Before I get into doing more Expert Advisor Reviews lets get some of the basics out of the way first.
The first step is to go get MetaTrader and download the package. There are two versions of MetaTrader available, MT4 and the newer MT5. The majority of commercial systems developed in the market are for MT4 but either version will suffice.
The MetaTrader application is free to download and install. During the installation process you will have the choice to select your broker. It would appear all of the supported brokers offer demo accounts (not verified!), so it's quick to get started. Alternatively, you can go direct to a supporting broker and download the software.
At this point you should have a demo account, access to a few currency pairs and the application itself. There will likely be a software update to do before you can stuck in which is just a matter of updating and restarting.
Adding a chart is easy. You can do so from the menu items (this will offer the main currency pairs), you can right-click a price quote in the Market Watch list and add a new Chart Window, or a chart can be added directly from the File menu.
You can change the viewable timeframe, add annotations, price overlays and technical indicators, or enter a trade from a chart.
The timeframe can be adjusted from the toolbar, a right click on the chart or from the Charts dropdown menu.
To add annotations, click on one of the toolbar options; then mark a point on the chart and either drag or complete the text box. Annotations can also be added from the Insert menu
There are a number of ways you can indicators to the chart, including custom indicators. The core set of indicators are available from the toolbar, Insert menu or from the Navigator.
Custom and core indicators are available in the Navigator menu. When adding an indicator from Navigator menu it's necessary to left-click on the indicator and drag it on to the chart before releasing the click. Any custom indicators you purchase or create will be listed in a seperate folder to your primary (or default) indicator set.
You can enter trades (in this case, demo trades) either from the Market Watch window. Right click on the instrument you want to trade and select New Order (short cut F9)
or right click on the chart and select Trading - New Order.
So what about trading strategies?
Trading Strategies in MetaTrader are called Expert Advisors. These are listed in the Navigator window. Your MetaTrader program offers two sample trading strategies: MACD Sample and Moving Average.
There are a few things you can do with a strategy. You can add it to a chart (i.e. market) and trade it or you can edit it in the supporting MetaEditor application.When you Attach to a Chart the strategy will be added to the currently viewed chart.
Only one strategy can be added to a chart/market. If you wish to trade your strategy across different currencies you need to add the strategy to each chart. Likewise, if you wish to trade a strategy across different timeframes, you need to add a strategy to each timeframe chart.
When a strategy is added to a chart it will appear with a smiley face icon in the top-right of the chart. If there is an 'x' instead of a face it's because the Expert Advisors button in the menubar is switched off (red).
When the face is 'unhappy' the strategy is switched off and no trades are generated. Clicking on the face to make it 'happy' activates the strategy for that market. If you are trading a live account with real money then trades will be generated using real cash.
Focus should be limited to one chart and one strategy. Trading across multiple markets or timeframes is taxing and it's easy to overlook where signals are generated.
Initially, you will be dealing with a demo account, but once you are ready to go Live you will need to ensure the Live Trading Button is checked in the Strategy setup menu. This menu appears when a strategy is added to a chart. This menu is also available by right clicking on the chart attached to the strategy and selecting Properties.
It may also be prudent to check the Ask Manual Confirmation button to see how the strategy works in practice. Any strategy can be traded Long & Short, Long or Short.
When trading commercial strategies it's frequently necessary to click the Allow DLL Imports check box.
The real-fun with MetaTrader comes with tinkering your strategy (which will be a seperate articled). Part of this process is Back Testing a strategy. To do this we need to open the Strategy Tester, available from the View Menu and Toolbar.
This opens the Tester menu located beneath the chart.
In order to back test a strategy, first select your Expert Advisor from the dropdown menu (at the start there are just two to choose from). Start with MACD Sample. Then select Expert Properties next to it.
Expert Properties allow a top-level control of the strategy. Optimization allows another level of protection to a strategy.
For Inputs, individual values (like the Lot size) can be changed by clicking on the Value and editing it.
For the purpose of this Back Test we will leave all of these variables unchanged.
Next it's necessary to set the Symbol and Period. Both of these choices are dropdown menus.
The Model adopts Every tick and may (or may not) involve Optimization. The Modify Expert opens MetaEditor and allows you to tinker under the hood of the stategy.
The Use date defines the lookback period of the Back Test.
In order to maximise the number of ticks the Back Test should avail of the most amount of data available. To access data go to Tools - History Center
Select a Market on the left, including the timeframe, then click Add. This will upload the data to the window on the right, then click Download to bring the data into your MetaTrader application. You may need to do this for each timeframe you are interested in trading.
You may not need to this for EURUSD, but for other Markets - like indices - it may be necessary to import this data.
Next, go into Options of the Tools menu. Then go to Charts and change the Max bars in history and Max bars in chart to 9999999999999.
Once you have the data you can then run your Back Test.
The next step is to hit [Start]
This will run the strategy with your settings as selected in Expert Properties.
An Alert will sound once the strategy is complete. The various tabs show the differerent outputs of the Back Test.
The Results shows each signal, including the initial entry (buy/sell), the modification when the stop loss changes (modify), the exit by stop (s/l) or target price (t/p).
The Graph shows the Cash:Equity Return for the Strategy
The Report summarises the strategy returns, including number of trades, returns from Long positions and returns from Short positions.
The full report can also be downloaded and viewed in a browser. To do this, right click in the Report/Results/Graph screen and click Save as a Report.
With this, you now have the basics to start playing around with MetaTrader...
You can download MetaTrader Here
Cutting Through The Noise To Find The Most Profitable Trading Strategies. Forget the Hype - Read It Here First
Tuesday, August 30, 2011
Wednesday, August 24, 2011
GPS Forex Robot Review
In my next strategy review I took a look at GPS Forex Robot. The strategy is promoted through a very lengthy splash page (I wonder what the record is for this...) but this includes nearly 900 comments - all of which look to have been responded to by the developers, which I'll give them credit for at least. The splash page offers the typical hyperbole and pointless picture of creators next to their cars but here Metatrader will do the talking.
This strategy retails at $149 and offers an additional service, charged per month, which I did my best to avoid clicking. This point at which the additional monthly offer appears is after you complete payment for the strategy. You have to scroll to the end of the second lengthy splash page to complete purchase of the basic strategy system without the monthly charge. This review will only focus on the basic robot, not the monthly service offered.
GPS Forex Robot offer a 60-day money back guarantee which is good and certainly gives you time to test it out.
Installation
This was easy and the documentation was good. GPS Forex Robot offer an automated installer all of which you have to do is to select the MetaTrader directory to drop the files into. A restart of MetaTrader will see three systems labelled for EURGBP, EURUSD and USDCHF. As you can probably tell, only these three pairings are supported. So if you are looking for a dollar-yen strategy you will be out of luck.
The documentation also does a good job of explaining how to set up MetaTrader to trade the Robot.
They recommend trading on H1 and M15 timeframes (this test is on the M15 timeframe). They also suggest using a lot size of 0.01 for every $100 of your account and this is easy enough to test before you start. The document even outlines how to set up a Backtest. They suggest checking the Optimization parameter for Balance as part of the Backtest routine (this test will not use Optimization).
Expert Properties Setup
In the Inputs tab of the Expert Properties you need to add in the Email address and CBReceipt reference. You will not be able to Backtest or trade live without doing this. Note: In their documentation they use a numeric example for the CBReceipt reference but this can be a sequence of letters too. Don't delete your CB Receipt!
It's also necessary to set the "AutoGMTOffeset" to False. I received an error when I tried to run a Backtest without doing this.
For a $10,000 account, GPS Forex Robot recommend a lotsize of 1.0, but my comparative test will use the lot size of 0.1 (recommended for a $1,000 account).
Comparative Test Period
No Optimization was used in the testing and we have three, 12-month periods which were initially randomly selected beween Jan 1st 2000 and Jan 1st 2009. It just so happened the selected periods clustered in the early part of the decade:
March 31st 2000 to March 31st 2001
June 29th 2003 to June 29th 2004
May 17th 2001 to May 17th 2002
The M15 time frame was tested.
Comparative Results
In the documenation GPS Robot recommend testing from April 2007 and in repeating their settings you can get signal data.
However, using the aforementioned Comparative test periods there were a number of issues:
[1] The March 2000 - 2001 test only showed signals in March 2001 and nothing before then.
[2] June 2003 - 2004 test offered no signals. When optimization was selected, 6 signals were discarded as insignificant.
[3] May 2001 - 2002 test only showed signals in March and April 2002.
So in three years there were only ten signals. This is too small a sample size to verify performance, despite the good performance of those ten trades.
Random Test Period
Three additional 12-month periods were selected for testing. Again, EURUSD was the pairing and the periods were selected between January 1st 2000 and January 1st 2009. Tests were done on 15-minute data. The three periods were:
June 3rd 2005 to June 3rd 2006
May 18th 2004 to May 18th 2005
October 5th 2001 to October 5th 2002
Random Test Results
The same issues which were raised in the Comparative test emerged in the Random test too.
[1] The June 2005 - 2006 period offered no signals, but optimization suggested 6 signals which were ignored.
[2] May 2004 - 2005 no signals. Again, 6 overlooked signals on optimization.
[3] October 2001 - 2002 offered 4 signals, but the randomized period results were similar to the Comparative test May 2001 - 2002 results.
Conclusion
How good is the system? Without a more robust data set no real conclusion can be drawn. There was nothing in the Journal outputs from the tests to suggest a problem in the data (and this is the same data used in other Metatrader EA comparisons). 327 trades were generated for back test run on GPS Forex's recommended time frame of April 2007 to the current day. Of those 327 trades, 94% were profitable with an average return of $88.8 per trade (effectively 8.8% per winning trade). So in that respect the back test holds true. However, in the same backtest, the strategy exhaled all of its gain on the last day of the test??? Not surprisingly, this plunge is not visible on any of their promotional material. The bizarre equity curve (line) of the backtest from start to finish actually looks like this:
There are two important issues which aren't hidden in GPS's promotional material. The first is the large Drawdown for the strategy. The second is the high loss per losing trade.
In their promotional example the average Return Per Profitable Trade of their USDCHF system is $147.21, but the Return Per Losing Trade in the strategy is over ten times greater, at $1,564 (and in their 4 year test period there were 25 losing trades and 565 profitable trades). In one case, there were two consecutive losing trades for a combined loss of $3,046. So this is not a system for the feint of heart, despite the high win percentage of trades.
Obviously, GPS Forex Robot don't draw attention to the system's losses or drawdown risk, but at the same time, they don't hide it either. I don't know what the equity curve I show is different to theirs, or whether the returns hold true if traded live. As stated at the start, GPS Forex Robot offers a 60-day money back guarantee. So it may be worth a trial and see how you get on. The one piece of advice I would give because of the large drawdown risk and loss per trade is to use a smaller lot size. So while they may recommend 0.01 lot per $100 I would be more inclined to start with a 0.01 lot per $1,000 capital.
Our Current Strategy Leader is:
This strategy retails at $149 and offers an additional service, charged per month, which I did my best to avoid clicking. This point at which the additional monthly offer appears is after you complete payment for the strategy. You have to scroll to the end of the second lengthy splash page to complete purchase of the basic strategy system without the monthly charge. This review will only focus on the basic robot, not the monthly service offered.
GPS Forex Robot offer a 60-day money back guarantee which is good and certainly gives you time to test it out.
Installation
This was easy and the documentation was good. GPS Forex Robot offer an automated installer all of which you have to do is to select the MetaTrader directory to drop the files into. A restart of MetaTrader will see three systems labelled for EURGBP, EURUSD and USDCHF. As you can probably tell, only these three pairings are supported. So if you are looking for a dollar-yen strategy you will be out of luck.
The documentation also does a good job of explaining how to set up MetaTrader to trade the Robot.
They recommend trading on H1 and M15 timeframes (this test is on the M15 timeframe). They also suggest using a lot size of 0.01 for every $100 of your account and this is easy enough to test before you start. The document even outlines how to set up a Backtest. They suggest checking the Optimization parameter for Balance as part of the Backtest routine (this test will not use Optimization).
Expert Properties Setup
In the Inputs tab of the Expert Properties you need to add in the Email address and CBReceipt reference. You will not be able to Backtest or trade live without doing this. Note: In their documentation they use a numeric example for the CBReceipt reference but this can be a sequence of letters too. Don't delete your CB Receipt!
It's also necessary to set the "AutoGMTOffeset" to False. I received an error when I tried to run a Backtest without doing this.
For a $10,000 account, GPS Forex Robot recommend a lotsize of 1.0, but my comparative test will use the lot size of 0.1 (recommended for a $1,000 account).
Comparative Test Period
No Optimization was used in the testing and we have three, 12-month periods which were initially randomly selected beween Jan 1st 2000 and Jan 1st 2009. It just so happened the selected periods clustered in the early part of the decade:
March 31st 2000 to March 31st 2001
June 29th 2003 to June 29th 2004
May 17th 2001 to May 17th 2002
The M15 time frame was tested.
Comparative Results
In the documenation GPS Robot recommend testing from April 2007 and in repeating their settings you can get signal data.
However, using the aforementioned Comparative test periods there were a number of issues:
[1] The March 2000 - 2001 test only showed signals in March 2001 and nothing before then.
[2] June 2003 - 2004 test offered no signals. When optimization was selected, 6 signals were discarded as insignificant.
[3] May 2001 - 2002 test only showed signals in March and April 2002.
So in three years there were only ten signals. This is too small a sample size to verify performance, despite the good performance of those ten trades.
Random Test Period
Three additional 12-month periods were selected for testing. Again, EURUSD was the pairing and the periods were selected between January 1st 2000 and January 1st 2009. Tests were done on 15-minute data. The three periods were:
June 3rd 2005 to June 3rd 2006
May 18th 2004 to May 18th 2005
October 5th 2001 to October 5th 2002
Random Test Results
The same issues which were raised in the Comparative test emerged in the Random test too.
[1] The June 2005 - 2006 period offered no signals, but optimization suggested 6 signals which were ignored.
[2] May 2004 - 2005 no signals. Again, 6 overlooked signals on optimization.
[3] October 2001 - 2002 offered 4 signals, but the randomized period results were similar to the Comparative test May 2001 - 2002 results.
Conclusion
How good is the system? Without a more robust data set no real conclusion can be drawn. There was nothing in the Journal outputs from the tests to suggest a problem in the data (and this is the same data used in other Metatrader EA comparisons). 327 trades were generated for back test run on GPS Forex's recommended time frame of April 2007 to the current day. Of those 327 trades, 94% were profitable with an average return of $88.8 per trade (effectively 8.8% per winning trade). So in that respect the back test holds true. However, in the same backtest, the strategy exhaled all of its gain on the last day of the test??? Not surprisingly, this plunge is not visible on any of their promotional material. The bizarre equity curve (line) of the backtest from start to finish actually looks like this:
There are two important issues which aren't hidden in GPS's promotional material. The first is the large Drawdown for the strategy. The second is the high loss per losing trade.
In their promotional example the average Return Per Profitable Trade of their USDCHF system is $147.21, but the Return Per Losing Trade in the strategy is over ten times greater, at $1,564 (and in their 4 year test period there were 25 losing trades and 565 profitable trades). In one case, there were two consecutive losing trades for a combined loss of $3,046. So this is not a system for the feint of heart, despite the high win percentage of trades.
Obviously, GPS Forex Robot don't draw attention to the system's losses or drawdown risk, but at the same time, they don't hide it either. I don't know what the equity curve I show is different to theirs, or whether the returns hold true if traded live. As stated at the start, GPS Forex Robot offers a 60-day money back guarantee. So it may be worth a trial and see how you get on. The one piece of advice I would give because of the large drawdown risk and loss per trade is to use a smaller lot size. So while they may recommend 0.01 lot per $100 I would be more inclined to start with a 0.01 lot per $1,000 capital.
Our Current Strategy Leader is:

Monday, August 15, 2011
Forex Breakout System Review
I'm going to take a look at some of the (many) automated strategies currently available for MetaTrader on the market. Seperating the wheat from the chaff is a time consuming process and I'll start with some of the cheaper products out there.
First on my list is a Forex Breakout System. The system has the usual corny splash page promotion, but is it any good?
The strategy retails for $125 and offers the core Expert Advisor strategy in addition to Candlestick and Pivot indicators to add to your charts, but the real interest is the strategy. Documentation is provided to help you set up and get started.
Expert Properties Setup
For the purpose of the test I will use a $10,000 starting account trading 0.1 Lots both Long & Short. In Expert Advisor Options, the boxes to 'Allow DLL Imports' and 'External Experts Imports' were checked and if Live trading was to be supported these should be checked too.
There was no Optimization employed.
This particular strategy trades in three different time zones so in the Expert Properties / Inputs it was necessary to set the Lot size for each time block (the image shows the Asia_FixedLots configuration); both 'Value' and 'Start' were set to 0.1 for Asia, London and New York.
Other Expert Properties Inputs were left unchanged and unticked. Further adjustments may improve the performance.
Comparative Test Period
Three 12-month periods of 15-minute data on the EURUSD were initially selected randomly from a 10 year period starting January 1st 2000. These dates will form the basis for comparative tests across different MT systems. The periods tested were
March 31st 2000 to March 31st 2001
June 29th 2003 to June 29th 2004
May 17th 2001 to May 17th 2002
Comparative Results
Despite a very high win percentage the strategy failed to perform at deployed settings. The strategy frequently took small profits and in many cases broke even, but individual losses were large.
Comparative Graphs
The FBS strategy nickle-and-dimed the test account lower.
Random Test Period
A second set of tests were conducted a fresh set of data. I had hoped to test on a different currency pairing but the Model Quaility of results wasn't high enough to qualify. The start dates were randomly selected from the period of January 1st 2000 to January 1st 2009 and covered 12 months. The test was conducted on 15-minute data. The test periods were:
April 1st 2009 to April 1st 2010
November 12th 2000 to November 12 2001
March 25th 2002 to March 25th 2003
Random Test Results
Only one of the three random test periods didn't show a loss, although if slippage was a factor it would likely have registered as a loss. The remainining two random sample dates were close to the Comparative test period, so the results were similar (and therefore counted as losses).
Random Test Graphs
The April 2009 - 2010 test period did manage to work a small profit before giving it all back at the end.
Conclusion
On their website they give two examples of their system. One which shows a win percentage of 74% in the EURUSD trading in March 2010. The other with an impressive 96% winners trading the GBPUSD in December 2007. During March 2010 the EURUSD was effectively range bound, while in December 2007 GBPUSD was in a strong downtrend. So while they may have cherry picked their promotion sample period, returned results weren't necessarily attributed to a given trend (or lack thereof).
The win percentage expectation between their sample set and the results presented here are comparable, as was the likely loss per trade - which was quite heavy in their sample period relative to their profits. But where they showed strong profits per trade, analysis here showed little of that, despite all the winning trades. If there is an issue with the strategy 'out of the box' then better documentation may help improve this return.
At the core there may be a useful strategy to deploy, but users adopting it will need to get under the hood to work the profit side of the returns.
Our Current Strategy Leader is:
First on my list is a Forex Breakout System. The system has the usual corny splash page promotion, but is it any good?
The strategy retails for $125 and offers the core Expert Advisor strategy in addition to Candlestick and Pivot indicators to add to your charts, but the real interest is the strategy. Documentation is provided to help you set up and get started.
Expert Properties Setup
For the purpose of the test I will use a $10,000 starting account trading 0.1 Lots both Long & Short. In Expert Advisor Options, the boxes to 'Allow DLL Imports' and 'External Experts Imports' were checked and if Live trading was to be supported these should be checked too.
There was no Optimization employed.
This particular strategy trades in three different time zones so in the Expert Properties / Inputs it was necessary to set the Lot size for each time block (the image shows the Asia_FixedLots configuration); both 'Value' and 'Start' were set to 0.1 for Asia, London and New York.
Other Expert Properties Inputs were left unchanged and unticked. Further adjustments may improve the performance.
Comparative Test Period
Three 12-month periods of 15-minute data on the EURUSD were initially selected randomly from a 10 year period starting January 1st 2000. These dates will form the basis for comparative tests across different MT systems. The periods tested were
March 31st 2000 to March 31st 2001
June 29th 2003 to June 29th 2004
May 17th 2001 to May 17th 2002
Comparative Results
Despite a very high win percentage the strategy failed to perform at deployed settings. The strategy frequently took small profits and in many cases broke even, but individual losses were large.
Comparative Graphs
The FBS strategy nickle-and-dimed the test account lower.
Random Test Period
A second set of tests were conducted a fresh set of data. I had hoped to test on a different currency pairing but the Model Quaility of results wasn't high enough to qualify. The start dates were randomly selected from the period of January 1st 2000 to January 1st 2009 and covered 12 months. The test was conducted on 15-minute data. The test periods were:
April 1st 2009 to April 1st 2010
November 12th 2000 to November 12 2001
March 25th 2002 to March 25th 2003
Random Test Results
Only one of the three random test periods didn't show a loss, although if slippage was a factor it would likely have registered as a loss. The remainining two random sample dates were close to the Comparative test period, so the results were similar (and therefore counted as losses).
Random Test Graphs
The April 2009 - 2010 test period did manage to work a small profit before giving it all back at the end.
Conclusion
On their website they give two examples of their system. One which shows a win percentage of 74% in the EURUSD trading in March 2010. The other with an impressive 96% winners trading the GBPUSD in December 2007. During March 2010 the EURUSD was effectively range bound, while in December 2007 GBPUSD was in a strong downtrend. So while they may have cherry picked their promotion sample period, returned results weren't necessarily attributed to a given trend (or lack thereof).
The win percentage expectation between their sample set and the results presented here are comparable, as was the likely loss per trade - which was quite heavy in their sample period relative to their profits. But where they showed strong profits per trade, analysis here showed little of that, despite all the winning trades. If there is an issue with the strategy 'out of the box' then better documentation may help improve this return.
At the core there may be a useful strategy to deploy, but users adopting it will need to get under the hood to work the profit side of the returns.
Our Current Strategy Leader is:

Subscribe to:
Posts (Atom)