Rise of the Machines: Algorithmic Trading in the Foreign Exchange Market
This purpose of this study is to analyze the impact of algorithmic trading on price discovery in financial markets. The authors use a time series of high-frequency data to model the potential effect which algorithmic and non-algorithmic trades have on the efficiency of price discovery in the foreign exchange markets. The data is comprised of one-minute price data from September 2003-December 2007 for EURUSD, USDJPY, and EURJPY. The authors investigate whether algo trading contributes to the volatility of asset prices caused by a temporary diversion in prices from their fundamental values. The results suggest algorithmic trading improves price efficiency in the frequency of triangular arbitrage opportunities and autocorrelation of high-frequency trading returns.
A vector auto-regression (VAR) specification model is used to determine the effect of algo trading on triangular arbitrage opportunities, controlling for time, trading volume, and exchange rate volatility in each pair. The results of this model indicate there is a direct negative correlation between algo trading and triangular arbitrage opportunities as the frequency of these opportunities for EURUSD, USDJPY, and EURJPY fell significantly as algo trading increased. These models are shown in the appendix and their statistical significance is indicated.
Previous literature on this subject suggests there are two main differences between computer and human trades— firstly, computers react to new information faster than humans and secondly, computer trades may be more highly correlated than human trades, as computers need to be pre-programmed and may react similarly to an input. Chaboud et al. suggest that generally, computer strategies are correlated so computers do not trade with one another as frequently as computer-human trades. Human strategies are also correlated, as humans do not trade with another as frequently as computer-human trades. This would imply the effect of algorithmic trading on price efficiency may depend ultimately on the strategy type used by the trader, not merely the presence of algorithmic trading. The appendix contains models measuring the correlation between human and algo trade strategies and provides the results of the regressions run on the data to derive statistical significance of these results.
Risk Warning: The FXCM Group does not guarantee accuracy and will not accept liability for any loss or damage which arise directly or indirectly from use of or reliance on information contained within the webinars. The FXCM Group may provide general commentary which is not intended as investment advice and must not be construed as such. FX/CFD trading carries a risk of losses in excess of your deposited funds and may not be suitable for all investors. Please ensure that you fully understand the risks involved.