Algorithmic trading is a trading method that uses advanced mathematical models to make instant decisions on the time and conditions of transactions in financial markets.
How to create an algorithm for a trading robot?
At the first stage, the trader creates his/her own mechanical trading strategy. The strategy is tested using historical data to understand the profitability level of the strategy, as it is exactly the historical data that contains all the conclusions and reactions of a huge number of market participants, and those conclusions and reactions characterize exactly the moment when traders and computers were betting.
High-frequency algorithmic trading is mostly executed at the expense of the Direct market access and at the same colocation centre, where the exchange trading system is located. The most common solution is the FIX/FAST protocol. FIX-protocol is supported by the majority of electronic platforms of the world. To obtain the minimum delay, high-level languages are used, such C++, C#, Java, etc. Due to this approach, the time for passing the order is minimized and the efficiency of algorithmic trading is increased.
Of course, creating a long-term profitable trading algorithmic system is not easy. Investment funds spend millions of dollars a year to develop such systems. This requires a lot of effort and time, understanding and knowledge, endless searches for new algorithms and new ways to improve the old ones.
So what is better to implement using trading robots, and what is better to leave to a human being?
Computer will be trusted the following tasks.
High-frequency trade. A human being is physically incapable of performing several operations per second, doing calculations at the same time.
Scalping. People can certainly scalp, but fatigue, for example, for sure will come. The person gets tired, attention fails, and emotions get accumulated. The algorithmic robot will easily scalp 24 hours a day for 30 pairs.
System technical analysis. A human being cannot be compared with a computer in the question of speed and possibilities of searching different patterns and market inefficiencies. The automatic trading system has so many advantages that justify the risks of using:
- The trader practically does not interfere with the trading process, which saves time;
- A market player can combine both manual and automatic trading;
- The trading adviser constantly works according to the incorporated Forex strategy;
- The trader, not being in front of the monitor, minimizes the emotional load;
- Minimal risks (use of conservative trading advisers).
Automatic trading at the exchange also has its drawbacks, some of which are significant:
- Trading robot cannot take into consideration the actual entire market situation.
- Trading robot is unable to use, for example, a fundamental analysis.
- The fact that the algorithmic robot trades according to the embedded algorithm, has the opposite side: it does not see the whole market situation; it cannot estimate the political and economic events.
The work of the trading robot needs to be constantly monitored. The trader should periodically update parameters for the algorithmic work in different markets. After all, as it was said, the robot does not know how to conduct various analyzes to adjust independently to the current situation.
The decision-making speed of the algorithmic system is much higher. How much will it take for a human to count 826 * 6983 / √892, and how much for a trading robot? How many actions will it take for a human to open a deal, and how many for a trading advisor? How much time does a trader need to make a trading decision, and how much an algorithmic robot advisor need? Everything is obvious here – the trading robot processes information much faster, and thus, executes very quickly the algorithm.
The trading robot is emotionless. Everyone is aware that 95% of the losses in the financial markets are not due to a bad trading strategy, but to the trader’s emotions. The trader might have been underexposed, overexposed, early entered into the deal, “crazy”. There are many synonyms to how to get a loss without controlling emotions. Since the trading robot cannot think, it does not have emotions.
How is the algorithmic trading arranged?
An algorithm is how we process data. It can be simple, difficult, or it can be a completely different thing. It can be a program that performs a simple task: two plus two – buy, two plus one – sell. Or it may be a big team of programmers, mathematicians, physicists, giant servers, supercomputers that receive the same data, but process it in their own way and make a decision on the basis of this complex analysis.
In addition to complex strategies, there are also strategies that are very simple from the algorithm point of view. For example, there is a classical arbitration: when in Frankfurt some share or a currency pair went up, we buy it, but in London. It’s all very simple, but due to the fact that many understand this logic, you need to be faster than others. Accordingly, the competition in this case is moving towards technologies. The one who has a faster communication channel, signal processing, operating system and so on is the winner. The question is literally in micro- and nanoseconds.
So, for such simple strategies, profits can turn out to be huge. But such strategies cannot work with big amounts. Because if you execute a huge number of transactions for a billion dollars per second, it becomes obvious that it won`t be possible to find counterparties for such large amounts in such a short time frame. A problem of liquidity rises up. Accordingly, such algorithms exist, but they are piece-wise – you cannot earn more money. But such level of profitability affects amateur traders mostly – there is a feeling that all this is very easy and profitable.
Was it a mistake or a wrong trading algorithm?
You work in a competitive environment. Can your competitors look at how your algorithmic assistant trades, figure out your algorithm and use it themselves?
Now about whether it is possible to go broke. Theoretically, in all algorithmic systems should be an option of risk management, in addition, they are always monitored by a human being. But still sometimes even very serious companies face problems. Do you remember, for example, the story with the Knight Capital, which lost $ 400 million in a couple of hours, while it was about their two or three year’s revenue?
Sometimes yes, but not always. For example, there is a simple strategy, everyone knows how it works, but still for some reason it works better for us. Hence, there is some kind of trick, and very often it is not marketable – just as an example: we may use gold cables instead of copper ones. This can be found out through espionage only, but this is not an algorithmic advantage of course.
How common is espionage in your business?
Espionage is incredibly developed, as well as secrecy. It is very common for companies trying not to “show off” – no office, no evident contacts, nothing.
According to your forecast, how will the algorithmic trading affect the traders’ labor market?
Right you are here, as today there is a clear tendency to robotize manual labor. And this is manifested not only in the trading sphere. Probably, there will be a category of people who will lose their jobs, it always happens, but a massive outflow of traders will not take place. They will tend to market analysis and trend studies. At the same time, buying and selling will become more a function of trading robots.
Human being traders will not disappear. It is more likely that the scalpers performing operations in manual mode will not be useful, since the exchange algorithmic robots execute trades with greater efficiency. Algorithmic robots themselves do not earn money. A fundamental idea with its basic points is required. Therefore, we hardly can talk about the success of absolutely all algorithmic systems that will appear in the market. In most cases, they will experience the same fate as ordinary human being traders, meaning the vast majority of them eventually will “die” of losses.
No technical analysis carried out with the help of the algorithmic system is able to predict the future and earn Money. Only a human being possesses intuition, while algorithmic systems don`t.