Identifying moments of decision making on trade in financial time series using fuzzy cluster analysis

Abstract

The article investigates the problem of identifying trading decision points in financial time series using the Fuzzy C-Means (FCM) method. The authors argue that classical forecasting methods have limited effectiveness for decision-making in trading, as they do not take into account market structure and nonlinear patterns. The proposed methodology involves analysing time series using additional features derived from technical indicators (MACD, Stochastic) and further clustering based on FCM, which allows identifying market entry and exit points. In contrast to traditional approaches based on the assessment of forecasting accuracy (e.g. MAE, RMSE, MAPE), this study focuses on financially oriented metrics such as Net Profit, Max Drawdown, Win Rate and Profit Factor, which more accurately reflect the effectiveness of trading strategies in real market conditions. Experiments on the currency pairs EUR/USD, AUD/USD, USD/JPY, USD/CAD on daily and four-hour timeframes have demonstrated that the use of the proposed approach can improve the efficiency of trading strategies. The simulation results showed fairly high stable profitability results with low risks (drawdown). The proposed approach can be useful in developing automated trading systems and further research in the field of financial analytics.