I’ve written a lot about Carry, and its uses in trading FX recently. In this article, I will look at how most efficiently to structure a carry portfolio across the varying risk levels. A simple re-cap of what FX carry is, is that you buy high yielding currencies, while simultaneously selling low yield currencies, in order to receive the difference in interest. The idea behind this comes from interest rate parity pricing of forward FX contracts (I went into detail here), but basically, the future value of a currency should discount any interest rate difference. I.e. at its heart, when you buy a higher yielding currency, the main objective is not that the currency increases in value, just that it stays above the forward value of the currency so that you gain Carry. However these particulars are not too important, what is important is how we structure a portfolio to suit our needs most effectively.

As discussed previously, the generic go to carry portfolio for G10, is buying the top 3 yielding currencies, whilst simultaneously selling an equal amount of the 3 lowest yielding currencies.

As a reference point, here is a table of 2Y swap rates for the currencies within the universe I consider.

So currently, if you were long carry, you would be holding; AUD, NZD, NOK while being short CHF, JPY, EUR. Normally these would all be of equal USD size.
This strategy has returned 4.17% so far YTD as shown below.

There are ways to tweak this however depending on risk tolerance. First of all though, we must consider that with a higher interest rate, comes, at the margin at least, higher risks.

This is because, investors demand a higher return to accompany a risk they are going to take on, hence interest rates on lower credit rating companies command a higher interest rate. (This is not always the case of course, but in a sell-off, investors tend to think irrationally and sell the highest yielding asset first, so regardless, there is more intrinsic risk to them)
So the simplest thing to do, when considering how we want to position with regard to risk, is to adjust the weightings of the six currencies in the portfolio.

My preferred weightings are 50%, 33.3%, 16.7% for use in the defensive, and aggressive portfolios.
For example here;

Over a backtested 3yr period, we can, as we would expect see the following.

These are the 3 equity curves as well:

*defensive portfolio, where the least extreme interest rates hold the highest weighting

*normal portfolio, each currency equally weighted

*Offensive portfolio, most extreme interest rates, have the highest weightings

One thing to consider here, while beta is difficult in FX, carry often correlates with risk, which in turn often correlates with equity markets. Now I know today, the relationship is no-existent, but it will return at some point, and also throughout a lot of the back-testing period, it was correlated. Hence comparatively to equity markets, we can potentially introduce leverage to our portfolio. Normally I would take the two extremes and equate them.

In this example then, that would mean comparing the carry max DD (~10%) to the S&P500’s which was 22% over the testing period. Thus, a potential of 2.2x leverage could be employed to bring it up to equity markets. This is however optional and completely dependent on the investor, more or less can be used. But we must remember, leverage only acts to amplify the waves, it does nothing to improve the risk-adjusted returns, as per the ideas in this article (i.e. weightings)

Bear in mind however these were un-hedged performances, one could employ options to protect against downside, while only paying a small premium. This point was highlighted in the last article, with the implied skew at all time lows, and implied volatility pretty much at all time lows. However, so far we have only really considered a simply G10 strategy, and how we manipulate weightings to improve performance.

The next way I will look at is using volatility to filter out some noise from the respective interest rates. By taking the expected interest rate that you’d receive, and dividing it by the expect volatility of the spot rate, you can get the risk adjusted return.

These vary over time, and can be used to selectively choose one currency over another. Here is a chart from BNP Paribas showing the respective carry/vols of each currency

Over the test period, this improves performance of the offensive strategy from about 17% to about 21%, while reducing the maximum drawdown to 9%. Clearly this works well, but we can do better.

As highlighted earlier, carry tends to correlate with general risk trends, so we can set parameters based upon risk sentiment and complacency to optimize when we are in the market.

One simple way is to consider global volatility. If the specific volatility index is below a certain requirement we buy carry, and if volatility is above a specific level we exit our carry positions. Setting this is really where the manager earns his money, and is going to be very dependent on risk profiles.

However using a moving average as the benchmark Above/below level, such as the 200DMA works well in a simplified model. You could also extend this idea far enough that you look to sell you carry positions once volatility gets really low, as it can be a sign of complacency in the market.

Employing this strategy on a basket of the S&P500 and AUDUSD, we see how returns are doubled, with barely higher volatility and lower drawdowns.

So, we have got two further ways with which to optimize our portfolio, but we can also look to add the Emerging markets into our Carry basket. I would suggest, for a balanced portfolio to have some EM, but maybe 25% of the net exposure. Once again though, EM trades at a higher beta to G10, and also has a higher yield, so that is always something to bear in mind.


In conclusion, we have looked at various ways of optimizing a carry portfolio to maximize returns, to sum up they include
- Weighting the currencies according to interest rates. Dependent on whether we are aggressive or defensive.
- Adding a volatility filter to see the carry/vol ratio of each currency so that we can get better risk adjusted returns
- Adding a further sentiment based filter, so that we are long carry at the right times. (this step can also be the discretionary step if one prefers)

But what we can see, is that we've taken a basic carry strategy, which anyone can quite easily do, and make it a bit more sophisticated and we've managed to reduce drawdowns and improve risk adjusted returns.

Thanks for reading, and I hope this can help in any way to optimize portfolio performances.
Translate to English Show original