Last Updated on November 28, 2022 by Mark Ursell
In this article I will show the results of my backtest of a EUR/USD trading strategy. What is interesting about this test is that I used a random entry signal. I also used a trailing stop based on the ATR to close the positions.
The results showing that a random entry trading strategy can be profitable using a trailing stop.
Market and Timeframe
I am using my favourite forex pair, the EUR/USD, and trading it on the 4-hour timeframe. Each run of the analysis is over 10,000 4 hour periods. The study runs from May 2006 to October 2012.
Entry Signal
Good trading entries are generally considered to be an important part of trading. I strongly agree with this however, different types of entries work well in different types of markets. It is not possible to tell what the market is going to do, therefore the very best entries can often result in unprofitable trades.
In this test, I have used a random number generator to trigger the trade entries. The great thing about random entry is that we can test it again and again on the same historic data. Because it is random entry the results do vary quite considerably. However, by repeating the test many times, we obtain a good level of consistency.
This method of repeating an analysis is often referred to as a Monte Carlo method after the famous casino. I have previously recorded a YouTube video on using Excel to create a Monte Carlo simulator.
Exit Signal
Each trade is exited using a trailing stop. Trailing stops tend to be associated with trend following type strategies. I do not use a profit target.
The trailing stop is calculated based on the Average True Range (ATR). The ATR is a measure of the recent volatility of the market. I then multiply the ATR by a factor to calculate the trailing stop distance. This method of calculating the stop-loss is often called the Chandelier Exit and was developed and popularised by Chuck LeBeau and Dr Alexander Alder. In my analysis, I am using a 20-period ATR.
As an example: For a long trade with an entry point is 1.3500, an ATR of 45 pips and a multiplier of 2. The stop-loss will be 1.3500 – (0.0045 * 2) = 1.3410. As the trading position moves into profit the trailing stop will move with it to lock in the profit.
Simulations
For each ATR factor, I have run the analysis 100 times.
Results
For these results, I have compared three different metrics. Firstly, the average profit generated over the tests. Secondly, the Profit Factor, which is the absolute value of the total value of the winning trades divided by the total value of the losing trades. Finally, I have calculated the percentage of tests that were profitable.
Trailing Stop | Average Profit | Average Profit Factor | Percentage of Winning Tests | |
---|---|---|---|---|
1 | 51,610 | 1.06 | 67.7% | |
2 | 22,016 | 0.99 | 35.4% | |
3 | 111,394 | 1.04 | 60.6% | |
4 | 25,762 | 0.98 | 43.4% |
The results confirm that it is possible to trade a random entry trading system and be profitable. All four scenarios were profitable but there was a big difference in terms of the consistency of the results.
The results indicate that trading using an ATR factor of 3 yielded the most profit. An ATR of 1 had the most scenarios that were profitable. Interestingly an ATR factor of 2 and 4 actually had more losing scenarios than profitable scenarios but were still profitable overall.
Video
You can check out the accompanying video on YouTube to get more information and see the spreadsheet in action. Please note that I have retested the analyses from the video doing additional simulations.
Conclusions and Future Tests
The above results are very interesting to me. The results push against the ‘random walk’ theory of market behaviour and lean towards the chaotic ‘misbehaving markets’ theory.
On the basis of these results, I will not yet be including blind random entries into my own trading. However, we can be fairly clear that trailing stops have shown very good results in the EUR/USD in the past and should be considered as part of any trend following trading strategy using this pair.
In future tests I am going to expand on these tests by looking again at random entries and also see if it is possible to get more consistency using more traditional technical analysis entries.
Mark,
Do you plan on creating an ebook based on the Monte Carlo in Excel video ? Thanks
Jack B.
Hi Jack, thanks for the comment. I’m not planning on doing an eBook about that just yet. But I have recorded some new videos about backtesting random entry signals. These use the Monte Carlo method to investigate the historic profitability of various strategies. When I have finished this series of videos and articles I will look at writing an eBook about the process and results.