Zwift bans two more virtual racers for ‘data manipulation’ 

Zwift has banned two more riders from virtual racing for alleged data manipulation. 

The online training and racing platform has released details of two more sanctions it has handed to riders accused of manipulating their power data after competing in online races. 

Both riders, Antonina Reznikov and Selma Trommer, have been banned from competing in official Zwift e-sports events for six months, after the Zwift Performance Verification Board found anomalies in the numbers. 

Reznikov, from Israel, has been sanctioned after competing in the Zwift Racing League (Season 2) Women’s Race 1 event, where she finished in fourth place.

Announcing its decision, the Zwift performance board said following the race Zwift’s automated systems found that dual-recorded data from the rider’s power meter may have been edited, leading to further investigation. 

Zwift then found that the power values submitted were 32 per cent higher than those originally recorded, according to the board.

After discussing the issue with the rider, Zwift Reznikov initially denied editing the file but later admitted to changing the power data before submitting it to Zwift. 

Cycling Weekly contacted Reznikov, an Iron Man triathlete and regular Zwift racer, who said she did not edit the power file, but instead deleted the warm-up segment of her ride, which meant the power data from the second recording didn’t match the original file. The file could not then be recovered to prove her innocence, she said. 

Reznikov added that she pushed 4.2 watts per kilogram in this race, but has previously recorded 5w/kg in races with verified power data, meaning this was not a performance where she would need to deliberately edit the power file to gain an advantage. 

She said: “I didn’t even argue and didn’t try to prove anything because I was sure that everything was fine and trimming the warm-up part file is not something that could ruin everything.

“I’m very sorry that it happened. I can only hope that those who know me, they know that for me a race of 4.2w/kg is not a problem and I can ride much stronger.

“It is very sad and painful that such technical nonsense can denigrate an athlete.” 

Reznikov added: “I believe that every person has the right to personal space and not everyone has the right to report in the media everything that he heard somewhere. They hurt me with my family and everyone who believes in me as an athlete” 

The Zwift Performance Verification Board announced on January 26 that Reznikov would be banned from Zwift e-sports for six months for “bringing the sport into disrepute” by “fabricating or modifying data.”

Her ban has been back-dated to the date of race on January 11 so will expire on July 10, 2021. 

In the second ruling, rider Selma Trommer was banned after her participation in race two of the Zwift Racing League (Season Two) Women’s Race on January 18, where she finished sixth. 

Zwift said its automated systems once again found anomalies in dual-recorded data that warranted further investigation.

Further analysis found that the power file was nine per cent higher  than those in the original recording, Zwift said.

Trommer denied editing the file, but her team said that she had problems uploading the file via Garmin Connect so instead used another application to import and export the files before sending them to Zwift. 

The verification board says it considered the possibility that the files were edited accidentally, but decided this was not the case because the power had been increased by a fixed percentage.  

>>> ‘I want to clear my name, it’s devastating’: Zwift racer speaks out after being banned for data manipulation 

Zwift said Strommer did then acknowledge the file had been edited, but that she did not admit any personal fault or explanation about how the data was changed, Zwift said. 

Strommer has been banned from e-sports for six months, back-dated to January 18 so she will be able to race again after July 17.  

Cycling Weekly has reached out to Strommer for her response.