U.S. Markets close in 2 hrs 35 mins

Up to 86% of total reported cryptocurrency trading volume is likely fake, according to analysis of exchange website visits

Larry Cermak

This free preview of The Block Genesis is offered to our loyal readers as a representation of the valuable research and journalism our Genesis members receive daily. If you’d like to receive all Genesis content on our site and via daily newsletter, join today!

Bitwise released a 104-page white paper on Friday, which examined exchanges for fake volume, and found that roughly 95% of reported trading volume in bitcoin is fake or non-economic in nature. Bitwise used screen scrapers to collect live trading data from over 80 exchanges for a period of several months. The only 10 exchanges that passed Bitwise’s tests are Binance, Bitfinex, Coinbase, Kraken, Bitstamp, bitFlyer, Gemini, itBit, Bittrex, and Poloniex.

Still, Bitwise's study ignores the real volumes on the hundreds of exchanges which they deem to have majority fake volumes. The asset manager estimated that there is an increase of 12% in volume if the approximation of real volume on HitBTC, Huobi, and OKEx is included. But none of the other exchanges are considered. Bitwise says that while the real volume on exchanges that engage in faking volume is "non-negligible," the firm does not believe it would materially change the conclusions of the paper.

[related id="1"]In a wide-ranging study, The Block analyzed the website traffic of the most frequently used cryptocurrency exchanges from data provided by SimilarWeb. SimilarWeb, while far from perfect, is a good indicator of real activity on the exchanges. SimilarWeb uses multiple algorithms and data sources but it’s important to recognize that the data are still estimates. An analysis of SimilarWeb accuracy found that it was the most accurate tool available outside actual analytics access. 

While web traffic could theoretically be manipulated as much as volume, it’s fairly hard to do at a large scale and in a convincing manner. It’s also possible that the web traffic coming from China could be understated because of the firewalls and VPNs.

The Block analyzed 48 exchanges and their monthly traffic in the last 6 months (ranging from November 2018 to April 2019) to help paint a picture of real volumes in the cryptocurrency market. Binance had the largest number of estimated visitors in the examined period with nearly 185 million; followed by Coinbase with 143 million. All other exchanges each had less than 50 million visits in the last six months.

In total, there were nearly 800 million visits on cryptocurrency exchanges in the last six months. April's traffic was the highest in that period but very close to the traffic in November. In November, prices were tanking across the board and volume was 23% lower than in April, which could mean that most of the buying activity is not retail investors on exchanges but rather over-the-counter (OTC) or through APIs, which is usually done by larger traders.

If we look at the reported traded volume, the volume numbers shared by the exchanges themselves, the data shows a much different trend than the number of website visits. The top 10 exchanges sorted by reported volume in the last six months are BitMax, Binance, OKEx, DigiFinex, Huobi, ZB.COM, Bit-Z, CoinBene, LBank and BitForex; only one of which does not fake volume according to Bitwise. The next exchange vetted by Bitwise is #20 on the list.

By dividing the traded volume by the number of estimated visitors, we can get a rough idea of which exchanges are faking volumes. Again, results can be thrown off by understating the traffic on Asian exchanges or by sophisticated exchanges that fake web traffic in addition to volume. But directionally, this method should be accurate to spot the suspects. Nearly every exchange deemed by Bitwise not to fake volume saw less than $1,300 of volume per website visit.

ItBit stands out as an outlier, but there's an explanation. ItBit focuses on large clients and most of them trade through APIs, which obviously doesn’t register as a visit, thus resulting in lower web traffic.

Still, itBit hardly stands out when you compare the Bitwise approved exchanges with the ones likely engaging in significant volume faking (Coineal, Fatbtc, BW, BitMax, LBank, DOBI Exchange, Bit-Z, DigiFinex, Idcm, DragonEX, ZB, CoinTiger, IDAX, Bibox, CoinBene, BitForex, Bithumb, Negocie Coins, Liquid and OKEx) and the exchanges likely engaging in some volume faking (Huobi, Upbit, HitBTC and Hotbit). Coineal leads the way with a whopping $98,000 of volume per one website visit, 740 times higher than Coinbase.

It is very hard to estimate the magnitude of volume faked by each exchange, but we could use itBit ($5,700 of volume per website visit) as the ceiling and estimate for the maximum amount of real volume on a given exchange (assuming website visits are not severely understated for Asian exchanges). For example, Coineal's maximum real volume is 5.8% of what they have shown ($5,700/visit divided by $98,000/visit). 

We could also use Coinbase ($132 of volume per one website visit) as the absolute minimum of volume per one website visit and get an estimate for the minimum amount of real volume on a given exchange. By following this methodology, it appears Indodax, BX Thailand, Luno, CEX.IO, Zaif, and KuCoin are not faking any volume unless they are also faking traffic. YoBit, EXMO, GOPAX, Gate.io, Coinone, Bitbank and Korbit are also likely not faking significant volume. Again, using Coineal as an example, the minimum amount of real volume is 0.13% ($132/visit divided by $98,000/visit). 

By multiplying the minimum share of real volume by the self-reported volume figures, we can get an idea of the minimum total amount of real trading volume. In the last six months, the total reported trading volume was $1.96 trillion, out of which a minimum of $272.5 billion appears to be real. In other words, a maximum of 86% of the trading volume appears fake.

At the very least, there is an increase of 10.2% in volume on top of the Bitwise-vetted exchanges if the approximation of real volume on all the other exchanges is included. What's worth noting is that 65% of the total real volume in the last six months came from Binance and Bitfinex, which both have virtually no regulatory oversight.

While analysis of web traffic likely won't yield precise results, dividing the self-reported volume by web traffic currently appears to be a good method for identifying exchanges that fake volume. At least until the exchanges catch up and start reliably faking traffic as well. There are other, more precise, methods such as analyzing slippage, measuring total available liquidity within X% of best price or analyzing order book patters. But simply dismissing the real volume on exchanges that fake some portion will not yield indicative results for the whole industry.