A News-based Machine Learning Model for Adaptive Asset Pricing

TL;DR

This paper introduces the NEUS model, which leverages financial news and machine learning to improve stock return prediction and asset pricing accuracy over traditional models.

Contribution

It develops a novel news-based asset pricing model using machine learning to derive company embeddings and select basis assets for better return prediction.

Findings

NEUS outperforms the Fama-French 5-factor model in fitting stock returns.

The model effectively captures news-driven variations in asset prices.

High-dimensional statistical methods enhance prediction accuracy.

Abstract

The paper proposes a new asset pricing model -- the News Embedding UMAP Selection (NEUS) model, to explain and predict the stock returns based on the financial news. Using a combination of various machine learning algorithms, we first derive a company embedding vector for each basis asset from the financial news. Then we obtain a collection of the basis assets based on their company embedding. After that for each stock, we select the basis assets to explain and predict the stock return with high-dimensional statistical methods. The new model is shown to have a significantly better fitting and prediction power than the Fama-French 5-factor model.