Dual-CLVSA: a Novel Deep Learning Approach to Predict Financial Markets with Sentiment Measurements

TL;DR

This paper introduces dual-CLVSA, a deep learning model that combines trading data and social sentiment to improve financial market prediction accuracy, demonstrating the value of sentiment measurements in forecasting market movements.

Contribution

The paper presents a novel dual-CLVSA model that effectively fuses trading data and sentiment measurements for enhanced financial market prediction.

Findings

Dual-CLVSA outperforms models using only trading data.

Sentiment measurements provide additional profitable features.

The approach is validated on eight years of SPDR SP 500 Trust ETF data.

Abstract

It is a challenging task to predict financial markets. The complexity of this task is mainly due to the interaction between financial markets and market participants, who are not able to keep rational all the time, and often affected by emotions such as fear and ecstasy. Based on the state-of-the-art approach particularly for financial market predictions, a hybrid convolutional LSTM Based variational sequence-to-sequence model with attention (CLVSA), we propose a novel deep learning approach, named dual-CLVSA, to predict financial market movement with both trading data and the corresponding social sentiment measurements, each through a separate sequence-to-sequence channel. We evaluate the performance of our approach with backtesting on historical trading data of SPDR SP 500 Trust ETF over eight years. The experiment results show that dual-CLVSA can effectively fuse the two types of data, and verify that sentiment measurements are not only informative for financial market predictions, but they also contain extra profitable features to boost the performance of our predicting system.