Which Cryptocurrency Exchanges Are Faking Their Trading Volumes?

Finance

Here at HodlBot, we make it easy for anyone with an exchange account to create, diversify, and automatically rebalance their portfolios.

Our users use many different exchanges, and as such, we support many exchange integrations. Before we roll out any exchange integrations, we always evaluate it for liquidity.

Liquidity can be difficult to ascertain, as one cannot take reported trading figures at face value since many exchanges fake their numbers to appear busier than they actually are.

**Assessing Liquidity via Order Book Data **

While trading volume may lie, order books don’t.

You can always verify the integrity of an order book by executing orders against it. As far as I know, no exchange has been as bold or as stupid to fake bids & asks. If you can’t execute against an order book, then the exchange completely loses all validity.

Once you have order book data, you can evaluate metrics such as:

  • slippage (how much does the price rise/crash when you buy/sell $X)
  • bid-ask spreads (% spread between the lowest bid and highest ask)
  • order book depth (sum of all bids & asks in the order book)

Slippage, bid-ask spreads, and order book depths are all correlated with trading volume.

Binance order book depth vs. trading volume Binance slippage vs. trading volume Binance bid-ask spreads vs. trading volume

As trading volume increases, slippage decreases, bid-ask spreads decrease and order book depth increases.

If we’re able to identify a subset of exchanges that are reporting real figures and use this to build a predictive model between:

  • slippage vs. trading volume
  • bid-ask spread vs. trading volume
  • order book depth vs. trading volume

We can then use slippage, bid-ask spreads, and order book depth (metrics that an exchange cannot fake), to estimate what the true trading volume should be.

Establishing Trustworthy Exchanges

Binance, Bittrex, Kraken, and Bittrex are all BTI verified exchanges.

Data Source

We’re taking order book snapshots every hour for 10 days via the exchange API. Order book snapshots can be volatile, so we’ll be taking the average of all individual measurements

Time Period: September 17- 27, 2019

Training Our Regression Models

book depth vs. trading volume slippage vs. trading volume bid-ask spreads vs. trading volume

Spotting a Fishy Exchange

Notice how slippage & bid-ask points fall significantly higher than the prediction line that we trained on legitimate exchanges.

slippage vs. trading volume bid-ask spreads vs. trading volume

For order book depth it falls lower.

To Be Continued…

In the next article, we’ll put the following exchanges to the test in order to evaluate whether they are faking their exchange volume:

  • Lbank
  • Bitfinex
  • Poloniex
  • BiBox
  • Okex
  • HITBTC
  • Bitmart
  • Bitstamp
  • Gemini
  • COinex
  • Lbank
  • Liquid
  • Digifex
  • LAToken
  • HuobiPro
  • Binance
  • Kraken
  • KuCoin
  • Coinbase Pro
  • Bittrex

Written by Anthony Xie

I’m the founder of HodlBot.

I’m a big data nerd. I like to talk about all things data, finance, and crypto. You can find me on Twitter here.