Python Code FTX Used to Randomly Fabricate Insurance Fund Exposed

• FTX ‘s software code for manipulating insurance fund data has been exposed.
• FTX used a random number function to generate the insurance fund it published to the public.
• Recently, a former CTO testified at the SBF trial that the insurance fund figure was inaccurate and entirely fabricated.

The crypto community has unearthed the software code that the now-defunct FTX crypto exchange allegedly used to manipulate its insurance fund, misleading the public about its actual values. In a recent tweet, verified X account BitMex Research shared a screenshot of the alleged FTX database code.

According to BitMex Research, FTX used a random number function to generate the insurance fund it published to the public.

Notably, the screenshot BitMex Research shared contained lines of code of the Python Programming language. The part of the code that indicated random number generation was “return f2d(numpy.random.normal(7500, 3000)) * daily_volume / Decimal(‘le9’).”

Specifically, that line of code generates a random number from a normal distribution with a mean of 7500 and a standard deviation of 3000. It then scales it to the size of its daily volume, divides it by one billion, and returns the output.

In parallel, FTX’s former Chief Technology Officer, Gary Wang, testified against FTX insurance fund on the fourth day of the trial involving FTX founder Sam Bankman-Fried (SBF). Amid rigorous questioning, Wang conceded that the insurance fund figure displayed on FTX’s platform was inaccurate and entirely fabricated. 

“For one, there is no FTT in the insurance fund. It’s just the USD number. The number listed here does not match what was in the database,” he stated.

When asked about the source of the fabricated number, Wang stated that FTX had been generating the figures independently, without relying on the actual assets held in the insurance fund.