The Long Memory of Order Flow in the Foreign Exchange Spot Market

We study the long memory of order flow for each of three liquid currency
pairs on a large electronic trading platform in the foreign exchange (FX) spot
market. Due to the extremely high levels of market activity on the platform,
and in contrast to existing empirical studies of other markets, our data
enables us to perform statistically stable estimation without needing to
aggregate data from different trading days. We find strong evidence of long
memory, with a Hurst exponent of approximately 0.7, for each of the three
currency pairs and on each trading day in our sample. We repeat our
calculations using data that spans different trading days, and we find no
significant differences in our results. We test and reject the hypothesis that
the apparent long memory of order flow is an artefact caused by structural
breaks, in favour of the alternative hypothesis of true long memory. We
therefore conclude that the long memory of order flow in the FX spot market is
a robust empirical property that persists across daily boundaries.