Over the years, analysts have spent countless hours looking for an edge in the markets. Much of that time has been devoted to finding new ways to measure momentum or find overbought/oversold extremes in the market. These are the most common use of indicators, and have led to formulas for RSI and MACD, among many others. They are well defined and the formulas are readily available on a number of web sites. Trading software makes it possible to trade an indicator without even knowing what it really is, but traders who are just starting out and not able to invest in software can find a broker that has charting capability built into their platform. In a worst case, you could use spreadsheet software to calculate either of those indicators and almost any other indicator if you are patient enough to do that.
Trading sites that offer lessons in how to use an indicator will usually show “well selected” examples on charts. Invariably, the indicator works exactly as expected. It’s rare to see any test results and that’s not because it is difficult to program indicator signals. Back tested results usually aren’t shown because indicators often don’t work as well in real life as they do in theory. RSI and MACD do have among the best track records in testing, but it is still difficult to make a living using the default setting with standard indicators.
Some test results can be used and we can use the tests to learn more about both indicators and trading systems. We’ll use RSI with a level of 30 signaling a buy after the market becomes oversold and selling when the RSI falls below 70. To calculate RSI, the default setting of 14 periods will be used. Tests will be run on the Euro/US Dollar currency pair, just because it’s a liquid and tradable currency pair. Testing will start in October 2003 and run through the end of August 2011.
The goal is to see if an indicator-based, mechanical system can deliver profits and then to learn a little more about developing a trading system by testing some of the parameters. All test results will include a deduction of $5 per round trip for trading costs. There is always a cost to trading and testing needs to recognize that.
Initial test results will be summarized by starting with a brief description of the test followed by the total profit or loss over that period and the maximum drawdown as a percentage of profits.
- Signals based solely on RSI levels and always in the market, 5-minute chart: Losses of more than $14,000 resulted This timeframe was tested because forex trading is usually considered to be very short-term. The reality is that it is difficult to overcome trading costs as the frequency of trading increases. Short-term trading might be something profitable for high frequency trading firms which have lower costs and technology specifically designed to capture the smallest price movements. But, in practice, short-term trading rarely works for individuals trying to trade from home.
If trading costs could be ignored, this system would make about $2,800. It’s hard to overcome the costs of about 40 trades a month, or just two trades a day on average.
- Signals based solely on RSI levels and always in the market, 30-minute chart: A gain of almost $1,000, after trading costs, shows that it’s easier to be profitable by trading less. But drawdown is more than $3,000 so the simple idea carries too much risk. Trade frequency, and associated costs, is reduced to only 10 trades a month.
- Signals based solely on RSI levels and always in the market, 60-minute chart: Profits of $3,259 exceed the maximum drawdown of $2,338. With about four trades a month, this is a good starting point for a tradable system.
- Same system as in number 3, but adding a profit target of $300 per trade: Total gains increase to $6,398 while maximum drawdown is reduced to $2,063. The profit target of $300 was selected because the average run-up in winning trades using system 3 was $366. This additional rule was intended to lock in profits when they occur.
- Same system as in number 4, but adding stop losses: No stop loss makes a significant improvement in system performance. Many traders like to target a reward to risk ratio, and with a profit target of $300, a 2:1 ratio would lead to a stop loss of $150. This reduces profits by more than $2,300. Other stop levels either reduce profits or have no effect if they are large enough to allow the system to function. Some traders will always insist they need a stop loss, and any stop greater than $500 works with this system because it is unlikely to be triggered and therefore can’t detract meaningfully from the performance.
- Same system as number 4, but confirming RSI signals with a trend following indicator: Profits are improved slightly, to $6,667, and risk is unchanged. RSI is confirmed with a 14-period moving average in this case and long signals are only taken if the price is above the moving average while shorts require confirmation with a close under the moving average. The time frame of 14-periods in the moving average was used only for consistency with the RSI calculation and to avoid curve-fitting. Using more complex trend following indicators, like ADX, degrades the system performance. Complexity in this case adds a delay to the signal and that results in lower profits. That is a general rule that applies to trading – the more delay introduced by an indicator, the less likely it is to be profitable.
- Same system as in number 4, daily charts: Profit of $5,960 with a maximum drawdown of $1,633. There is an average of one trade every three months and the winning percentage is 90.9%.
There are a few conclusions to draw from these tests. First is that a simple indicator can work, but adding additional indicators won’t necessarily increase profits or reduce risk. Second is that profit targets help more than stop losses. This is an implementation of the idea that no one ever went broke taking a profit. Finally, reducing trading costs is important to profitability.
While the results of the daily strategy are promising, they really aren’t enough to allow the small trader to earn a living solely from trading even though the system delivers an average annual return of about 20%.
This is a fairly simple system, and wouldn’t take much time to implement. Results using daily data are close enough to the hourly chart that the increased workload of the shorter time frame doesn’t seem like the best use of time for a part-time trader. Trading with a fairly simple money management strategy would be a better way for the part-time trader to maximize profits.
Money management ideas can be a little boring, so I’ll try to grab your attention with the numbers – profits of $14,141 from a $250 unleveraged account with a maximum drawdown of 33.54%, using system 7 as described above with fixed fractional money management. The risk is high, this is actually about the highest percentage of drawdown a trader should accept for a tradable system, but the potential rewards seem to make this a system that’s worth looking at more closely.
There are a number of money management ideas that you can apply to a system – money management is basically the concept of varying the trade size or you can think of it as how much you bet on each position. A few of these techniques are fairly complex and require you to take on a lot of risk to capture the upside. They also require a larger account size than the fixed ratio method. For example, optimal f is a widely known money management technique that would grow $10,000 to more than $57,000 with this system, but the maximum drawdown would be more than 70% and most traders would prefer less risk. Optimal f can’t be applied with an account balance of less than $10,000 for this strategy.
With fixed fractional trading, an additional unit is traded as the account balance grows. In this example, because of the very high winning percentage and low starting balance, 100% of the account was risked. That means when the account balance grows to be greater than the maximum drawdown, two units would be traded. When the account size increases so that it could absorb twice the maximum drawdown, three units would be traded.
There are variants of this technique, such as risking less than the full account size, but they aren’t as effective with small accounts. But, this idea alone is powerful enough to more than double the back tested profits. We still can’t trade for a living with a $250 account in the forex market, but this system shows that with a good idea, controlled trading costs, and aggressive money management, you could turn that small account into enough that you could trade for a living within a few years.
Michael J. Carr, CMT