Monthly Archives: September 2010

Why back testing is misleading: History doesn’t repeat in forex



Forex is a risky business, and one of the major preoccupations of forex traders is the reduction of uncertainty through every possible means. In the case of live trading, the tools and options that are available for this purpose are limited. In one way or the other, any conjecture about the future must involve a degree of risk, and with all the unknowns and variables that go into the formulation of a strategy, the extent to which risk can be reduced in real trading is not very large. This fact has made many traders willing to consider back testing on historical data as an alternative to testing in live trading. The presumption is that the market action repeats itself, that a method that was successful consistently over a period stretching back decades into the past, in some cases, is likely to be successful in the future as well. But how true is this presumption? We will take a look at this subject in this article.

1. Forex price action is chaotic

Let’s first recall that forex price action is a chaotic process. Other chaotic processes include the Brownian motion of molecules, weather events, and earthquakes. The common property of all these events is that their response to different initial conditions is vastly divergent. A small difference in the input, a seemingly insignificant alteration of the data fed to these systems results in a vastly different outcome. What this means from our point of view as forex traders is that each days dynamics, and rules governing the price movements is difference from the others, and as such, success at one point in the past implies nothing whatsoever about success in the future.

2. Back testing a strategy ignores crucial broker-specific factors.

There are other factors that make back testing an unrealistic tool. Broker-specific issues are some of them. What would happen if the software crashes? Would back testing reflect those cases where the broker misquotes a price, or refuses to execute a trade? What about those situations where the broker experiences very brief but severe liquidity problems, which widen the spreads rapidly while triggering stops? Such short-term, temporary issues are commonplace, but are not always discounted in the back testing process since they are not observed by all traders. (You may trade with firm A, while back testing your strategy with the software of firm B, which would lead to misleading results.)

3. Back testing is a marketing tool

All this being the case, why is back testing so popular? Since back testing makes many promises and claims possible for strategy developers, and peddlers of automated systems, it infuses an air of credibility and reliability to a trading method being advertised. Thus, it is effective as a marketing tool, and in order to promote their own products, the sellers of online software choose to promote back testing as well, as a certification mechanism for a valid and profitable strategy.

4. Back testing optimization is dangerous and futile

One of the worst approaches that can be taken with back testing involves the notorious back testing optimization method where the strategy is tweaked in such a way that it generates optimal results when it is applied to past input. Yet we have already noted that past patterns are highly unlikely to be repeated in the future. They were the result of particular combination of factors that is extremely unlikely to be repeated for the foreseeable future. If this is the case, back testing optimization ensures that we’ll blow up our account sooner or later, especially if the results of back testing do indeed work to inspire some false confidence in our trading decisions.

5. Back testing is only useful for educational, and entertainment purposes

What is the best use of back testing, then? Back testing can be excellent tool for the forex novice who doesn’t know much about indicators and patterns, and is even more ignorant of how a strategy can be constructed. In that case, back testing can be used to create, and evaluate strategies not with respect to future performance, but to show whether the creator is fully in command of the tools that he employs, whether he understands how they were created, and what purpose they serve.

Perhaps the most popular back testing platform is the MetaTrader software popular with both online forex brokers,and trader. The merits of this platform are in fact numerous, but its back testing capabilities are not one of them. While making use of the platform, we advise that you focus on its analytical capabilities while minimizing the historical aspect of testing. Trading is risk taking; there’s no way around this basic fact.