Monday, May 25, 2009

The Vicious Cycle of Expert Advisor Testing

A few days ago I read a comment by someone on a blog saying that he had nothing against expert advisors but that he never participated in discussions or reviewed expert advisors because of the way they are currently being tested and commented on. I couldn't agree more with him. The way most expert advisors are currently being tested and reviewed by the general public is not only wrong in my opinion but also problematic, misleading and fruitless. After studying several community review sites and forums I came to the conclusion that there is a vicious cycle happening which people hardly ever notice. The cycle always seems to go through the following steps :

1. Discovery - A new system is discovered and people start to hype around it. This can be a free or a commercial system. People start to do backtests and to encourage the development of the system as a group. There is a high expectation for the system and the way it performs. Generally the strategies are short term profitable, with high risk to reward ratios.

2. Early Forward Tests - Some people now start to put the experts on demo accounts to see if the backtesting they have done has anything to do with reality. Generally the experts produce good amounts of profit in small amounts of time, further encouraging the hype produced on the earlier moments.

3. Failing Period and Reoptimization - A period starts when the ea begins draw down accumulation and people start to change it, introduce new parameters, optimize current settings, etc. The hype starts to go down because people find that the system did not work across all market conditions.

4. System Trashing - By this moment all the hype has gone down and people are already in the discovery phase of a new trading system. All testing of this expert advisor is discontinued and testing is focused on new expert advisors. This ea will probably go through this cycle once more when someone rediscovers the system in a year or two and becomes exited by the initial forward and backtesting results.

The problem with this cycle ? The people here lack a good understanding of how automated trading systems actually work in the long term. People tend to generalize results that are obtained in very reduced time spaces (2 months for the longest, less than a week for the shortest) and they often try without any success to modify the ea in order to prevent draw downs on those small time periods. The problem I guess, is that most of this people are not really profitable with any automated system and they are just desperately trying to find one that works, trying to extrapolate very small trading tests towards the whole profitability of the expert advisor. Their efforts are fruitless because no meaningful trading information is gathered and they also misguide people since some of the systems they trash are actually profitable in the long run but experienced a period of draw down (which can last months for long term profitable systems).

My readers should know that this is the complete opposite of how I do my testing. All of the experts I trade have their logic rigorously examined and backtested before testing, they then have extensive periods of forward and even live testing in which I know the draw down that is expected (hence I do not run if the system enters draw down). I do not trade systems with high risk to reward ratios and I always look for systems that are consistent performers in the long run (at least 10 years). I am not interested in short term profitable systems because I am interested in long term profitable systems which often have draw downs that last for 3-5 months and I am ok with this because I know, it is just the way the market works with forex automated systems.

If you would like to learn more about the systems I have tester or I am currently testing 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: