ETFs, High-Frequency Trading, and Flash Crashes

Abstract

This article presents a model of distributional properties of returns on financial instruments tied to exchange traded funds (ETFs) via high-frequency statistical arbitrage. As the author’s model shows, the securities subject to an ETF arbitrage exhibit a well-defined behavior, largely dependent on the behavior of other securities comprising the ETF. The model can be used to improve the risk management of long-term portfolios and, in particular, allow hedging of flash crashes. Furthermore, the author shows that in electronic markets that allow high-frequency trading, the intraday downward volatility for the underlying securities constituting an ETF is bounded from below; as a result, it is less extreme than that of securities not included in any ETFs. Also, downward price movements are more extreme in markets that restrict high-frequency trading than in markets in which it is present.