For every trading pair mentioned in Table 1, we calculated two time-series: price volatility hourly logarithmic returns (ln∆P), and trade volume hourly logarithmic returns (ln∆V).
Price Volatility Return Formula:
Volume Return Formula:
You can find the time series descriptive statistics for Price Volatility Returns and Volume Returns in the Annex (Table A1).
Based on the obtained results we computed the correlation between the two time-series using the following formula:
The resulted correlation coefficients we weighted according to each pair’s volume decomposition in particular exchange total volume structure, and calculated the Weighted Price-Volume Correlation (Wcorr) for all crypto exchanges:
The scatterplot on Fig. 1 displays Wcorr calculated for each covered exchange.
The value of Wcorr 0.7 and above we consider as a strong positive correlation between volume and price volatility indicating presumably natural trading activity. Only 4 out of all analyzed exchanges have Wcorr greater than 0.7: BitMEX, Coinbase Pro, Bitfinex, and BitFlyer. At the same time, there is a cluster of exchanges (17 out of 31) that have the Wcorr value of less than 0.5. We consider this case to be an illustration of a weak positive correlation indicating trading activity presumably influenced by non-market mechanisms such as trade volume manipulations. The lower the Wcorr value the more questionable the authenticity of trading activity on the exchange is.
For visual comparison, let’s match scatterplots of analyzed time series values of the best and worst performers of our analysis, BitMEX (Wcorr = 0.836) and EXX (Wcorr = -0.040) (see figs 2-3). There is an obvious positive correlation on BitMEX’s corr values scatterplot.
In contrast, based on the EXX’s scatterplot we can conclude that the trade volume performance does not depend on price fluctuations on the exchange. This fact is confirmed by the extremely high kurtosis ratio for the volume changes of EXX (on average 173.7 vs 9.0 for BitMEX), which means that the distribution is characterized by a large number of values close to zero.
Just for illustrative purposes, let’s look at BitMEX and EXX official charts (Fig. 4-5).
Apparently, something incomprehensible is happening with EXX`s trade volume – it holds perfectly stable despite price fluctuations.
To analyze price and volume manipulations on the crypto market, various methodologies can be applied. In the current analysis, we used the correlation approach to estimate the relationship between price volatility and volume performance on the largest crypto exchanges. We calculated weighted correlation coefficients and displayed them on the scatterplot. As a result, we identified a cluster of crypto exchanges with weak Price-Volume Correlation values suggesting suspicious trading activities, as well as a small number of presumably honest players.
The goal of this article is to examine a new research method for assessing the crypto exchanges’ trade volumes authenticity and encourage other researchers to investigate the issue using profound approaches.
What should the global crypto community do to eliminate the trade volume manipulation and make the industry mature?
Enter your email address to subscribe to Hacken Reseach and receive notifications of new posts by email.