Is it better to express a currency position through multiple pairs?

Question

I use a trend-following approach where I look for trends in various currency pairs such as GBP/USD or EUR/USD and then take a position in the Spot currency. I measure the performance of my strategy by looking at the ratio between the % returns of my trades and the standard deviation of the returns of those same trades. A positive ratio above 0.2 actually leads to excellent returns over a long period of time.

Someone suggested to me that rather than following one pair, I should take one base currency such as GBP or EUR and generate some kind of composite average (like the US Dollar Basket) and search for trends in that. Once I have signal, I should then take positions in the Spot with the USD and other crosses eg. for GBP it would be 50% GBP/USD + 25% GBP/EUR + 25% GBP/JPY. It was suggested that this would improve my returns/volatility metric (ignoring the fact that non-USD crosses generally have slightly higher spreads and transactions costs than major FX).

Why is this? I understand diversification but I thought that the idea was that you combined positively expectant strategies which were poorly correlated with each other to diversify. Here, you are just combining random currency pairs where you have no particular evidence whether its going to move favourably for you or not. eg. your research signal indicates GBP will move up, but doesn't say EUR/JPY/USD will move up. So essentially you are just dampening down your returns - but if I wanted to do that, could I not just take a smaller position size?

Answer 1

My 10 cents is to think about how, if the market is buying GBP, for example, then it will buy GBP against any or all other currencies. It's not a matter of buying GBP only against USD or EUR although those are, say, the major terms currencies. Alternatively, if the market is selling GBP then they will sell against all other currencies. Thus if you use a currency basket for some base currency then that is a better reflection of a "trend" relative to that base currency alone, to reduce a bias in general trends going on with different terms currencies. So if there's really a trend in a base currency then it makes sense that trend is better reflected using a basket of terms currencies.

This is also the way traders will cover positions. You run a book in a base currency against all other currencies, and if you decide you need to reduce risk in GBP positions, for example, then you will sell GBP against any other currency in your book. It just so happens you probably have bigger positions in the majors, so, GBPUSD. For example, any different GBPXXX deals (say GBPAUD or GBPJPY) are probably "decrossed" into GBPUSD and USDXXX where possible so your trader has fewer positions to have to manage. Thus a major pair like GBPUSD has many more position changes than GBP against other currencies as trades that were originally done in GBP against another currency are decrossed out into GBPUSD and that other currency against USD, or EUR, for risk management of all currencies in their most basic pairs.