Are Bitwise’s stunning findings on Bitcoin trading trustworthy?

On March 19, in support of its exchange-traded fund (ETF) proposal, Bitwise Asset Management has submitted to the U. S. Securities and Exchange Commission a report with a staggering claim that up to 95% of trading volumes reported by Bitcoin exchanges were actually fake.

Bitwise’s core motivation behind filing their report to the SEC was to show that the genuine BTC market is rather more orderly and resembling the established financial markets than many believe. If this is established, then the SEC’s reluctance to approve properly structured Bitcoin ETFs appears to be unfounded. The report’s conclusions also arguably have broader implications about how one should judge what happens in the crypto space, and whether the predominant perception of it as a chaotic jungle is reasonable.

They are also relevant to the question of the prevalence of holding BTC vs. speculating. If valid, they suggest that there is much less speculative activity involving BTC than it is usually believed.

Bitwise’s key findings and justification, and the report reception

Coinmarketcap reports Bitcoin daily trading volumes in BTC/USD and BTC/USDT (Tether) currency pairs in the order of several billion U. S. dollars (6 billion on average for Bitwise’s analysis purposes). At the same time, when one looks at the list of top Bitcoin exchanges and prices reported by them, one immediately notices two things even without detailed analysis. First, the names of those exchanges are barely recognizable, and then, there is suspiciously high variance in price given the reported volumes.

Going deeper, Bitwise’s main finding is that only ten BTC exchanges of those it analyzed have actual volume. Those include Coinbase, Binance, Kraken, Bitstamp, Bittrex and Poloniex. In contrast, the other exchanges Bitwise studied report an extent of activity that cannot be considered genuine. From this, Bitwise concludes that only the volumes reported by the good actors should be counted and that they represent only 5% of what Coinmarketcap claims.

Bitwise put forward an impressive array of converging lines of evidence in support of this conclusion. One is the number of round trade sizes that should be higher than its expected value because of human psychology. Another involves the bid-ask spread that should decrease with the rising volume. And the legit exchanges do show these features, while the suspect ones do not. Especially striking is the difference in spreads between Coinbase Pro and Coinbene. The latter purports to have more than 30 times more volume than the former, yet the spread is enormously larger.

Probably the most convincing piece of evidence is the fact that trading patterns on all the ‘legit’ exchanges can be seen to be very similar in three important respects. Histograms shown in the report demonstrate that the ‘legit’ exchanges follow natural patterns. They mirror each other in terms of the distribution of trades according to their sizes, volume spike alignment (when activity peaks and falls during the day) and spread patterns over time. In contrast, the suspect exchanges analyzed are all over the place with each having its own idiosyncratic behaviors.

In addition to the statistical evidence, there are also other reasons for believing that much of the Bitcoin trading volumes are not legit. The key of those is that projects seeking to promote their tokens usually receive price boosts and generate a lot of excitement when their tokens get listed by exchanges classified by Bitwise as honest, especially Binance and Coinbase.

The reaction to Bitwise’s report has been mostly positive, with reputed crypto information resource and data aggregator Messari even rolling out the so-called “Real 10 Volume” metric. Coinmarketcap’s representatives admitted in an email to Bloomberg that the volume data concern was valid without opining on the exact accuracy of the real volume estimates by Bitwise, however.

Collins’s criticism of Bitwise’s findings

Despite the impressiveness of the research presented by Bitwise, not all the commentators are fully convinced. On May 1, Clay Collins from data provider Nomics penned a challenge to Bitwise’s findings on several grounds.

First, he argued that the report has received too little scrutiny given the gravity of its claims, and Bitwise’s inherent bias in favor of its conclusions. Especially damning, in Collins’s view, has been the tendency of commentators to extrapolate Bitwise’s conclusions to the cryptocurrency markets in general, which is far beyond their scope.

Collins also suggests that Bitwise did not directly examine all the important BTC exchanges and it is thus improper to conclude that only ten exchanges have actual volume. Some of the exchanges not considered may actually have activity that matches the one they report.

The most serious objection that Collins makes is that the methodology used by Bitwise is not falsifiable. In other words, according to him, a third party cannot take the same data and use Bitwise’s methodology to attempt to reproduce the findings. What he may mean by this is that there is no systematic statistical methodology in the report for randomly selecting a sample of exchanges and analyzing it.

A potential middle ground

Who is right? Bitwise’s report’s findings should be taken very seriously, especially the histograms showing similar trading patterns on the exchanges identified as trustworthy and the unique patterns on the untrustworthy ones. While this is not an ironclad proof, there are reasons to believe that economic activities, including Bitcoin trading, should exhibit a certain degree of regularity. Especially given that there are a lot of traders participating and the trading is conducted at scale.

However, what is probably not justified is the logical leap that Bitwise makes from finding that the volumes reported by the untrustworthy exchanges cannot be trusted to saying that they should, therefore, be completely discounted. First, this conclusion just does not follow logically. The irregularities in trading patterns probably indeed reflect made-up figures and wash trading but this does not necessarily mean that no real trading has been happening on suspect exchanges whatsoever. More importantly, as Collins rightly noted, this certainly does not imply anything about the exchanges that Bitwise did not directly study. In fact, for what it is worth, the author of this article personally used one of the top 15 non-legit exchanges, HitBTC, to make small trades in Swarm City tokens.

Hence, we can conclude that Bitwise’s report strongly suggests that the BTC trading volumes reported by Coinmarketcap are probably highly exaggerated but its recommendation to count only trading volumes from 10 exchanges does not withstand scrutiny.

This post was first published on Proof of News.