On "Real" Crypto Trading Volumes

One of the most maddening things for professional investors regarding the crypto markets is that it’s obvious where liquidity is pooled and price discovery happens, but it’s not well understood or properly calculated anywhere.

Efforts like Bitwise’s herculean study into trading volume manipulation helped pierce the veil on legitimate vs. spoofed order books and led to the creation of metrics like Messari’s “Real 10” Volumes.

But that new metric omitted some of the largest markets in the world. Particularly, venues with *some* overstated volume, but *enormous* legitimate trading volume at the same time. Giants like Bithumb, UpBit, and CoinOne in Korea, Liquid in Japan, and most importantly, Huobi, OKEx/OKCoin, and Gate.io in China.

Other data services usually either ignore rampant overinflated volumes, conflate spot and derivatives volumes, or both. We decided to curate the data curators and take a common sense, admittedly imperfect, but *directionally accurate* approach towards estimating real trading volumes. But we knew we needed to add some missing exchanges to our methodology, so have decided to score exchanges for trading volume legitimacy based on a review of 10 different third party stats: exchange rankings, ratings, and liquidity estimates from CoinGecko, CoinMarketCap, CryptoCompare, CryptoWatch (Kraken), Nomics, and FTX’s volume monitor, plus onchain transaction data from Chainalysis and other on-chain data providers, and anecdotal data from top traders.

As a result, we’ve added 10 exchanges to our "real volume” metric, but applied a 50% haircut to the Korean and Chinese exchanges to better approximate their true volumes. We believe this better reflects the *magnitude* of adjustment necessary vs. simple web traffic comparisons, which usually discount these volumes by ~90%.

(For what it's worth, this methodology also better aligns with scoring from FTX’s volume monitor, and the Alameda guys are pretty fucking good at trading.)

The results of our methodology change propel Huobi, OKEx, Liquid, Bithumb, and UPBit into the top 10 by volume without catapulting them past original “Real 10” constituents inaccurately. (Binance remains #1 by real volumes across most listed assets.)

Importantly, we will review our methodology quarterly in order to add or remove constituent exchanges, and tweak volume adjustment parameters.

In a perpetual game of data cat and mouse, we believe the most conservative yet accurate method for updating real exchange volume estimates will be analogous to the “hire slow, and fire fast” mantra in recruiting: slowly add additional exchanges as their volumes become material *and* multiple independent data services verify their data integrity, while quickly removing exchanges whose volumes spike outside of normal bands or raise red flags under third-party review. (Ours included.)

This is a necessary change, but one that may seem counterintuitive at first glance. I can assure you, though, that it is more accurate, if less precise, than alternatives.

-TBI

P.S. We welcome your feedback! And if you'd like to hear more about the challenge of "Calling Balls and Strikes" like this in crypto, then you should join us for this exact session at Mainnet on June 1 with Dave Balter from Flipside, Ambre Soubiran from Kaiko, Nic Carter from Castle Island Ventures, and our own Qiao Wang. :)