Zwift bans two eRacers for data fabrication

Zwift on laptop
(Image credit: Josh Croxton)

Zwift has issued an eRacing suspension to two riders after data discrepancies suggested 'fabrication or modification of data'.

The virtual riding platform, which has seen a surge in popularity during 2020, is attempting to enact protocols that discourage data manipulation for gain.

Zwift’s latest communication on the matter of digital cheating concerns Lizi Duncombe and Shanni Berger. In both cases, Zwift made a decision to sanction riders from competing on its platform for six months, backdated to the date of the alleged offences.

With digital racing now becoming a meaningful competitive outlet, with prize money, the possibility of file manipulation was perhaps inevitable. The first UCI eSports World Championships is scheduled for December and Zwift is signalling that it will have a zero-tolerance policy.


Certain eSports race organisers stipulate that power data from a rider’s smart turbo trainer needs to be validated by a secondary power source, ie a power meter. The comparison of the two files is what helps the platform - and organisers - identify poorly calibrated trainers and power meters. 

Lizi Duncombe

The Duncombe investigation was triggered by a fourth-place she received for a Zwift Racing League Women’s Qualifier event, on 17 September. Zwift requires reference files from riders, and, upon closer interrogation of the data, a series of inconsistencies were revealed.

According to the Board Decision released by Zwift, Duncombe originally failed to fully dual-record her race, after she submitted a 90-second recording of her warm-up as the verification data, and despite judging this as a mistake, made the decision to annul her result from the race based on a lack of an viable dual-recording. 

Duncombe subsequently supplied a different dual recording, which came from Favero Assioma power meter pedals paired to a Garmin Edge 820. Zwift contacted Duncombe, to which she responded: 

"The first [dual-recording file] was the warmup file which saved before I started the Garmin again. When I charged my Garmin after the race, it then uploaded the race file which I sent a couple of hours later, which managed to get the whole race before the battery died."

On initial inspection, the data looked to be legitimate, but further examination highlighted anomalies. 

The dual-reference file sent to ZADA (Zwift Accuracy and Data Analysis) by Duncombe, contained a .FIT file tagged with a Version ID of '562'. 562 is the Version ID found in a FIT file created by Zwift, whereas Garmin Edge 820 computers use a Version ID value of '1250'. This suggested to ZADA that the file had been manipulated. 

There was also significant timing discrepancy with her support data file, as well as GPS data that a Garmin device wouldn't be able to read. Finally, the data was perfectly correlated (there was a Correlation Co-efficient of R=1.00) with the in-game Zwift recording, in that every time the original file read 297 watts, the Garmin read 294. 

Duncombe's subsequent responses included a claim that the Garmin had in fact paired with the turbo trainer, not the pedals, as well as suggestion of corrupted data. Both claims were disproven by ZADA. 

Shanni Berger

The Berger case arose after a second place in an Off the MAAP women's race. According to the Zwift release, the original annulment came due to Berger's use of a power meter as the primary power source for the race, despite the event stipulation that riders use a turbo trainer. In addition, there was a variance of 0% for the dual recordings, suggesting the second file was, in fact, a duplicate of the original.

This suggested that Berger had connected her power meter to both the Zwift device and her Garmin computer. An easy error to make, but without a viable dual recording, Berger's result was annulled. 

In response to the annulment, Berger's team manager and team owner provided a Zwift log.txt file claiming the power source was, in fact, a Saris smart trainer. 

However, the same log.txt file is also stored on Zwift's servers, upon cross-referencing, ZADA found that, alongside 26 other discrepancies, the power source was a Stages power meter, suggesting that the log.txt file from Berger had been edited. 

Berger repeatedly denied wrongdoing, with both herself and family members contacting Zwift stating that she was innocent and didn't have the technical capability to manipulate the files. However, according to Zwift, Berger later sent an email expressing her "apologies for any misunderstanding," and that it's "very possible that I [Berger] made mistakes with the software due to human error."

Both riders can still use the Zwift platform, but are suspended until March 19th and February 17th 2021 respectively. 

Today's best Black Friday deals

Today's cycling-related highlights from the Black Friday sales. For even more, view our guide to the best Black Friday bike deals where you'll find on all things cycling, including bikes, helmets, clothing and more. 

Today's best deals: USA

Today's best deals: UK

Thank you for reading 5 articles in the past 30 days*

Join now for unlimited access

Enjoy your first month for just £1 / $1 / €1

*Read any 5 articles for free in each 30-day period, this automatically resets

After your trial you will be billed £4.99 $7.99 €5.99 per month, cancel anytime. Or sign up for one year for just £49 $79 €59

Join now for unlimited access

Try your first month for just £1 / $1 / €1