

Wouldn’t these riders have received a blanket ZRVG disqualification for any races on Zwiftpower before Zwift took over the system? Why shouldn’t the ban apply to community races now as well? Or is it just up to each race organizer keep track of these and decide for themselves for each race?

thewannabeironman
Super Member
thewannabeironman (@wanna_b_ironman)
6 days ago

Was one of them called Lance?

Chris
Chris
5 days ago

Except at the highest level, there not a lot of money in pro bike racing. In women’s racing, it’s worse; only the top riders at the highest level can really race as their only occupation. So the following question is begged: Did these ladies’ sponsors pay for them to have both a power meter and a smart trainer? Did they even have sponsors?

If not, you might be having riders just because they don’t have the same level of pecuniary resources as the country club set.

Zwift is expensive

James
James
5 days ago

Interesting that Zwift prefers Smart Trainer data as the primary power source. It makes sense, since unless you’re riding with a powertap hub on a wheel-on trainer, your pedal or crank-based PM will not account for drivetrain losses, which can be 10-20 watts. That seems to be quite fair to me, so good on Zwift.

Happily, I cleaned and lubed my drivetrain for the first time in like 8 months and gained a free 10 watts!

Benjamin
Benjamin (@benjamin_pitt)
5 days ago
Reply to  James

I actually think it’s not because of the accuracy – my kickr actually reads higher than my assioma duos. But because the next gen trainers are becoming less prone to calibration issues/cheating. The neo of all generations and kickr5 being what I’m talking about. My version of the kickr still has a spin done function that can manipulate the readings from accident or deliberate means (spindown at the wrong operating temp, or interfer with the spindown itself. Same with my assioma duos (input incorrect crank arm length). Given the top riders are generally provided with trainers (not always the case… Read more »

Shaun Harris
Shaun Harris (@vgrntbeauxner)
5 days ago
Reply to  James

This is my point. The accuracy of my pm is much better than the Kickr (which doesn’t use strain based measurement) and reads typically 3-9% lower than any of my PMs. So is the exercise now to replace my Kickr with one that (is hopefully similar accuracy of my PMs)? Or do I start competing off my Kickr (which is going to fk with my accumulated zwift power data and potentially put me back a cat – while most certainly blunting my performance).

Chris
Chris
5 days ago
Reply to  Shaun Harris

I had a KICKR Core that consistently read 200 watts higher than my Stages power meter, even after multiple firmware upgrades, factory resets, calibrations, etc., etc.

Besides, it strains credulity that the 2% to 3% power gain from reading at the crank arm or pedals is the primary issue here. After all, Zwift provides people with virtual bikes that simulate light weight and aerodynamics even if Zwifters themselves are riding a supermarket Huffy. Why not virtual mechanics?

Chris
Chris
5 days ago
Reply to  Chris

But wait there’s more. The whole weight issue is even more laughable. Everyone wants people to submit to a video weighing. But why? I have no space indoors to Zwift in air-conditioned comfort, and I don’t have the money right now to do anything with my garage. So in the summer, it’s 95-degree heat for several months, and I can drop 6 to 10 pounds in a single Zwift session, even if I’m hydrating like crazy. Why not force everyone else to compete under the same conditions, or at least leave me alone if I take this weight-loss into account?… Read more »

Graham de la Mare
Graham de la Mare
5 days ago

I agree with the cost issue. I’m not buying another power source when i occasionally race online. How about Zwift automatically DQF riders on the leader board at the completion of the race, for those who exceed the w/kg for that category. Currently only Zwift power shows the real standings. I’ve been in B grade races where the winner has a higher output than the top placed A grade winner!

Cat D ZPower Hero
Cat D ZPower Hero
5 days ago

*masters racer clinging onto any shred of validation through winning a community race – visibly sweating
Funny how many people think this has anything to do with them. Read the article, premier races. 99% chance that’s not you.

LHarding
LHarding
4 days ago

People getting wound up about a computer game that’s supposed to be a bit of fun and a training aid is just about the stupidest/funniest thing I’ve read today

Graham
8 hours ago

I use a Quarq power meter for zwift and my turbo is a Tacx Neo. I did once notice that the quarq read a lot higher than the Neo when I did a ramp test and noticed my power output was far lower on the Neo. My Quarq d-zero was reading about 30 Watts higher in the 250 Watts range than the Neo. The question is why should the Neo be more accurate than the Quarq? I calibrate the Quarq before every ride and would expect it to be more accurate than the Neo which is after all a turbo… Read more »

Free Zwift Trial

Subscribe to Posts

Receive each new post by email!

Newsletter Subscription

54
0
Would love your thoughts, please comment.x
()
x