Take a close look at trading activity on BKEX—a cryptocurrency exchange founded in 2018 and registered in the British Virgin Islands—and you’ll see something odd.
Compare its transactions side-by-side with those of Binance, one of the largest crypto exchanges in the world, and you’ll notice BKEX’s trading history is a replica, printing the same numbers delayed by a few seconds.
According to CoinMarketCap, BKEX has $1.1 billion in daily volume, making it the 20th-largest exchange on the planet. Yet it seems to be simply copying Binance’s trade history and passing it off as its own, in perhaps the laziest attempt in history to fool people into thinking it’s a lively place to trade digital assets.
In the wake of other reports on phony trades, including one by digital asset manager Bitwise indicating that 95% of all transactions are bogus, Alameda felt it could create better research by leveraging its trading data and experience.
The startup was co-founded in 2017 by Sam Bankman-Fried, 27, an MIT alum and former trader at high-frequency trading outfit Jane Street. Gary Wang, 26, a fellow MIT grad and former Google software developer, is his co-founder.
The firm has $100 million in assets, and over the past month, it has traded $1 billion a day on average, making it one of the largest crypto trading firms in the world.
Exchanges make money by charging users to trade, and they have many reasons to artificially inflate volume. More activity means a higher rank on the still-popular website CoinMarketCap, which can attract new users.
Exchanges also charge fees to new cryptocurrency projects that want to get listed in their marketplace, and the perception of popularity helps them command higher rates.
Since an exchange’s place of business is just a website or an app, and many located outside the U.S. are unregulated, it can publish any numbers it wants and call them trades.
Meanwhile, CoinMarketCap continues to insufficiently vet exchanges’ transaction volume, often taking companies at their word and publishing dubious numbers.
According to Alameda’s research, another method exchanges use to juice their statistics is sneaking in large, fake transactions amid a flurry of smaller ones.
CoinEgg, a Hong Kong-registered exchange that trades $1.1 billion a day reported by CoinMarketCap, recently employed this tactic with litecoin (LTC) trades.
During a period when Alameda observed 15 different offers to buy and sell litecoin in a maximum quantity of 134 LTC, several trades printed as large as 2,000 LTC, as if a buyer appeared out of thin air.
Trading marketplaces typically publish their “order book,” showing a list of bid prices at which people are willing to buy an asset, plus a separate set of offer prices where people are willing to sell.
For instance, Bill might be willing to buy bitcoin at $10,000, while Mary wants to sell at no less than $10,050.
For a trade to happen, a new buyer must be willing to pay the $10,050 that Mary is offering, and the vast majority of trades that clear will align with orders that previously showed up in the order book, unless two users place offsetting orders at the same exact time.
Yet on some exchanges, trades get executed at prices and sizes that fall outside anything sitting on the order book.
On Digifinex, a Singapore-based crypto trading venue, Alameda observed bids and asks for bitcoin between $8,296 and $8,298, but several trades printed at $8,290 and $8,293, prices lower than what anyone was willing to sell it.
On LAToken, a Moscow digital exchange, Alameda saw bids and offers with a maximum size of 1.6 bitcoin in the order book.
Implausibly, several trades sailed through at sizes up to 20 bitcoin. LAToken founder Valentin Preobrazhenskiy says his platform only has a “tiny share” of 20-bitcoin orders and that exchanges use inflated volumes as a marketing tool.
“The situation would change when large exchange-ranking sites would add a section for trading volumes based on trades reported to regulators,” he says.
On Singapore-based ABCC, the best bid and offers Alameda saw were for sizes less than one ether, yet several transactions materialized with sizes of up to 11 ether.
Among trading venues, there’s also the well-worn method of simply printing transactions that fall in the middle of the bid and ask prices, which Alameda’s research spotted in IDAX and Coineal.
In total, Alameda’s report gives examples of fishy trading patterns on 60 different crypto exchanges. Aside from LAToken, none of the exchanges named above responded immediately to Forbes’ request for comment.
The methodology behind Alameda’s research was to test each exchange on six different criteria. First, they manually looked at an exchange’s order book and observed where trades printed. If more than 10% of transactions didn’t appear on the order book, it failed on this dimension. Another test involved observing the percent of an exchange’s trades that took place at the best available bids or offers.
A third criterion was to analyze how much Alameda itself traded on a given exchange, since the startup deals in “virtually every cryptocurrency,” considers its algorithms “exchange-agnostic” and estimates that it trades 5% of all global crypto volume.
“If we trade more than .5% of an exchange’s reported volume, we consider that exchange to pass according to this criterion,” the report reads. For more details on its methodology, see the full report.
Beyond exchanges’ bad behavior, the report has other provocative insights. It claims that crypto—including both “spot” trades of actual digital assets and derivatives, like bitcoin futures—trades $38 billion in real volume a day, and 87% of that happens on Asian exchanges, with just 9% happening on U.S. venues.
The strict regulatory environment in the U.S. is likely a contributing factor in Asia’s dominance, Alameda says.
Compared with Bitwise, which released a follow-up fake volume report in May 2019, Alameda thinks more crypto volume is real.
For large exchanges like OKEx and Huobi, which were founded in China, Alameda estimates about 70% of their transactions are authentic.
Bitwise is much more skeptical, as is the Blockchain Transparency Institute, which has estimated that more than 60% of Huobi’s volume is fake and more than 90% of OKEx’s volume is fabricated.
A Huobi spokesperson says it doesn’t engage in wash trading, but that it has observed some market-makers doing so on its platform, and it takes steps to stamp them out.
An OKEx spokesperson says the company isn’t involved in and doesn’t tolerate wash trading, adding, “Recently we have joined the Data Accountability & Transparency Alliance (DATA) led by CoinMarketCap, as a commitment to revealing as much data as possible.”