The beta of an investment measures its sensitivity to “market returns”. Unlike in equity, in FX the relevant benchmark for a beta cannot be a long-only index. Instead, an FX-specific beta can be based on common types of currency strategies, such as carry and trend. Currency betas measured against such benchmarks can be valuable for portfolio construction and measuring positioning risk.
The tracking of relevant “beta” for FX positions ties in with the subject of setback risk in general and positioning risk in particular, as summarized here.
The below are excerpts from the paper. Headings, links and cursive text have been added.
The beta problem
For some markets, it is fairly easy to find market beta, i.e. what would constitute a proxy for the typical return for an investor. In equities, we could for example use S&P500. Similarly, in bonds one could use bond indices from various investment banks. Whilst the investment community tends to be dominated by ‘long only” investors in bonds and equities in FX traders cannot be long only. In order to buy a currency, they simultaneously need to be sell another one.
“Without an obvious beta which can represent the returns of a typical FX investor, we instead need to think about creating proxies. The idea of these proxies is to represent the typical types of generic FX trading styles that are used by many FX investors”
Carry, trend and value
“Foremost amongst…generic FX trading strategies…is the carry trade. Purely on a visual basis, there does appear to be some relationship with the S&P500….the largest drawdown in both strategies occur around the Lehman crisis in 2008… The idea of a carry strategy within FX… is to buy high yielding currencies, funded by selling low yielding currencies”
“Trend following can be seen as a subset of technical…strategies, which also encompasses mean-reversion based strategies. If many investors are jumping on to a trend, then it attracts more investors in a virtuous cycle. If we want to create a generic FX trend following model which proxies the behaviour of most FX market participants, one approach could be to use a combination of…popular moving averages. We… follow a similar approach to Lequeux and Acar (1998), who noted that a combination of using 32D, 61D and 117D moving averages for FX, effectively acted as a reasonable proxy for FX fund returns. 1/3 of capital is invested in each trading rule. In our case, we use a different universe of currencies, namely the USD, EUR and JPY crosses in G10.”
“The concept of a ‘value’ model is totally different from that of trend following. With trend following, we essentially buy high and sell low, on an expectation of a continuation of the trend. With value meanwhile, we seek to buy ‘cheap’ currencies and sell ‘expensive’ currencies. However, how do we determine this? One of the simplest currency valuation models is PPP (purchasing power parity), which can accomplish this.”
Combining FX styles
“We have examined the various FX styles in some detail, noting both the benefits and drawbacks of them. FX carry can suffer from severe drawdowns, whilst FX trend can be impacted by ranging markets. FX value meanwhile needs to have a long term time horizon to be traded. At the same time, we have seen that we can for example use FX trend as a hedge for FX carry during times of severe risk aversion. Furthermore, FX trend can profit from large moves in FX markets.”
“In Figure 11, we have plotted the returns of the FX styles portfolio, we note that the long term risk adjusted returns of 0.64 exceed that of any of our three strategies individually. Furthermore, the maximum drawdowns of -14% are far lower.
In Figure 12, we have plotted the long term correlations between the various styles and also the portfolio. Given it uses daily data, the correlations might understate the relationship between the various strategies. Indeed, the correlation between S&P500 seems relatively low with the FX styles.”
Proxy of FX fund returns with trend and carry
“We use as our proxy for FX fund returns, HFRX Macro Currency Index. In Figure 16, we display the results of a linear regression between the FX fund index (y variable) and FX trend and carry models (x variable). We use a 12-month window (using monthly data). We have then created a portfolio [of trend and carry strategies] using these weights…In Figure 17, we have then given the rolling 12-month correlation between the FX fund index, the weighted portfolio, FX trend and FX carry. We find that for nearly all the whole sample, our weighted portfolio has a higher correlation, suggesting that this combination explains FX fund returns better than FX trend or FX carry in isolation.”
 P. Lequeux and E. Acar (1998) – A dynamic index for managed currencies funds using CME currency contracts – The European Journal of Finance 4, 311–330