Development of Decision Support Systems Based on Fuzzy and Binary Logic for the FOREX Foreign Exchange Market

Introduction. Most authors considered the use of only binary logic and technical analysis, which does not allow for effective consideration of market uncertainty and rapid dynamic. Other researchers considered the use of fuzzy logic, but these studies are limited to local markets or do not provide integration with more flexible types of analysis such as ML. It was also found that no comparative analysis of the effectiveness of different logical approaches (fuzzy, classical, probabilistic logic) is conducted, which creates a gap in the scientific justification of the choice of a particular method. The potential of multi-timeframe analysis is also practically not taken into account, although it can increase the accuracy and stability of decisions made. The above indicates the need for a comprehensive study that would combine the advantages of various logical approaches, machine learning and multi-timeframe analysis within a single hybrid DSS. This would also allow a reasonable approach to the choice of a specific method. Research objective. The aim of this work is to develop multi-timeframe hybrid DSS for algorithmic trading based on fuzzy and classical binary logic with probabilistic elements. This will make it possible to increase the efficiency of algorithmic trading systems. Results. The study consisted in the development of multi-timeframe hybrid DSS based on binary and fuzzy logic with probabilistic elements, as well as their comparative analysis. As a source of signals for further decision-making, the system uses forecasts made by the Random Forest model. Cross-Validation was used to train the model to predict not only the opening, maximum, minimum and closing values ??of the position (Open, High, Low, Close – OHLC), but also the level of confidence of these predictions. The Mamdani fuzzy logic system [13, 14] was used as a fuzzy logic system for DSS. Both DSS were implemented in the MQL5 programming language. The backtest was carried out on the MT5 platform. As a result, the decision support system based on fuzzy logic showed a significant advantage over the decision support system based on classical binary logic with a Win Rate of 60.81%, and an annual return of 58% and a Sharpe ratio of 1.33. While the decision support system based on binary logic showed the following results: Win Rate of 34.16%, and an annual return of –95.46% and a Sharpe ratio of –5. An applied aspect of using the obtained scientific result is the possibility of improving DSS for making trading decisions. Conclusions. The study showed that multi-timeframe hybrid DSS based on fuzzy logic with probabilistic elements allows making more effective decisions than DSS based on binary logic. This study allows for a reasoned approach to choosing a specific method. In addition, the proposed methodology and constructed models can be used by other researchers in the field of financial technologies for the further development of decision support systems in financial markets. Future research will be aimed at improving time series forecasting methods in order to improve the quality of input signals for the trading system. Keywords: algorithmic trading, FOREX, Machine Learning, fuzzy logic, Mamdani.