Real Bitcoin Volume: The Other 30%
By Darius Avens on ALTCOIN MAGAZINE
On March 19th, Bitwise Asset Management delivered a 226 slide report to the SEC as part of their ongoing efforts to gain approval for a Bitcoin ETF. Of note to many who read this report has been the revelation that Bitcoin spot trading volume is a mere 5% of what most known price aggregators quote, as well as the assertion that Bitcoin’s “real” trading volume is split between just 10 exchanges. For those who have not yet read this report, you can check it out here: https://www.bitcointradevolume.com/
While the existence of fake volume is no surprise to anyone, the report is striking due to the order of magnitude of volume that is classified as such. As Clay Collins points out in his essay on exchange transparency ratings (https://blog.nomics.com/essays/transparency-ratings/), the question of trading volume is a complex one, and for a community that is usually fairly critical, Bitwise’s report has not been met with much scrutiny. As he says, while the quality of Bitwise’s research is very high, it is also in their best interest as a potential ETF provider to portray the Bitcoin spot market as concise and easily measurable.
As of the publishing of this post, Bitwise suggests the 24hr spot trading volume of Bitcoin is $675m, while price aggregators such as Coinmarketcap and Coingecko put that number around $19b. We believe Bitwise is overly conservative in their estimate, while most price aggregators wildly inflate it. In reality, the true number is somewhere in the middle of these two. The analysis presented below attempts to provide a better estimate of this true volume using an objective metric that any third party can verify.
The Bitwise Report
Let’s begin by breaking down the 10 exchanges identified by Bitwise as representing real trading volume: Binance, Bitfinex, Coinbase, Kraken, Bitstamp, bitFlyer, Gemini, itBit, Bittrex, and Poloniex. A closer web traffic analysis via Similarweb.com gives the following monthly (March 2019) web traffic figures for each exchange:
Given that Bitcoin is a currency without borders, one would expect global trading volume to come from a variety of different countries. Yet, it appears the vast majority of users for these 10 exchanges (with the exception of bitFlyer) are located in Western countries, mostly residing in the US and EU. Indeed, 6 out of these 10 exchanges (Coinbase, Kraken, Gemini, itBit, Bittrex, and Poloniex) are US based, with only 2 from Europe (Binance & Bitstamp), and 2 from Asia (Bitfinex & bitFlyer). The numbers above are aggregated in the chart below across all 10 exchanges:
As expected, web traffic for these 10 exchanges is almost entirely dominated by the US and European countries. Moreover, this distribution does not account for discrepancies in GDP per capita between each country, which would only further skew supposed trading volume towards this region (USA & EU) that is already entirely dominant. Notably absent are China & India, the world’s 2nd and 6th largest economies respectively.
This leads to two possible conclusions:
1) Bitwise’s report is accurate, and there is almost no trading volume contributed by non-Western countries.
2) Bitcoin trading volume is globally distributed, and looking at only 10 exchanges provides an insufficient measure of it.
Believing the latter to be true, we decided to take a closer look at what other exchanges might be legitimate.
Another Way Of Looking At Things
At a fundamental level, an exchange’s spot trading volume should be more or less in line with the number of individuals who are visiting the exchange’s website. That is to say, there should be a rough ratio between exchange volume and exchange web traffic over the same time period that is similar across between different exchanges.
Using the 10 exchanges identified by the Bitwise report, it is easy to calculate what a baseline for this ratio should be for a legitimate exchange. Furthermore, we can also see what this ratio looks like for sites that are known to be faking volume (the Bitwise report mentions Coinbene, Bitforex, Lbank, and Exrates as a few examples). By comparing these two baselines to the universe of known exchanges, we can extrapolate which exchanges, if any, should be brought up for consideration alongside these 10 as legitimate sources of volume.
Exchange volume for this analysis was sourced from CoinGecko’s 24hr reported exchange volumes for the top 150 exchanges for the day of March 31, 2019. A snapshot of this data is provided here: http://web.archive.org/web/20190331020232/https://www.coingecko.com/en/exchanges
Web traffic volume for this analysis was sourced from SimilarWeb’s monthly web traffic overview for each exchange for the month of March 2019. An example of Binance’s web traffic is provided here: https://www.similarweb.com/website/binance.com#overview
Displayed above is a chart (scaled logarithmically) depicting each exchange’s monthly web traffic and daily trading volume. Highlighted in red are the 10 exchanges (Binance, Bitfinex, Coinbase, Kraken, Bitstamp, bitFlyer, Gemini, itBit, Bittrex, and Poloniex) highlighted by Bitwise as having real volume. The numbers for these 10 exchanges are as follows:
Average daily volume per user: $553
Sample standard deviation: 618.9883759
Here are the same numbers for the exchanges identified by the Bitwise report as “suspect”:
With the exception of itBit, which is a bit of an outlier due to its low web traffic, all confirmed legitimate exchanges have a daily volume to daily traffic ratio below $1,000. On the flip side, all the highlighted “suspect” exchanges have a ratio at least 18 standard deviations above this, with many being much higher.
Given the clear difference between these two groups of exchanges, it suggests that a volume to web traffic ratio is, in fact, a good predictor of whether an exchange is a legitimate or faking volume.
Using This Ratio As A Predictor Of Potential Legitimate Exchanges
Out of the 150 exchanges included in this sample, 92 falls outside 2.262 standard deviations of the sample means (the threshold for a 95% confidence interval with a sample size of 10). That is to say, if this ratio is 100% determined by whether an exchange’s volume is inflated or not, there is only a 5% chance one of these 92 exchanges would be that far off the mean if they are in fact legit. Therefore, we can say with certainty that the majority (61%) of exchanges are indeed faking their volume.
However, there are 48 exchanges not included by the Bitwise report that pass the ratio test. They are as follows:
A cursory inspection indicates a few exchanges that should be excluded from any Bitcoin volume calculation. These are as follows:
Max Maicoin — linear exchange volume over time is indicative of manipulation or misreported data
Coinroom — exchange is no longer operational
Tidex — insignificant Bitcoin trading volume
Bancor Network — no Bitcoin trading pair
Idex — no Bitcoin trading pair
Instant Bitex — historical trading volume anomalies
Bitex Live — insignificant Bitcoin trading volume
Crex24 — insignificant Bitcoin trading volume
EtherFlyer — no Bitcoin trading pair
OpenLedger DEX — insignificant Bitcoin trading volume
Infinity Coin — no Bitcoin trading pair
UEX — exchange is no longer operational
Tokenomy — no Bitcoin trading pair
Crypto-Bridge — insignificant Bitcoin trading volume
Further analysis will need to be done to clear the remaining 34 exchange’s volumes as truly being “real”. However, given their ratio of web traffic to reported volume, we believe it is likely most of them would stand the test of more scrutiny.
An Alternative Measure Of Real Bitcoin Trade Volume
Using these 44 exchanges (the original 10 + 34 smaller exchanges) to measure Bitcoin spot 24hr volume, we get a number of $875m (as of this post’s publication). To see this figure updated every 15 minutes, as well as a historical chart, we have created the following page: https://www.mahaadvisors.com/volume
Bitwise’s methodology provides an estimate of $675m (via https://www.bitcointradevolume.com), which is a 30% difference.
It is worth noting that both of these figures still vastly underestimate true total volume, as they are calculated under the assumption that exchanges with fake volume have 0 real volume. This is not the case. Many excluded exchanges, such as Huobi and Hitbtc, still garner millions of users per month in web traffic, who undoubtedly provide some real volume. Although the same type of traffic-volume ratio analysis might allow for proportional incorporation of suspect exchange’s trading volume, we have chosen not to include them in the calculation above.
Finally, while we believe real Bitcoin trading volume is higher than that suggested by Bitwise’s report, we completely agree with their analysis that the vast majority of the volume is fake. By highlighting the good and denouncing the bad, we hope that exchanges will begin to be held accountable for their actions, as well as act with more transparency.