Feature-Enhanced for Price Prediction: Validating Incremental Contributions of Fundamental, Technical, and Macro-Sentiment Composite Volatility Indicators

DOI

10.2991/978-94-6463-823-3_22How to use a DOI?

Keywords

Transformer; ConvTrans; Interactive Indicator; MSCVI; Dynamic Fusion

Abstract

This study combines traditional financial theory with deep learning technology to provide an efficient solution for stock price prediction. Based on the Transformer and ConvTrans, this study validates the incremental contributions of fundamental, technical, and macro-sentiment composite volatility indicators (MSCVI) in predicting future prices. To capture nonlinear relationships and dynamic market behavior, this study calculates the interactive indicators with fundamental and technical indicators. MSCVI aims to provide a more sensitive and robust market risk assessment tool, by using the dynamic fusion mechanism to combine the complementarity of macro and sentiment data. This study integrates MSCVI into the architecture of the classic Transformer model and Conv-Trans, from attention computation to positional encoding. ConvTrans combined with MSCVI-Guided Attention Bias shows the best performance, reaching an R² of 0.9716 and a MAPE of only 2.87%. This demonstrates the effectiveness of incorporating market-state volatility into attention mechanisms. Although the model overcomes the multi-source driven limitations of GARCH-type models by dynamically integrating macro and sentiment fluctuation mechanisms, it is sensitive to extreme sentiment noise and has a single data source bias. Future improvements will further adopt Transformer hybrid architecture, combined with a large language model to optimize sentiment analysis, and verify the universality of the two-way volatility feedback mechanism through multi-asset market tests.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Liangzhou Qu
PY  - 2025
DA  - 2025/08/31
TI  - Feature-Enhanced for Price Prediction: Validating Incremental Contributions of Fundamental, Technical, and Macro-Sentiment Composite Volatility Indicators
BT  - Proceedings of the 2025 3rd International Conference on Image, Algorithms, and Artificial Intelligence (ICIAAI 2025)
PB  - Atlantis Press
SP  - 236
EP  - 244
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-823-3_22
DO  - 10.2991/978-94-6463-823-3_22
ID  - Qu2025
ER  -