Friday, April 10, 2015


I found this strategy using StrategyQuant Pro v3.8, and I trade it in the Principle Analysis Portfolio.  I like using larger timeframes for cross pairs like this one since I believe they are more volatile and erratic pairs, and therefore harder to identify a long term pattern on short timeframes to trade from.  The larger timeframes you use, the more smoothed out that volatility and erratic behavior becomes.  I also feel it's hard to find trending strategies in these pairs, so for this pair I looked for, and found, a range strategy that is fairly complex.

So here I found a nice 4hr timeframe strategy that captured 23,172 pips over the past 9 years.  Like all my strategies, I use .01 lots on MT4 and a $5,000 account balance to base this data on.  You can see that drawdown is minimal, only $126, or 2.5%.

I also like to make sure that a lot of trades have been executed because the more trades executed with proven success means the more likely it is that this strategy will work moving forward.  In the larger timeframes like the 4hr and daily, this can be difficult.  But here I have just over 500 trades over 9 years which works out to be about 55 trades a year, or about 1 trade a week.  For a 4hr strategy, this is perfect.  Also note the outstanding Return/Drawdown ratio of 18.38.  This means that this strategy makes $18.38 for $1 of drawdown.  This is excellent in my view.

This chart shows exactly what I'd expect when reading that initial data; a solid move from the bottom left to the top right of the chart with minimal drawdown.  The stagnation period was short and occurred at the start of the data period tested (shaded in red).  The blue line represents the optimized strategy that has a slightly better performance than the original strategy (gray line).

This is the Trade Analysis and is something I like to see, which is a lot of green and blue.  You can see that every year this strategy made money; granted, 2006 was basically flat, the following years show solid and steady gains as reflected in the previous line chart. This strategy trades well in almost all hours of the day except in what appears to be the early hours of the European session.  However, since this is a 4hr strategy it is difficult for me to filter those two hours out without eliminating the profitable hours surrounding them as well.  So I'll leave them in place.  Friday is the most traded and most profitable which tells me this strategy likes the Friday morning volatility of the US session, probably due to big data releases and positioning prior to the weekend from the EURUSD, which will effect this pair.  The fact it is a range strategy also makes sense why this day is a good day for this strategy to trade.

The main thing I look for here is to have profitable years every year.  I will tolerate one, or maybe 2 losing years, especially if the losing years were early in the sample period.  But I do not want to see that most of the overall profits were made in just a year or two as this probably just means that the strategy worked well in only a specific market condition that occurred during that period, but that it will not endure through all types of market conditions moving forward, and therefore it is not robust.  So far, this strategy looks good as it can profit over a long period of time consistently.  So let's move this strategy forward.

Now on to the tough tests, the robustness tests.  Here are the results form the Monte Carlo analysis I performed on the strategy.  These are not fantastic results, but still overall pretty good as all of the runs went from the bottom left to the top right of the chart, and there is a 95% probability that the strategy will have resulted in a gain of $1,984 if the data was altered to simulate the strategy moving forward.

However, the Net Profit still dropped over $300 from the original strategy, and the Drawdown % more than doubled.  What's most concerning here though is that the fantastic Return/Drawdown I talked about earlier declined from 18.38 to 5.45.  That is a huge drop and suggests more volatility and drawdown with less returns will occur moving forward.  Not what I wanted to see, but taken in the proper context, this is still a viable strategy.  But let's see further test the robustness of this strategy with the Walk Forward Optimization results.

Walk Forward Matrix Optimization is a very rigorous optimization and robustness test for strategies.  Because of this, it is very difficult for many strategies to pass this test.  Walk Forward Matrix Optimization is a set of Walk Forward optimizations performed with a different set of optimization periods, parameters, and out-of-sample percentages (robustness tests).  The results will tell if your strategy will benefit from periodic optimizations, and if the strategy is robust enough to sustain a solid profit in the future.

You can see from the results that this strategy performed well during Walk Forward Matrix Optimization, and actually had a small increase in Net Profit from $2,317 to $2,510.  The report shows that 25 out of 30 parameter combinations passed, the best grouping of combinations passed 9 out of 9 times, and all runs produced solid in-sample and out of sample net profit results.  Lastly, the test tells me that this strategy is best reoptimized every 239 days with a history of 717 days on 11 runs with 25% out of sample.  Basically what this means is that every 8 months I will reoptimize the strategy on about 2 years' worth of history data.

This strategy shows a steady profit with minimal drawdown over a 9 year period, the currency pair and timeframe along with the type of strategy make sense, and robustness tests show this strategy can do well moving forward.

Follow the performance of this strategy and others at Principle Analysis Portfolio.
Find and develop your own strategy: StrategyQuant Pro v3.8


No comments: