Apr 03 2008
How Do I Back Test A Discretionary System
Mark asked me, “If I wanted to use the 4T’s system or any system like it and add Elliott Wave to it or fibonacci, then how am I supposed to back test it to determine my average profit etc.”
That’s a great question and the simple answer is you can’t. Well not technically anyway. A lot of the attraction to mechanical systems is that they are easy to back test, not only because there’s no decision making, but also because you can just plug a system into a piece of software and presto!
But there are several ways in which you can test a discretionary system.
If you are only taking your decisions from one time frame, then you can start from a pre-determined date back in history and move the chart forward one bar at a time, allowing you to make your decisions based on what you see and not what occurred in the future. Of course if you have to make decisions on more than one time frame it can become very difficult to do this. If you leave one time frame to a mechanical rule (i.e. the trend) and have your decision making (i.e. the trigger) on the lower time frame it does make it simpler to do the above.
The next solution is to paper trade for a month or even more, but when doing so make as many trades as you can. The good thing about paper trading is you’re not limited to your capital, and if you run out you can re-set your account. You want to make as many trades as you can because one of the ideas of the paper trading is to determine your metrics which are you average profit per trade, your largest use of capital per trade, the average length per trade, and the largest drawdown both in value and length. You must stay within your risk limit and position size limits when paper trading, however don’t worry if you use all of the capital up.
However I would add the following: paper trading for only a short time does not give you the benefit of various market conditions, so it is imperative you diversify the markets you are paper trading in, or you can try the following, use a mechanical system as a base for back testing.
By this I mean you are taking a mechanical system and trying to improve it by adding a decision making process to it, either with the trend, the trigger, the trade, the termination or a combination. You are doing this because you have a need to (psychological) and/or believe you can improve the systems performance.
If you take the results of back-testing the mechanical version of your system (let’s say the 4T’s system), and then apply the results of paper trading a discretionary version (i.e. you have added a decision making process to the trigger and trade section), you can compare the results. If you found that adding discretion to the system created a higher average profit per trade with all else being relatively unchanged, you’ve reached your goal.
What you must do though is compare the mechanical system’s results during the same period so as not to distort results. There is no point comparing paper traded results over the past 3 months to a 3 month period 5 years ago when the mechanical system had its greatest drawdown.
I hope I have answered your question Mark. If you or anyone else has more questions or feedback, or would like to see something presented here please let me know.
BlogMarks
Furl
Newsvine
Rojo
Email
If you like this then please subscribe to the