PEN: Prediction-Explanation Network to Forecast Stock Price Movement with Better Explainability

Abstract

Nowadays explainability in stock price movement prediction is attracting increasing attention in banks, hedge funds and asset managers, primarily due to audit or regulatory reasons. Text data such as financial news and social media posts can be part of the reasons for stock price movement. To this end, we propose a novel framework of Prediction-Explanation Network (PEN) jointly modeling text streams and price streams with alignment. The key component of the PEN model is an shared representation learning module that learns which texts are possibly associated with the stock price movement by modeling the interaction between the text data and stock price data with a salient vector characterizing their correlation. In this way, the PEN model is able to predict the stock price movement by identifying and utilizing abundant messages while on the other hand, the selected text messages also explain the stock price movement. Experiments on real-world datasets demonstrate that we are able to kill two birds with one stone: in terms of accuracy, the proposed PEN model outperforms the state-of-art baseline; on explainability, the PEN model are demonstrated to be far superior to attention mechanism, capable of picking out the crucial texts with a very high confidence.

Learning to Generate Explainable Stock Predictions using Self-Reflective Large Language Models

“…(1) a Summarize module, which generates a summary of factual information from the unstructured text inputs; (2) an Explain module, which generates explanations for its stock predictions and refines them through an iterative self-reflective process; and (3) a Predict module, which generates confidence-based predictions after fine-tuning a LLM using its self-generated annotated samples. [67], which is a popular benchmark used in many stock prediction works [22,40,56]. The duration of the original dataset ranges from year 2014-2016, and we collect an updated version for year 2020-2022.…”

“…[...] refers to truncated text. However, due to their complex and quantitative nature, traditional deep-learning methods in stock prediction are black box models and do not address the explainability of their predictions arXiv:2402.03659v3 [cs.LG] 29 Feb 2024 [40]. This reduces their usability in practical applications, as users might not be able to trust [4] the results to invest their capital.…”

“…This reduces their usability in practical applications, as users might not be able to trust [4] the results to invest their capital. Even among works that deal with explainable stock predictions [6,40], the "explanations" are often simply defined as the specific texts that caused the price movement, which are usually obtained by analyzing learnt attention weights [15,56]. For example, these models could analyze a series of texts regarding Apple stock and determine that its Positive prediction is attributed to the text "Apple reported revenue of $90.1 billion, beating expectations".…”

An Advanced Explainable Belief Rule-Based Framework to Predict the Energy Consumption of Buildings

“…Moreover, it does not consider domain knowledge and data uncertainties. Li et al [31] proposed a Prediction-Explanation Network (PEN) to predict stock price movement with better explainability. They employed a shared representation learning module to identify the correlation between text and stock price with a vector of salience.…”