English

GPT-Signal: Generative AI for Semi-automated Feature Engineering in the Alpha Research Process

Computational Engineering, Finance, and Science 2024-10-25 v1

Abstract

In the trading process, financial signals often imply the time to buy and sell assets to generate excess returns compared to a benchmark (e.g., an index). Alpha is the portion of an asset's return that is not explained by exposure to this benchmark, and the alpha research process is a popular technique aiming at developing strategies to generate alphas and gain excess returns. Feature Engineering, a significant pre-processing procedure in machine learning and data analysis that helps extract and create transformed features from raw data, plays an important role in algorithmic trading strategies and the alpha research process. With the recent development of Generative Artificial Intelligence(Gen AI) and Large Language Models (LLMs), we present a novel way of leveraging GPT-4 to generate new return-predictive formulaic alphas, making alpha mining a semi-automated process, and saving time and energy for investors and traders.

Keywords

Cite

@article{arxiv.2410.18448,
  title  = {GPT-Signal: Generative AI for Semi-automated Feature Engineering in the Alpha Research Process},
  author = {Yining Wang and Jinman Zhao and Yuri Lawryshyn},
  journal= {arXiv preprint arXiv:2410.18448},
  year   = {2024}
}

Comments

13 pages, 16 figures, 1 table, accepted by FINNLP 2024