FinReport: Explainable Stock Earnings Forecasting via News Factor Analyzing Model

TL;DR

FinReport presents an explainable stock earnings forecasting framework that fuses financial news analysis with a multi-factor return model. It introduces a News Factorization Module leveraging Semantic Role Labeling (SRL) and Semantic Dependency Parsing Graphs (SDPG) together with stock factors via an adaptive weight $W_\alpha$, feeding a 2-layer MLP to produce a three-class return signal. The Return Forecasting uses a FF5-News model by incorporating a News Effect Factor $M_t$ into the Fama-French 5-factor framework, while the Risk Assessment module employs an EGARCH-based VaR calculation $VaR = \mu_t - \sigma_t Z_\alpha$ to quantify downside risk. An LLM then generates readable reports from multi-dimensional forecasts and VaR, with experiments showing improved ROI, Sharpe ratio, and VaR accuracy, and real-world backtests demonstrating strong profitability and risk control.

Abstract

The task of stock earnings forecasting has received considerable attention due to the demand investors in real-world scenarios. However, compared with financial institutions, it is not easy for ordinary investors to mine factors and analyze news. On the other hand, although large language models in the financial field can serve users in the form of dialogue robots, it still requires users to have financial knowledge to ask reasonable questions. To serve the user experience, we aim to build an automatic system, FinReport, for ordinary investors to collect information, analyze it, and generate reports after summarizing. Specifically, our FinReport is based on financial news announcements and a multi-factor model to ensure the professionalism of the report. The FinReport consists of three modules: news factorization module, return forecasting module, risk assessment module. The news factorization module involves understanding news information and combining it with stock factors, the return forecasting module aim to analysis the impact of news on market sentiment, and the risk assessment module is adopted to control investment risk. Extensive experiments on real-world datasets have well verified the effectiveness and explainability of our proposed FinReport. Our codes and datasets are available at https://github.com/frinkleko/FinReport.

Figures (6)

• Figure 3: The processing results of SRL and SDPG.
• Figure 4: Examples of both postive and negateive report
• Figure 5: Comparing the performance of different models in real-world scenarios
• Figure 6: Details of backtest on Self-supervised SRLP with factors
• Figure 7: Details of backtest on Chinese Pert large