Abstract
This article develops a model for forecasting short-term price movements for the EUR/USD currency pair using historical analogies, taking into account candlestick patterns and trading volumes. It proposes a mathematical formalization of classic reversal patterns (Doji, Hammer/Hanging Man, Engulfing) as a system of quantitative inequalities, as well as a method for confirming their significance through relative tick volume using Fibonacci retracements. A dynamic procedure is introduced for determining the length of the analyzed window and classifying the trend/correction to assess market structure, based on the average market volatility indicator ATR (Average True Range), which reflects the typical range of price fluctuations over a certain period. The model identifies historical periods that are most similar to the current market situation and uses them to generate a forecast based on a neural network (Weight-Net) to determine which factor (price movement, candlestick patterns, or trading volume) should be prioritized at a given moment. Empirical testing on a sample of 15,788 candlestick bars (30-minute timeframe, January 2025 - April 2026) using a step-forward approach demonstrated an average price direction prediction accuracy of 60 % with an MAE of 9.8-11.2 points. The model performance was significantly dependent on the trading session.
Keywords
technical analysis; candlestick patterns; tick volume; historical analogue search; Forex forecasting; neural network feature weighting.
References
Nison S. Japanese Candlestick Charting Techniques: A Contemporary Guide to the Ancient Investment Technique of the Far East. New York, NY: New York Institute of Finance, 1991.
Murphy J.J. Technical Analysis of the Financial Markets: A Comprehensive Guide to Trading Methods and Applications. New York, NY: New York Institute of Finance, 1999.
Wilder J.W. New Concepts in Technical Trading Systems. Greensboro, NC: Trend Research, 1978.
Box G.E.P., Jenkins G.M., Reinsel G.C., Ljung G.M. Time Series Analysis: Forecasting and Control, 5th ed. Hoboken, NJ: Wiley, 2015. DOI: 10.1002/9781118619193
Bollerslev T. Generalized Autoregressive Conditional Heteroskedasticity. Journal of Econometrics, 1986, vol. 31, no. 3, pp. 307-327. DOI: 10.1016/0304-4076(86)90063-1
Hochreiter S., Schmidhuber J. Long Short-Term Memory. Neural Computation, 1997, vol. 9, no. 8, pp. 1735-1780. DOI: 10.1162/neco.1997.9.8.1735