Recently, I came across an interesting trading strategy, intended for futures trading but theoretically applicable to retail Forex trading. The strategy’s author claims that even with completely objective and straightforward rules, a simple and complete “trend-following” strategy traded across a widely diversified group of liquid futures markets has produced an average annual return of approximately 20% per year over the past two decades, significantly outperforming global stock markets and equating to the kind of returns produced by professionally managed trend-following managed futures hedge funds.
As a Forex professional, I took a closer look at strategy to see what kind of edge it might have historically provided to retail Forex traders. The results make interesting reading because they illustrate exactly why it can be so difficult for retail traders to exploit edges that exist within markets.
For the sake of full disclosure I reproduce the strategy rules in full:
Risk: the 100 day ATR (Average True Range) should equal 1 unit of risk.
Entry: long at the end of any day which closes above the highest close of the previous 50 days; short at the end of any day which closes below the lowest close of the previous 50 days.
Entry Filter: long entries only when the 50 day SMA (Simple Moving Average) is above the 100 day SMA; short entries only when the 50 day SMA is below the 100 day SMA.
Exit: A trailing stop should be used of 3 times the 100 day ATR from the highest price since the trade was opened (for longs), or the lowest price since the trade was opened (for shorts). The trailing stop must be recalculated constantly as a “chandelier stop” and it should be a soft stop: an exit is only made when a daily close is at or beyond the stop loss.
This strategy was tested against the most liquid and popular spot Forex currency pair, the EUR/USD, over a long and recent period of time (from September 2001 to the end of 2013), using publicly available spot EUR/USD data with the daily open and close at Midnight GMT.
The results show that the strategy provided a winning edge on EUR/USD during the test period. Over 366 trades, a total return of 33.85 units of risk was achieved, giving an average positive expectancy per trade of 9.25%. This means that the average trade produced a return equal to the amount risked plus an extra 9.25% of that amount. Considering the strategy is completely mechanical, and that it represents only one instrument within what is traditionally the worst-performing trend-following asset class (currency pairs), this is not such a bad result.
However, fees and commissions must be factored in, to determine the return that could actually have been enjoyed. Assuming that:
the trading was performed by a fund with EUR/USD futures contracts, and a quarter of the trades had to be “rolled over” before the contract expired, incurring an additional commission, and a “round trip” commission of $20 per trade had to be paid, and an account of $10 million was traded with each unit of risk equaling a fixed 1% of the starting asset size, then the total return would equal $3,385,000 minus 366 trades multiplied by $25 each, representing the commissions. This would mean a reduction of the return by only about 0.1%, giving a total return of 33.75%. It could be assumed that if the rolling over strategies were less than perfect, there would be some additional losses.
Imagine now a retail trader with an account of $10,000 who wants to trade this strategy using a retail Forex brokerage. Luckily for this trader, the brokerage allows access to some kind of approximation of a futures contract that can be traded with a very small lot size, as well as very small lot size spot Forex trading, so there is no problem with scalability.
The next step is to work out some likely costs of trading for the retail trader trying to implement this strategy over the same period on the EUR/USD. First of all we can look at the cost of using spot Forex:
Each trade incurs a spread of 2.5 pips, and
Each position that remains open at the New York close incurs an overnight swap charge that varies from position to position, but which averages out to, let’s say, three-quarters of a pip per night.
For the sake of simplicity we can perform a rough calculation based upon pips. The 33.85% return calculation was based upon a profit of 9,088 pips. The spread alone is 2.5 pips multiplied by the 336 trades, which are equal to 840 pips. Next, we must deduct the overnight swap charges. Our retail trader had a position open over 9,889 nights, which would account for 7,417 pips. So we must deduct a total of 8,257 pips from our total profit of 9,088 pips, which leaves a net profit of only 831 pips!
For the sake of this rough calculation, if we assume that the return is evenly spread over each pip, this represents a greatly reduced net profit to our retail trader of only 3.09%, as opposed to the 33.85% return achieved by the $10 million fund we looked at previously.
Our retail trader might have an alternative, which would be to buy synthetic mini futures contracts which do not incur overnight swap charges, but which have much wider spreads; something like 14 pips per trade for EUR/USD. Taking another look at the numbers and also assuming that one quarter of all trades must be rolled over, our retail trader would face a fee of 14 pips 458 times, equaling a deduction of 6,412 pips. This would represent a net profit of 2,676 pips. Assuming again that all return is spread equally over each pip, our retail trader ends with a net total return of 9.97%. So using synthetic mini-futures would have been much more profitable, but would still represent an annualized return over the test period of less than 1% profit per year! Furthermore, this return would be less than one third of the amount enjoyed by the large fund.
Breaking it down
Why are things so bleak for our retail trader? There are several reasons, and examining each reason carefully can help any aspiring retail trader understand how certain edges within the market can be effectively whittled away by the wrong choice of brokerage or execution methods.
Actual futures contracts are too large to be available to most retail traders, and position sizing cannot be achieved properly with amounts less than several million dollars in a diversified trend following strategy. Mini-futures are a potential solution, but if they are not very liquid then they are unlikely to present the same trend-following edge as ordinary futures. Exchange Traded Funds are another partial solution, but even so, the retail trader is going to have to pay some kind of spread for access to an appropriate market far in excess of the $20 round trip commission payable by a sizable client of a Futures Exchange.
This brings us to the topic of spreads. Frankly, there is no reason whatsoever why even a retail trader should be paying more than 1 pip for a round trip trade on an instrument as vanilla as the spot EUR/USD. Brokers charging more than this really have no valid excuse. It should be said that spreads in the retail sector have been going down in recent years. While this is good news, even if the retail trader in our example had been paying 1 pip instead of 2.5 pips, this would have raised profitability only by an additional 1.5%, and cannot truly be backdated all the way to 2001 in any case.
This brings us, finally, to the real culprit of the reduced return: the overnight swap rate, which is widely misunderstood, and so is worth a detailed examination.
Overnight Swap Charges
When you make a Forex trade, you are effectively borrowing one currency to exchange for another. You must therefore logically pay interest on the currency you are borrowing, while receiving in return interest on the currency you are holding in return. There is usually an interest rate differential between the two currencies, which means you should either be receiving or paying some extra fee each night representing the differential, and of course the exchange rate is a factor as currencies rarely trade at 1 to 1. The only time there would be nothing to pay or receive would be if the exchange rates were exactly equal at the rollover point, and there was no interest rate differential.
It would seem that sometimes you pay the difference and sometimes you receive it, so overall this swap cancels itself out. Unfortunately it is not as simple as that, for several reasons:
Currencies with higher interest rates tend to rise against currencies with lower interest rates, so you tend to find yourself in more long trades over time where you are borrowing the currency with the higher rate of interest, meaning you tend to be paying more often than receiving.
Retail Forex brokers charge or pay quite wildly different rates to their clients long or short of a particular pair. Many brokers are very opaque about this and do not even display the applicable rates on their websites, although the rates can be found within the brokerage feed on every MT4 platform. It is worth mentioning that, to be fair, there are legitimately different methods of calculating this charge. However if you look at the table compiled at myfxbook showing a range of overnight rates charged by some retail Forex brokers, you will get a sense of the wide variety in the market.
In addition to charging or paying the interest rate differential, some brokers also add an “administration” fee, which can mean that you will not receive anything even when the interest rate differential is positive in your favor! Ironically, these tend to be the same brokers that will bill you for account inactivity, and exactly what administration is involved when the trades are rarely even booked in the real market is highly questionable. The end result is to skew the fee even further against the client.
Most traders are highly leveraged, which means that they are borrowing the vast majority of the currency they are trading. Traders tend to forget that one of the negative consequences of leverage is to push up the overnight swap charges, as they must pay interest on all the borrowed money, and not just the margin that they are putting up on the particular trade. Of course, this is a legitimate element of the charge.
The practice of charging a fee for every night a client keeps a position open is not only open to abuse, but can be an effective way to dramatically reduce the odds that a trader might seek to move in their favor by an intelligent use of long-term trend trading, which usually pays off over time if executed properly. It might be said that some retail brokerages are using the widespread ignorance about these charges as a way to add to their balance sheets, and that regulatory agencies should be taking steps against this. On the other hand, it could also be said that a market maker cannot be expected to make a market in a way where they can be systematically put out of pocket by the long-term statistical behavior of the market. It might be that many of the differential rates between brokers are reflected by the currencies that their clients are long or short of at any particular time. It can be seen that one broker might be offering a better deal than another on one currency pair, but not on another, which seems strange.
A systematic study of this area would make a very interesting read. Meanwhile, a retail trader seeking to systematically hold positions overnight should make sure they fully investigate what is on offer when they are shopping around for brokers, and be aware that the speed of a price movement in their favor can have a big effect on the profitability of any trend or momentum strategies that they might be utilizing.