Monday, November 2, 2009

Trading on the Edge of the Charts...Why Visual Backtesting Does Not Work

It seems that the first thing that anyone does when they are starting to examine a potential trading system is to go down the charts and analyze the potential trades that would have happened from time to time. That is, people do what we call a "visual backtest" which is merely the optical evaluation of trading posibilities over a chart over a given (usually small) period of time. Visual backtesting brings many problems to the evaluation and creation of automated and manual trading system, often because profitability is over estimated by the person who does the backtesting who is already "biased" in a certain sense towards the profitability of the systems he or she develops.

What are the problems of visual backtesting ? Well, for one, most people seem to underestimate the problems related to how many indicators are portrayed, often traders will analyze systems with indicators that repaint (change past values due to future values) and because of that they get a greater profitability, there is also the fact that people tend to underestimate the exits when they already have access to how the whole picture developed. They say, I would get out here because "this" happened but "this" is usually a particular criteria applied to a single trade, a criteria that gets somehow twisted and adapted as the trader does the backtesting. Visual backtesters tend to hate taking loses and they usually modify the rules of the system "on the fly" to avoid the taking of loses. This of couse, means that in the end the trader has a system that may have worked very well for a small set of particular trades but will fail to work in the future or further away in the past.

What is the correct way to test a strategy then ? The best way to test a strategy is to simulate what real trading would have been like. The easiest way to do this is to use the metatrader 4 strategy tester as a "real life" market experience. You just need to edit any expert you have and place a // before the OrderSend function on any lines that start with it, this will make the expert unable to open trades. After this, just load the expert into any chart and do a visual backtest of the period you want to analyze. This will give you an "evolving" view of the market and you will see it how you would see the market in real life. With the advantage that you can "fast forward" and see where your trades would have ended. This allows you to apply any indicator/price criteria and see whether it would or it would have not generated you profit in the long term.

When I am teaching someone how to trade this is one of the first things I do, we will open up a strategy tester and we will "simulate" random entries on the visual backtesting and see where we would have goten out of trades (by simulate I mean that we will just draw lines on the screen and keep the record on an excel sheet since no actual trades can be entered while on a backtest). This teaches in a very practical way how to handle market exits and how the market is viewed "on the edge of the chart". Viewing the market before "the edge" is something that will never happen in real life so why would you ever practice trading a system on charts that have already been drawn ? If you think about it, it does not make any sense because the whole perspective changes dramatically. You will see that after doing this exercise you will get a very good perspective of how you interpret and react to the market. Generally I have several metatrader instances so that me and my students can analyze both short and long term trading perspectives.

If you would like to learn about automated trading systems and how they can be traded profitably in the forex market, specially how you too can trade or program a free, long term profitable expert advisor, 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!

Email Address: