AFX Index

In 1996 I co-wrote a paper which laid out the basis of a currency trend following index which we called the AFX. The paper went on to be published in 1998: Lequeux, P. and Acar, E. (1998) “A Dynamic Benchmark for Managed currencies Funds”, European Journal of Finance Vol.4. Our aim was to provide a framework to benchmark currency managers which at the time were primarily trend following in respect of their investment approach. Incidentally we probably coined the first FX smart beta index. To be clear the “raison d’etre” of the AFX was not specifically  to generate alpha but rather more to express a level of return and correlation consistent with the typical currency trend follower manager so as to be used as a tool for style analysis and performance evaluation. There clearly better techniques to accurately capitalise on currency trends. The paper we produced went on to be quoted in many research papers and the index to be used by many market practitioners and central banks in their own research. The methodology underpinning the benchmark was extending on previous published research which investigated the statistical properties of trading rules. The design which is transparent and relatively simple relies on 3 simple moving averages of order 32, 61 and 117 days to switch exposures to currency from long to short (and vice versa) as a function of the unfolding market trends.The simple moving average method was chosen amongst a large array of technical indicators because it was and still is one of the most popular trading rule amongst systematic traders. and therefore the returns of the index bear a high level of correlation to many currency manager peer group indices and also single managers.

Originally the index was computed using settlement price from the CME and also weightings derived from the triennial foreign exchange survey conducted by the BIS. As an attempt to make the index available to a broader audience of academics and market practitioners for research purpose I decided to somehow revisit some of the practical aspects of the AFX design. Namely data source availability , transparency and replicability. In that respect the re balancing of the weights in the index which are based on the most recent BIS triennial survey are now reset on the first business day of the year following the publication of the survey. Instead of using CME futures the index is now calculated using London market close (4PM) sourced from the Bank fo England database and short term interest rates sourced from the Federal Reserve Bank of St Louis database. The new iteration of the index is  strongly correlated to its predecessor whilst now having a greater level transparency and applicability which should be helpful to many academics and market practitioners who use it in their own research.

The most recent index monthly returns can be downloaded in a CSV format here whilst the return of the old version of the index can be found there. The R code that downloads the data and computes the returns of the index can be found here. The following link provides a list of most of the papers that have referenced and or used the index

plot of chunk arimaplot