The Problem With Trading Robots

My adventures with trading robots started many years ago, at time when few brokers of any kind would even consider allowing someone to connect or trade electronically with their system.

It was only through well-connected, tech-savvy friends, that I was able to do so at all. My custom software was written in the then quite new C++ programming language and every time the broker made changes to their system, which was often, my system would inevitably fail. I often spent long evenings after work figuring out how to get it going again.

Over the years the technology and facilities for automated trading slowly evolved.

I was an early adapter of the Metatrader Forex trading platform with its built in interface and programming language. At the time it seemed like a dream come true and it didn’t take me long to put together my first automated Forex trading program. Once I got it working I tested and adjusted it until it all seemed to be working properly before letting it lose on a live account. After watching it open a position that slowly increased in unrealised profit, I decided to go to bed and let it run for the night.

The next morning I was horrified to see that the robot had taken no wins and 10 consecutive losses at a cost of $200 each which amounted to a net loss of $2000 for a night of automated trading.

Just one of the many hard lessons I learned over the years about automated trading. Sigh…

It is not uncommon to find frustrated traders, particularly in the Forex market, turning to trading robots in the hope of making money. This is particularly the case in Forex markets which are probably one of the most challenging markets that you will ever come across.

These markets can toss you around emotionally like a cork on the ocean. It is not surprising that the promise of a device that can execute trades free from emotional and behavioural bias would have significant appeal.

While the trading robot itself is obviously free of emotions and irrational behaviour, its human operator remains unchanged. It is all too easy for a human to interfere with the robot either by turning it off completely or attempting to alter its operating parameters in the belief that something is not working properly.

Fig 1 – Equity chart of a live trading strategy

The above equity chart shows live trading results of a semi-automated strategy that I was recently invited to look at. I have no reason to believe that this is anything other than a genuine strategy, but it has a problem. The multiple periods of significant drawdown would scare the bejeebus out of many traders, especially if a significant amount of capital was involved.

I have marked up this chart to show the points at which a somewhat risk averse human operator might have reasonably intervened. If someone had actually followed this pattern they would have turned a modestly profitable strategy into a losing one.

Most of the robots that retail traders find themselves using are pretty basic. They inevitably consist of a handful of indicators used to trigger entries and exits together with some means for the user to set or adjust their level of risk by changing position size. They may also include other features which are usually little more than marketing gimmicks.

The vast majority of commercial robots sold to retail traders are known as “black boxes” which means that the way they work is hidden from the user. The marketers of these devices reveal only what they want you to know, claiming of course that they need to protect their valuable trade secrets. In reality, the vast majority of these products don’t have any secrets worth protecting.

While the marketers of these products are highly skilled at convincing people that they can earn fabulous fortunes by purchasing their product, the reality is that most of them perform very poorly. All the claims of amazing profits are little more than smoke and mirror tricks that I will endeavour to explain.

Price movement in all financial markets is predominantly random. A sequence of random events results in what is known as a random walk, which to the casual observer may at times look very much like a trend. A trading robot that of itself has very little ability or “edge” can be carried along by this mostly random movement and is sometimes able to make significant profits, at least for a while. It is simply a matter of having a run of good luck.

                    Fig 2 – A random walk

Critical to the outcome of any random behaviour is the initial starting conditions, so identical robots, started at different times in slightly different markets, can produce radically different results.

Another factor that affects a robots performance is risk. As you increase position size, more money is risked and the potential profit increases accordingly. The Forex markets provide access to extremely high levels of risk through highly leveraged margin trading. Many commercial trading robots use this risk capacity to an absurd degree. With ultra-high risk it is entirely possible to do as the robot vendors claim and double your investment in a month. What they don’t mention is that it is also possible to lose your entire investment in the same period of time or less.

You could start a robot trading on a $100 Forex micro account and just regard the whole exercise as an expensive lottery ticket. I see no problem with anyone who wishes to take that approach providing they realise that they are gambling, not trading, and are treating the financial markets like a giant casino.

Given enough time in the market, all these types of robots will lose money. They have an overall negative expectancy due to the broker’s and market maker’s spread or commission. Like the casino operators and financiers it is these people who ultimately benefit from the gambler’s losses.

The others who profit are the people who sell these robots. Many of these offerings are expensive for what you get, but it is easy to see how gullible people get sucked in by the claims of enormous profits. What you are really buying is hyped up hope.

If you want to play around with robots and verify some of these claims for yourself you can easily do so for nothing. There are quite a few places on the internet where you can freely download a trading robot for the Metatrader Forex platform. With nothing more than modest programming and technical skills it is possible to get one of these robots working without too much trouble. You can then set it up and run it on a free demo account with a broker of your choice.

One place to start looking is Metatrader’s own community website where you can find some quite informative articles as well as robot code that can be freely downloaded. The end result could well be just as good as and maybe even better than the commercial robot that you pay $1000 or more for.

Apart from the price, another advantage is that these free robots are not “black box” affairs, but open code that you can study to learn how they work. This knowledge can only help you in your trading journey.

You are probably wondering if there are any really “good” trading robots available that do consistently trade profitably. Yes there are. Commercial versions are typically marketed to large financial trading institutions and sell for hundreds of thousands of dollars, but that is just the beginning. These organisations then employ teams of people with PhD’s in mathematics and science to work with these software products, continuously developing, adjusting and tuning them so that they align with and profit from current market conditions.

You may also be wondering if it is possible for an ordinary retail trader to access technology at this level. The answer again is yes. However, it is difficult, but not impossible to do so.

Trading robots provide a fascinating and challenging opportunity that is ripe for creative exploitation. I intend to continue with this subject in my next article. In the meantime please feel free to leave any comments, thoughts or questions that you have. It is helpful to know where peoples interests lie.

2 thoughts

Leave a Reply

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