How to detect if a cryptocurrency exchange is fudging numbers

Huobi versus incumbents

Crypto volumes in China

This picture has been making the rounds.  The top section is the volume of LTC at OKCoin.  The middle of BTC volume at BTCChina and the bottom is the BTC volume at Huobi.  The yellow line is when the PBOC stepped in and told 3rd party payment processors / intermediaries they could no longer transfer RMB into crypto exchanges.

In the last post I mentioned that there are speculative reports that exchanges in China (and probably globally) are fudging their volume numbers (e.g., why the sudden dramatic drop-off at certain sites relative to others).

Why is this done?  Before answering that I should point out that it is quite simple to create a server-side app that dynamically reports a volume number during specific periods of the day (e.g., higher during the work day to mimic traditional stock exchange peaks and troughs).  Exchanges currently have an incentive to fudge these numbers in an effort to attract eyeballs by claiming they have the biggest volume.  This then becomes a self-fulfilling prophecy as market participants (e.g., speculators, day traders, etc.) discover this volume number through a news source and then setup accounts to begin trading on the exchange.

How can you tell whether or not this is the case in China or elsewhere?   Merely looking at blockchain.info would be pointless because the transactions at exchanges are internal and do not affect the blockchain until there is a deposit or withdrawal.  And after all, lots of investors like day traders never actually withdraw their bitcoins each and every day due to transaction fees.  Thus the only verifiable way is to actually go inside an exchange and look at their accounting / exchange database to see the true turnover.

How to fake (some) numbers

Because exchanges merely self-report whatever they want to, ultimately this kind of fakery is easy to spot if you publicly expose your market depth.

Another way is if you, the exchange operator, run your own bots which arbitrage to make extra profit yet your bots do not pay any fees.  Whether that is actually fake is arguable (see below), however since the trades are actually conducted this kind of activity is typically not practiced in many “real” securities exchanges (e.g., akin to NASDAQ operating the exchange and yet maintaining an internal trading desk whereupon it does not have to pay fees).

And another way is to simply mirror other exchanges and if you are mirroring, you may be fudging the numbers to disguise it, or not. You might execute the mirrored trades or just have them for show.  In fact, you do not copy all the orders exactly, you randomize the quantities and prices slightly.  The trick is to randomize the numbers, but keep your risk low by minimizing arbitrage opportunities (this can all be done via an HFT system).

So again, sites like BTC123.com aggregate Chinese volume numbers and assume that the self-reported numbers are valid.  Maybe they are, but there is no real transparency currently.  And I’m not sure how you can add transparency either.  And more seriously, how can you (the exchange) going forward publish accurate information and get the same audience that has been misinformed in the past to believe you once again?

It should also be noted, it is not always clear what is or is ethical / fake with this speculation.  As suggested above with the NASDAQ example, the lines between the trading exchange and the banking institutions that utilize it may be blurry (e.g., conflict of fiduciary responsibilities in the event you have duties at both).1.

Blatantly lying about your volume is obviously fraud.  But is trading in your own exchange (with bots or otherwise) wrong?  Is mirroring (as long as you are willing to execute the mirrored trades) wrong?  It is not clear whether that is wrong / fake, perhaps at some point a guideline of best-practices will become adopted industry wide that clarifies this.

At the end of the day, what counts is whether or not I can get a fill at or close to market price.  Thus, as a trader, I care if you just make up the ticker numbers.  But I may not care about mirrored trades, as long as I can still trade against them.  One last point: using bots and/or mirroring trades enough to have large volume is expensive and risky.  It takes millions of dollars to copy the same volume as the major exchanges and increases as the price of tokens increases.  And in all likelihood, many investors only have accounts with one exchange.  Otherwise the arb opportunities could not be so relatively frequent as they have been the past 6 months.

Market makers

Perhaps the only other individuals who have the ability to test and audit volume numbers are market makers.  This could be a HNWI or a coordinated group of day traders consisting of as little as a few bored housewives who have money to burn (e.g., xiaosan).  What a market maker could do is video tape the entire buying and selling of chunks of 30-50 or more BTC to see whether or not the buy and sell orders are filled (e.g., if the liquidity does not exist, the orders could not be filled).  Note: this type of empirical activity can only be done going forward in time.  Unless you have access to a Psychohistory device, there is no way to verify who was fibbing in the past.

If you are willing and able to fulfill this pro bono role, you could test out volume claims at the three exchanges listed above throughout the forthcoming days and weeks and report your findings which would either way qualify one variable known unknown.

In fact, it would be in the exchange owners best interest to implore investigative journalists to  randomly place and fill buy / sell orders throughout a series of video taped interviews.  Specifically, large orders at several random hours each day (that are unannounced to anyone at the exchange).  This then could be repeated over a period of weeks to prevent any kind of rigging (e.g., exchange owners depositing and withdrawing cash simultaneously as known trades occur).

Finding volunteers for such a task would be difficult but it would likely work with roughly $100,000 based on current token prices (~$750).  Any large trader (who is willing to show you the their trades) could tell you whether the market volume is legitimate.  And ultimately, if you can fill a $100,000 trade, it is legit to you, regardless of what fakery the exchange operator may be doing.

[Special thanks to David Veksler and Scott Freeman for their thoughts and comments constructing this post.]

  1. To clarify, this activity is not related to the actions of Bernie Madoff.  He committed fraud and violated his fiduciary responsibilities but not by using an internal desk at an exchange.  On the other hand, if the exchange utilizes its knowledge of customers book orders to front run, this could bring legal liabilites to the exchange.  Nor does this imply that HFT systems (small or large like the new one built by the NYSE in New Jersey) tied into cryptotrading are axiomatically front-running as they are unable to see other customer buy/sell orders. []

2 thoughts on “How to detect if a cryptocurrency exchange is fudging numbers

  1. Pingback: The Rise and Fall, then perhaps rebirth of BTCChina? | Imaginary Capital Markets (夢幻市場)

  2. Pingback: Tim Swanson, Of Numbers, on Namecoin, Mastercoin and other Cryptoledger developments, Coindesk | Imaginary Capital Markets (夢幻市場)

Leave a Reply

Your email address will not be published. Required fields are marked *