https://c.radikal.ru/c04/1804/8e/47cc57a374d6.jpgThe typical mistakes and delusions of the algo traderWe won’t concern the mistakes of program code strategy realization and we will concentrate the attention on the conceptual mistakes which the algo trader usually commits when thinks out the logic of the work of the trade system and when he estimates its results.
Ignoring the possible differences between the LONG and SHORT positionsOften, the algo trader doesn’t see any difference between “bull” and “bear” trade (that is unsurprising, many of us have graduated from the “school” of Forex where it is difficult to notice it), and meanwhile often happens it is useful
to work separately with positions for sale (SHORT) and purchase (LONG). Testing the algorithms, you will be able to see from time to time the positive result of trade. It is formed by two groups of transactions, for example, the profitable group of transactions of LONG in the set and smaller unprofitable group of transactions SHORT on the module (or the otherwise).
Ignoring the possible differences between the positions which are opened at different times of the dayOften, the algo trader doesn’t consider the fact that at a different time of the day the market can behave differently that will affect decisively success or fail of an algorithm work in general. It is possible to see statistics for an algorithm taking into consideration the time of day by means of the service of analytics available in the section of the website “Results of Modelling”. Having got acquainted with the statistics, perhaps want to add to the algorithm a time filter or to create a special subsystem which will distinguish the condition of the market (suitable for your strategy) from another (not suitable).
The aspiration to entirely rely on the technical analysis indicatorsReally, in the literature devoted to exchanging trade, especially in texts which carry a hidden (or obvious) advertising manner, it is possible to meet many optimistic judgments concerning the widespread indicators of the technical analysis accompanied by the numerous examples of their successful application on the most different tools and time intervals. Nevertheless, we will dare to claim that the only indicators of stable and profitable trade system can’t be constructed. Certainly, indicators should be used in many trade algorithms, however, their signals should be filtered or confirmed with an application of the most different mathematical apparatus: mathematical statistics, time series, various machine learning, neural networks, fractals, etc. Firstly, we wait for the algo- the new original ideas which will help to create and realize a successful trade model.
The values of parameters adjustment to an algorithmAlmost any algorithm has input parameters which serve for control of its work in the market. The period of moving average or the threshold value of some indicator serving as the criterion of an entrance to a position, etc. can be such parameters, for example. As a rule, the author of an algorithm quickly finds what values of input parameters yield the positive result on the time interval chosen by it for testing and considers the mission completed. However, almost any algorithm can be “adjusted” on a concrete time span of the past where it will show the profit that, nevertheless, doesn’t give any guarantee that the algorithm will show positive results and on other time spans, especially future. Therefore, testing of an algorithm should be made on different market spheres to capture as it is possible a bigger time gap, extend at least in few years. Most likely you will see that at different time intervals the best result is yielded by different input parameters. In this case, we can advise you to apply methods of machine learning for an automatic configuration of your trade system on constantly changing dynamics of the market.
The positive result of testing: what does it consist of?Having seen positive result of testing the algorithm, the study of the profitability schedule (which is available in the “Results of Modelling” section on pressing of the emerging icons with the image of the schedule in the table of results). The more uniform growth shows the profitability schedule, the better it is. It testifies the stability of trade system. To the other side, in the general positive profitability consists of two periods which are obviously different — unprofitable and profitable — it demonstrates that it is necessary to work on an algorithm as its success decisively depends on the nature of the market, and there is no guarantee that the period of profitability growth won’t be replaced by the period of its falling when market conditions exchange again.
#tradingene #tradingeneration
