BERTFOREX: Cascading Model for Forex Market Forecasting Using Fundamental and Technical Indicator Data Based on BERT

Abstract

Foreign exchange (Forex) rate forecasting is presently pursued by many researchers as it plays an important role in financial technology and business. The challenge of Forex research lies in its characteristics, fluctuation, non-linearity, and random walk phenomena. Several related studies generate forecasting signals using fundamental data (FD) and technical indicator data (TI) to support Forex. FD is an indicator of country economic conditions, while TI shows the price pattern-based signal. However, these two indicators pose two major limitations on their deployment. First, modeling a sequential neural network causes gradient vanishing and information loss. Second, FD exerts a significant impact on currency price upon its quarterly update and release. The second limitation is known as FD releasing problem. Moreover, Forex forecasting using FD and TI is typically conducted in an equal aggregation manner, resulting in inaccurate prediction due to unequal data changing frequency. In this work, we propose BERTFOREX, a cascading model for Forex time-series forecasting. The proposed technique uses deep learning Bidirectional Encoder Representations from Transformer (BERT) based on FD and TI data characteristics. The technique first applies FD to extract the hidden patterns over the designated period. Then, these extracted hidden patterns of FD are aggregated as additional weights to TI since FD frequency changes slower than that of TI. This yields a combined aggregated pattern of FD and TI. BERT again applies the aggregated pattern to discover underlying patterns within TI and FD over other influencing days. We demonstrate the efficiency of BERTFOREX aggregated representation using a simple neural network in forecasting. The proposed method outperforms other methods in terms of percentage of correct signals, sensitivity, specificity, precision, and negative predictive value.

FOREX Prices Prediction Using Deep Neural Network and FNF

“…The ensemble model takes all the signals from the different strategies and uses majority voting logic to decide what to do next [12] .The problem with this system is its low accuracy, and it was tested only on the EUR/USD dataset. Pornwattanavichai and Maneeroj [13] proposed a cascading model for the FOREX market. They made forecasts using Fundamental Data and Technical indications based on BERT.…”

Incorporating Multi-Source Market Sentiment and Price Data for Stock Price Prediction

“…In recent research, transformerbased natural language processing methods have shown promising results in financial text data analysis. Particularly Google's BERT model [22], as a transformer-based pretrained model, made remarkable progress in natural language processing and was applied to sentiment analysis of financial texts in many studies (e.g., [23,24]). For instance, Hiew et al's study shows that a BERT-based sentiment analysis approach is superior to models such as FastText or a multichannel Convolutional Neural Network (CNN) [25].…”

Modelo de análisis multivariante aplicado a la predicción de la tendencia del tipo de cambio euro-dólar

“…Algunos trabajos notables, de previsión de la evolución de los tipos de cambio se basan en técnicas de inteligencia artificial. Entre los enfoques más avanzados se encuentran el aprendizaje profundo por refuerzo [489], la memoria a largo plazo (LSTM) [490], el análisis de sentimiento a partir de contenido textual [491,492], el análisis wavelet [493], los algoritmos genéticos [494], las máquinas de vectores de soporte y otros enfoques [495]. En un intento por mejorar la precisión de estos algoritmos, los investigadores han propuesto técnicas para combinarlos y formar sistemas más robustos.…”