Building a Currency Trading System

About two weeks ago I got an email from a trader explaining his trading method and asking me for some advice on how to improve on it. On the surface of it, there wasn’t much to his approach but he had been trading profitably for a few years and that got me interested.

As it turned out, he was using the same charting package I use and had used the back test facility to go over years of data to come up with the method he now used. The method itself was a simple, sound little technique but his money management control was brilliant and that was the genius of his method.

So far there was nothing unusual about this other than he was using the back testing facility. If you have only started trading in the last few years this might not seem unusual to build a little code and go back and test lots of ideas. For me it is different. I started when most charting packages never had this facility and I did everything the hard way. I would sit with a notepad, come up with an idea and literally go over every bar on a chart for the last 10 years. This could take weeks.

Because that’s how I started I never really got out of the habit of doing it any other way. I have lost count of how many of these notepads I have dotted around my office. I have always continued doing it this way until a few weeks ago.

Anyway, He agreed to help me learn a little code in exchange for some help with his approach to the market. Even then I wasn’t particularly enthusiastic about the idea as I have become pretty set in my ways and comfortable with my trading style.

This was all done via email, but luckily he was very good at explaining things and after a few more emails I was beginning to get the hang of it.

Welcome to the 21st century — following his instructions to the letter, I opened up the box he told me to on the charting software and entered the code then pressed the button — BANG. In about 30 seconds flat this thing had worked out the profit and loss of 10 years of data. I sat looking at this for over 10 minutes thinking of the possibilities.

In the past, every time I had a new idea or combination I wanted to test I would write it down for testing later when I got time. With this new power I could do all this in minutes.

After that I idly started messing around with the code and couldn’t get it to work. When I put in my own formula and hit enter nothing happened or I got an error message. I knew immediately this would take some studying.

Luckily I am a fast learner and within an hour or two I had mastered a few basic formulas. I got so involved in this that I missed lunch, dinner and before I realized it, I had been at it for 20 solid hours.

The following days where no better. I stopped watching the markets and even phone calls and emails where beginning to bug me. All I wanted to do was test. This thing was becoming addictive.

The thing that got me hooked was actually a mistake. The first working formula I used gave a result of a 93% accuracy making astronomical amounts of money. It wasn’t until later in the day that I realized that I had made a mistake in the formal and I was calculating the wrong thing.

If you think back to when you started trading, you probably used one of the standard indicators in the standard way. After all, it was your only reference point.

My experience has taught me that most of the indicators don’t work well in the way that they where designed. Now, this is where I can be of some help.

I don’t claim to be a guru or a genius at anything but I do often have an unusual approach to things and here is an example.

Let’s take a fairly typical simple moving average of 50 periods and test it. If there is a close of the daily bar above 50 MA then we will go long and if there is a close below the 50 MA then we will go short. This is a SAR trade (stop and reverse), which means the long position will stay open until we get a short signal at which time we will reverse and go short. This is what you will come up with if you test this idea on the daily bars of the spot Pound/Dollar currency pair (GBP/USD).

Now as soon as you see these numbers it will make most trades walk away. There were a total of 196 trades and only 19.4% of them where profitable and to cap it all it was a loss overall. But wait, there is more here then meets they eye and this is the point I want to make. Be different. If you do the same as everyone else you will get the same results. You only need one little thing that works regardless of how weird it is to make serious money.

What would happen if we had done exactly the opposite of what everyone else would have done. What would happen if we went short when the close was above the average and long when it was below the average?

Again you would have made 196 trades but you would have been correct an incredible 80.6% of the time. The largest loss of -2208 wouldn’t have fazed me, as I know that once I start adding stops and limits it would take care of it. I also know that if something is profitable in its basic form it will generally become more profitable once good money management has been added.

Sounds impressive. However I wouldn’t trade something like this because for every trade you would pay a 5 pip spread, which means it would bring your profit down to 2770 over 10 years to give you an average of 277 pips per year or 14 pips per trade. That’s just not enough juice for the effort. Don’t try and trade this, I am just making a point with the example.

Over the two-week period I tested hundreds of combinations and came up with at least three viable methods. If an old fart like me can do it, you can do it to.

The point is that all the best results of my testing where when I tried weird stuff. Doing the opposite of what an indictor said I should do or using them in an entirely different way.

If your charting package doesn’t have the ability to do back testing or you are on a limited budget don’t worry. Just get out your notepad and start writing.

Start with an idea. Go back through as much data as you can get your hands on. A hundred trades are not enough. You really want to test thousands of trades if you can. You also want it to work on more than the market you normally trade. If you are trading stocks, test it on all the stocks you trade not just one. If it is a good system, it should work on anything.

Just because your method might be discretionary (you decided when to enter or exit a trade) doesn’t mean you shouldn’t back test it. Go through all the charts you can get your hands on and go through each scenario and make a note of whether you would have made a profit or a loss on each trade.

Some methods can’t be back tested because there are too many variables. Especially if each trade set up is different form the next. But test the parts you can.

New traders are obsessed with the entry point – don’t be. The object of trading is to get an advantage. There is nothing wrong with having a system that only has a 30% win rate if overall you have a profit.

There are two main approaches to building a system. You either have to have an advantage of more winners than losers or your winners have to greatly outweigh your losers. If you can invent something that has the advantage and your winners greatly out number your losers then you win first prize.

The common thread that ran through all the methods and systems I have tested is this — money management is easily the most important part of any approach. How you control your loses is probably the most important thing you can do in trading. Even a badly designed method or system can look good if it has good money management. On the other hand, there is no system that will survive bad money management.

Last thing — try and keep things simple. I know there are bright sparks out there that test banks of information with various types of sampling but your approach must be robust and robust approaches tend to be simple.

Martin Chandra

http://www.articlesbase.com/finance-articles/building-a-forex-trading-system-87766.html