Automatic Forex Trading Systems: Why Do They Fail?

By Jason Cline

Someone develops a new automated forex trading system practically every week now, it seems. They all show amazing results in the tests they show but when we get into live testing the results can be very different, as most of us know from bitter experience.

So why do the hopes crumble to dust? Is it the fault of the user and their settings? Did the promoters fake the results? Or is there some little known law of physics that says that the moment a trading system is automated, the forex market will turn around to prevent it from working?

I know that last one may sound a little crazy but but sometimes I have wondered and you too maybe.

But in reality I do not believe it's because of any of those reasons. I may be hammered for this but here is what I think actually happens ...

The way a forex robot tends to come into being is this: forex experts take a system that has been bringing in profits (or invent a new one and backtest it), pay a software developer to turn it into a robot, and then to recoup the cost of the programming and make something on it too, they market it to people like you and me.

The critical question comes in the very first step. If a system has been working for the developer for a long time, great. But most times they act far too quickly. They depend to a greater or lesser extent on backtesting. They know that people will buy a new robot, so they can surely cover the cost of the automation, so there is in fact very little risk in hiring a programmer as soon as they think up a system that gives the results on backtests. They may not wait for live test results.

So they go ahead and create a new automated forex trading system. Having done that, they must market it. They might do a little live testing, but that is risky! What if it made a loss? They wouldn't lie about the results so maybe it would be better not to run it live, but release it to the market right now. People are credulous and far too many of them will buy on the basis of backtesting alone. Quick! the trader thinks, Let's get it out there now while it still seems that it works!

So what's wrong with backtesting? Nothing, if you think that future results will mirror past results. But wait, isn't that the first thing they tell you in the disclaimer on all investment documents? "Past results are not an indicator of future performance ..."

Look at a simple example. You know that the odds of winning on black at roulette are less than 50%, don't you? It's less because of the zero. I think it's around 48.5%. But distribution patterns mean that if you recorded a few hundred spins you would probably not get exactly 48.5% blacks. You might have 51% black for example.

So imagine if you did that, took those results and said, Wow, 51% black in backtests! Excellent, so now I can develop a robot that always bets on black ...

It would lose.

Of course the foreign exchange market is a little more involved than a roulette wheel, but even so I think this is basically what developers are doing if they build a forex automated robot based on backtests. And I think that is why they often fail.

I do not mean that you shouldn't use robots and expert advisors, not at all. A forex robot can be a wonderful tool.

I am just saying that we should all consider how they have been tested. Do not rush to buy the latest forex robot the minute it comes out. Wait a few weeks at least, check the forums and see how real people like you get along with new forex trading systems before you thrust your money into the developer's hot little hands.

About the Author:

Jason Cline writes articles on automated forex trading systems and the foreign exchange market for many internet sites. Click through to his opinion of the top seller FAPTurbo in his FAPTurbo review.