Going with the FX flow

In this paper, we discuss using CLS intraday hourly flow data to understand FX markets. Using two different trading strategies (daily and hourly) against a generic trend model, we find that CLS’s flow data relating to funds and non-bank financial firms (NBFIs) tends to have a positive contribution to spot returns when viewed on an aggregate contemporaneous basis. By contrast, flows from buy-side and corporates tend to have a negative contribution. Later, we develop systematic trading rules for FX which use intraday hourly flow data from funds as an input between March 2013 and March 2018 as our in-sample period. Our daily flow based FX basket has annualised returns of 2.3% and risk adjusted returns of 0.63, and outperforms a generic trend basket which has returns of 2% and risk adjusted returns of 0.5 over the same period. We find that the drawdowns of an equally weighted portfolio of trend and daily flow based strategies were smaller, than any of the individual portfolios in isolation. Out-of-sample performance for the latter part of 2018 (April 2018 to October 2018) was slightly positive for the daily flow strategy, outperforming trend which was loss-making. Our hourly based flow trading FX basket in comparison, has risk adjusted returns of 0.81 and annualised returns of 1.6% in-sample. The risk adjusted returns of a combined daily and intraday hourly FX flow trading basket is 0.92.


To download the full Cuemacro paper, click here!