Forex: The Problem of Backtesting

Filed under: Learn Forex Trading |

In physics, the ‘observer effect’ is when the very act of measuring a phenomenon via observation leads to a variation in the results. In forex trading, when you trade, the very act of trading can also change the market conditions. Therefore, the very presence of yourself in the trading arena changes the market conditions. One variable that is not included in your backtesting is yourself. Therefore, backtesting can never be a true indicator of the market conditions that you intend to trade in.

 

You might think that the presence of one person in the vast forex market is negligible. However, if you have noticed a particular pattern that keeps repeating itself in recent markets, then other people have probably already noticed it too and will be using the same strategy. When enough people start to do the same thing, the edge will be lost and the system will no longer work.

 

Another reason for a discrepancy between back-tested results and actual results is that of curve fitting. Curve fitting should not be confused with optimisation (which helps to improve results).  Curve fitting is when system makers add too many rules, rules that have little or no theory behind them, and are there simply to improve results. The rules only affect a very small number of the most important trades, and as such, over-fitting occurs. For example, if a system had a bad drawdown, this drawdown could be smoothed out by adding a few more rules specifically designed to smooth out that drawdown. This is bad research, and should not be done. So be sure to only add rules that make some kind of logical sense. Also, a large sample should be used for backtesting, as this will make the results more reliable.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>