Saturday, July 24, 2010

Trading a Martingale Based on Backtests : A VERY Dangerous Road

During the past few months I have sadly seen a great increase in the number of Martingale systems flowing around on the internet and their use by new and inexperienced traders. I have also seen some people testing and using Martingale systems with great hope and some with what seem to be some very good results. Many of these people who use Martingales or systems with progressive money management seem to rely on backtesting results to test their theories and sometimes they claim that a 10 year backtest in which no wipeouts happen is "good enough" to consider a Martingale worth running on a live account since it is inherently "safe". Within the next few paragraphs you will see why this way of approaching Martingale trading is terribly dangerous and why people approaching trading in this manner are bound - sooner or later - to wipe out accounts and face the truth about progressive money management, in the end it never works.

The argument here seems to be pretty simple. You know that Martingale systems are dangerous but if you can get a 10 year backtests that shows no wipeouts it means that you are safe, right ? If during such a long trading period and such a varied array of market conditions your system survives then everything should be Okay. The truth is that there a few VERY large pitfalls to this approach.

The first problem with the backtesting of Martingale systems is that there is an inherent error in every backtesting result which can be explained in technical aspects of both the backtesting mechanism and the market itself. A 10 year backtest only gives you an approximation of the results during the past 10 years because - in reality - things like spread widening, requotes, broker differences and the absence of one minute interpolation would have made the results different to some extent.

In the case of Martingale and progressive money management systems the problem is that a very small error in the backtests can make the whole difference between wipeout and survival. Imagine that you have a Martingale system that wipes an account with 7 consecutive loses and the account achieves a maximum of 5 in a 10 year backtest. Now if the system only had 2 additional consecutive loses on any of those losing periods the account would have been wiped. While an addition of 2 consecutive loses to a given trading period for a trading system designed on sound principles is minimal, the effect on a Martingale or progressive money management system is bound to be devastating.

In the end the limitations of simulations make the "certainty of safeness" of any Martingale system a lie since the actual errors and limitations of the simulations are not only important but actually most likely determinant towards the evaluation of systems that are so sensitive to small increases in the number of consecutive loses. In reality, all trading systems are bound to be facing anything between one to three times the number of consecutive loses they have given in historical testing and this makes progressive money management systems always reach wipeouts in real trading.

To sum it up, using Martingales based on backtesting is a VERY dangerous thing to do due to the limitations of the simulations. In the end, ALL martingales are bound to wipe their accounts in the long term. This is a statistical certainty which does not change, no matter what short term results show or what simulations may appear to be telling you. It comes back to the old saying, there are bold and old traders but there are no old bold traders.

If you would like to learn more about the development of trading systems and how you too can design and trade your own systems based on sound money management techniques please consider buying my ebook on automated trading or joining Asirikuy to receive all ebook purchase benefits, weekly updates, check the live accounts I am running with several expert advisors and get in the road towards long term success in the forex market using automated trading systems. 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!

Email Address: