AFX Currency Management Index
The AFX index is a currency management index, which aims at a high correlation with other currency management indexes such as Parker or CISDM (formerly MAR), with correlation over the past 15 years being, on average, 0.74 (a correlation study can be downloaded from here). The traditional benchmark for currency traders has been nil return. This is misleading since the actions and trading strategies of many active currency managers are similar, with most active managers basing their decisions on trend following trading systems. Taking this into account, it becomes apparent that a more suitable index, against which to benchmark an active currency manager, would be the average performance of an active, trend following, currency manager. AFX is an index that replicates the trading actions of an active manager and provides a more realistic benchmark for active currency traders.
Extending previous work by Lequeux and Acar (1998), the AFX index relies on 3 moving averages of 32, 61 and 117 days. As mentioned in Lequeux (2001) "The moving average method has been chosen amongst a large array of technical indicators because it is the most popular trading rule among futures traders". As shown in Lequeux (2001), this method aims at a high correlation with real currency management performance indices such as Parker or MAR. The goal of this index is to represent the time horizons of investors, this it does with the choice of moving average time periods. Not all investors trade with the same time horizons and this can create a problem when forming a benchmark index. To represent adequately all time horizons, the statistical properties of technical indicators have to be used. It was decided to build the benchmark around ex-ante measurable criteria of risk reduction and transaction costs. Each day moving averages are calculated for each currency pair involved in the benchmark. These averages are then compared to the current price of the currency pairs. Should the currency price be greater than the moving average the benchmark assumes that a long position is to be initiated or held for the next 24 hours, otherwise the reverse will apply. The value of the benchmark is the equally weighted average of the returns earned by the set of 3 moving averages on each currency pair included in the benchmark.
Recently Dunis and Chen (2004) show how this index can be used, as an alternative to statistical accuracy benchmarks, to gauge currency trading models. This approach has been extended further by Dunis and Miao (2004), who have utilised the AFX index as a benchmark, in order to show the added value which volatility filter strategies can provide to currency traders, both at portfolio and single asset levels.
Lequeux, P. (2001) "Trading Style Analysis of Leveraged Currency Funds" Journal of Asset Management, Volume 2, Number 1, 1 June 2001 , pp. 56-74(19) (DOI: 10.1057/palgrave.jam.2240035)
Lequeux, P. and Acar, E. (1998), "A Dynamic Index for Managed Currencies Funds Using CME Currency Contracts", European Journal of Finance, 4(4) pp 311-33.
Dunis, C. L. and Chen, Y. X. (2004) "Alternative Volatility Models for Risk Management and Trading: An Application to the EUR/USD and USD/JPY Rates", Derivatives Use, Trading & Regulation, 2005, No 11/2, 126-156.
Dunis, C. L. and Miao, J. (2004) "Optimal Trading Frequency for Active Asset Management: Evidence from Technical Trading Rules", Journal of Asset Management, 2005, No 5/5, 305-326.
The above text is from the cached version of the AFX Site at Liverpool John Moores University, formerly maintained by Professor Jason Laws.
An XLS spreadsheet with monthly data from Jan 1984 to date on the AFX Index
is maintained by Pierre Lequeux.
KMEC 9-72, 44 West 4th Street, New York, NY 10012