Thursday, January 21, 2010

The Testing Time Problem... For How Long Should I Demo Test ?

When people start to get into the world of automated trading systems they soon find out that there is a sort of "general protocol" established by most of the online user community to use forex expert advisors. People are generally told to get an EA, demo it for a certain period of time and then, when if the demo gives satisfactory results, run the expert advisor on a live account with real money. Where did this idea come from ? It came from the notion that regular people have that a minor test of a demo account will most likely mimic trading on a live account later on. Today's post will focus on explaining why this criteria to evaluate profitabilty is wrong and why most of the time it will only lead to the discarding of profitable systems and loses on forex live accounts.

What is wrong with this testing approach ? You take an expert advisor you want to test, run it on a demo account for X weeks, then you are sure it is profitable so you run it in a live account. What ? From where did people get that a profitable or unprofitable period of a few weeks or months on demo is any indicative of the expert's profitability on a live account ? I certainly do not know. A demo test of a few weeks or months is only going to tell you if the expert is profitable under a very narrow window of market conditions, these conditions are bound to change once the market starts to evolve, leaving the actual usefulness of your demo period in the trash.

What if the demo testing period only generates draw down ? Does this mean the expert is unprofitable ? No, it doesn't, many long term profitable systems go through long periods of draw down (even one or two years) before the very profitable periods which generate most equity gains. Then you are bound to discard long term profitable system when using this demo testing small periods because long term profitable systems are always most likely to be within a breakeven/draw down period than within a profitable one. Now what if you have one year of demo testing ? good enough ? No ! Again, forward testing is not the only thing you need to take into account. Some systems with terribly unsound trading techniques can generate a lot of money on periods of even one or two years before wiping everything out. This is the case for some Martingale systems, you can also be using a curve fitted system which then goes out of date and starts to wipe your account.

So is there any amount of time you can have in demo testing which will inevitably ensure you that you will not lose money on a live account ? No ! Long story short, there is no substitute for analysis and sound evaluation of trading systems. No forward testing period, no matter how long, can cover for the tester's ignorance about the system or its potential for long term profitability. How can you be sure then that a system is going to be profitable in the long term ? You cannot ! However you can ensure that there is a high probability for your system suceeding in the future if you ensure that the system uses sound trading techniques, is able to adapt to varying market conditions and performs in a satisfactory manner in long (ten year or more) simulations with consistent results in live testing.

If you would like to learn more about my criteria for the evaluation of long term proftability and how you too can start your journey in profitable automated trading with realistic profit and draw down targets please consider buying my ebook on automated trading or subscribing to my weekly newsletter to receive updates and check the live and demo accounts I am running with several expert advisors. I hope you enjoyed the article !

No comments:

Daniel Fernandez's Expert Author Email Alerts
Sign up to receive email alerts of Daniel Fernandez’s latest articles from EzineArticles.com!

Email Address: