The Art and Science of Trading Carry

Let’s talk about carry trades.

First, what exactly is a carry trade?

A carry trade is a trade that pays you to hold it. A position where, if nothing changes except the passing of time, you expect to make money.

Let’s go through some examples.

FX carry

The classic example is the FX carry trade, where you borrow a low-yielding currency to buy a high-yielding one and profit from the interest differential.

Here’s how it works:  

  • Currency A yields 2% pa  
  • Currency B yields 10% pa  

Ignoring costs, we can:  

  • Borrow in currency A at a cost of -2% pa  
  • Buy currency B with the borrowed funds and receive 10% pa

If the exchange rate and the interest rates stay the same, we expect to make 8% pa. That is, if nothing changes, we expect to make 8%.

Of course, at this point, you should have some questions about these assumptions!

Surely, you would expect an efficient market to adjust for this differential, with currency B depreciating relative to currency A. And you’ll find a bunch of academic and economic literature to this effect – search “uncovered interest rate parity” to see what I mean.

And on face value, this seems pretty reasonable. You wouldn’t expect the market to pay you for doing something as naive as this.

Would you?

Well, a lot of traders have done this trade, borrowing the low-yielding thing to buy the high-yielding thing. And it turns out that, at least in the past, these traders were more right than the academics.

Here’s a plot of returns to a currency carry factor constructed from the G9 currencies (G10 minus USD, short the ones with the lowest policy rate, long the ones with the highest):

And this paper looks at the same trade across a basket of 20 currencies (and some other carry trades).

You can see that it was an embarrassingly good trade for something so naive, right up until the GFC.

You could argue that the returns have diminished since then.

Note the sharp drawdowns – you could argue that this skewness in the return profile is the real cost of doing this trade. When it whacks you, it whacks you hard.

If you dig into the data, you find that, contrary to economic theory, it suggests something like this:

We find that the lower-yielding stuff, instead of increasing in value to maintain parity with the higher-yielding stuff, actually tended to decrease.

And the higher yielding stuff, rather than decreasing in value, tended to go up. At least historically, on average.

So historically, you had carry returns not only coming from the differential interest yield but also from price change.

Which seems totally at odds with what we’d expect from an efficient market.

And currencies aren’t the only place where this trade has played out.

Yield curve carry

Another good example is yield curve carry. If we plotted interest rate futures’ implied yield against time to expiry, you’d find that it generally looks like this (at least most of the time, recent history notwithstanding):

We find that the further we go out, the higher the implied yield. So if we’re long bonds, we’re surfing down this yield curve and making money as time passes.

If nothing changes, we make money in this trade. If something does change, then we might lose money – for instance, the level of the curve might shift, or the shape may change, as we saw during the recent tightening cycle.

This is a plot of historical returns from about 1930 to being long US ten-year treasuries.

And this is a plot for the period 1994 to 2013 (the only period I could easily find) of returns to rolling global interest rate futures:

There’ll be some structural things going on there as well, but you can see that our naive carry strategy of rolling bonds has made money over time. It’s also been whacked about occasionally.

Selling volatility

Another example is selling options premia or selling volatility generally.

Options price in the possibility of big future price moves. So there’s a time value – if the underlying moves a lot in a specified time, then the option will pay a lot. So, the longer an option exists without anything exciting happening, the more it will decrease in value.

So, writing options could be classified as a type of carry trade – you make money selling them if nothing changes.

Buying stocks

You could also argue that buying stocks is a carry trade.

Imagine you borrowed some cash and bought stocks. You received dividends on those stocks but had to pay interest on your loan. You would say that your stock position carries if the dividends you receive exceed your financing costs.

If you look at stock index futures, the longer you go out in the future, the cheaper they are (usually). That’s because the futures don’t receive the dividend distribution. So, the distributions get priced into the futures via a carry effect – the future will increase in value as time passes until it converges with the spot price at expiry.

So, to the extent that financing costs are expected to be lower than distributions, holding stocks or rolling a stock index futures position can be classified as a carry trade.

And notice that this carry trade has the same tendency to get whacked as our other carry trades. This is a defining feature of carry trades – most of the time they pay you a little, but occasionally you get knocked about. And you’ll find that diversification will only help so much, as most carry trades tend to get whacked at the same time. So sizing becomes important.

Do we see carry effects today?

Yes, we do. But many of the traditional carry trades are a lot harder these days. Notice the FX carry performance from about 2010. Bonds as a carry trade were awful for the last couple of years.

One place where carry has recently been an exceptional return source is in crypto.

Many crypto exchanges provide markets on derivatives called perpetual futures, which are essentially futures contracts that never expire but instead transfer a funding cost between longs and shorts, depending on where the future is trading relative to spot. If the future is trading at a premium, then longs pay shorts. If it’s trading at a discount, then shorts pay longs.

The idea is that this funding mechanism forces the futures to trade close to spot. And in an efficient market, we’d expect the future to converge to the spot before the close of the funding period.

But we don’t see that. Instead, we tend to see autocorrelation in the premium or discount – that is, it tends to be somewhat sticky. Here’s an autocorrelation function for the funding rate on an hourly BTC perpetual:

There could be many reasons for this; for example, more people want to trade the futures (with leverage) in a particular direction than can be readily absorbed.

This feature makes for some interesting carry trades.

An obvious one is shorting perpetual futures trading at a premium and longing the spot to hedge the risk, thus collecting the funding.

To the extent that the premium is sticky enough that you accrue more funding than you pay in transaction costs, you make money.

The opposite trade is more difficult because shorting the spot requires you to pay to borrow it.

This was a fantastic trade for a while, but it’s gotten harder as more people chase it and the market becomes more efficient.

A messier variation is to create a long-short basket of perpetuals trading at a discount or premium, respectively. This trade will see a much higher return variance than the spot-perpetual version because the basket components will dislocate and do all sorts of weird idiosyncratic things. It also requires a lot of oversight and maintenance. And you’ll find yourself almost constantly rebalancing a basket of illiquid altcoins. The trade can easily become a full-time job. It’s been a great trade though.


A carry trade is a position that pays you if nothing changes. Traditional examples include:

  • Borrowing low-yielding currencies to buy high-yielding ones
  • Rolling stock and bond futures positions
  • Selling volatility

Carry trades tend to pay you a little most of the time but whack you hard occasionally. And they tend to be quite correlated, so diversification will only help so much. Sensible sizing that acknowledges the asymmetry in the return distribution is therefore critical in managing these trades.

While traditional carry has been difficult recently, crypto markets have provided a lovely source of carry returns via derivatives such as perpetual futures, particularly if you’re willing to do the difficult work of managing an illiquid basket and accept some return variance.

2 thoughts on “The Art and Science of Trading Carry”

Leave a Comment