Paper: arXiv 2411.18997
Authors: Yonggai Zhuang, Haoran Chen, Kequan Wang, Teng Fei
Abstract
The complexity of stocks and industries presents challenges for stock prediction. Currently, stock prediction models can be divided into two categories. One category, represented by GRU and ALSTM, relies solely on stock factors for prediction, with limited effectiveness. The other category, represented by HIST and TRA, incorporates not only stock factors but also industry information, industry financial reports, public sentiment, and other inputs for prediction. The second category of models can capture correlations between stocks by introducing additional information, but the extra data is difficult to standardize and generalize. Considering the current state and limitations of these two types of models, this paper proposes the GRU-PFG (Project Factors into Graph) model. This model only takes stock factors as input and extracts inter-stock correlations using graph neural networks. It achieves prediction results that not only outperform the others models relies solely on stock factors, but also achieve comparable performance to the second category models. The experimental results show that on the CSI300 dataset, the IC of GRU-PFG is 0.134, outperforming HIST’s 0.131 and significantly surpassing GRU and Transformer, achieving results better than the second category models. Moreover as a model that relies solely on stock factors, it has greater potential for generalization.
Complexity vs Empirical Score
- Math Complexity: 6.5/10
- Empirical Rigor: 8.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs Graph Neural Networks (GNNs) combined with GRUs, representing advanced machine learning architectures with non-trivial mathematical foundations. It also presents strong empirical results on a standard financial dataset (CSI300) with specific performance metrics (IC 0.134) and comparisons against established benchmarks, indicating a high degree of implementation and backtesting readiness.
Research Flowchart
flowchart TD Start["Research Goal: Predict stock returns using only factors, capturing inter-stock correlations without extra data"] --> Inputs["Data Inputs: Stock Factors Historical Data"] Inputs --> Method["Core Methodology: GRU-PFG Model"] Method --> Step1["1. Feature Extraction: GRU processes temporal factor sequences"] Step1 --> Step2["2. Graph Construction: Project factors into correlation graph"] Step2 --> Step3["3. Graph Learning: GNN extracts inter-stock correlations"] Step3 --> Step4["4. Prediction: Combine temporal & correlation features for return forecast"] Step4 --> Outcomes["Key Findings"] Outcomes --> F1["IC Score: 0.134 on CSI300 dataset"] Outcomes --> F2["Outperforms GRU & Transformer, surpasses HIST (IC 0.131)"] Outcomes --> F3["Matches 2nd-category models while using only standard factors"] Outcomes --> F4["Superior generalization potential"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a new stock prediction model called GRU-PFG that projects stock factors into a graph neural network to extract inter‑stock correlations.
Source sentence
Considering the current state and limitations of these two types of models, this paper proposes the GRU-PFG (Project Factors into Graph) model. This model only takes stock factors as input and extracts inter-stock correlations using graph neural networks.
Assumes
No explicit assumptions found in the text.
Data
- CSI300 market-prices unknown
Source sentence
The experimental results show that on the CSI300 dataset, the IC of GRU-PFG is 0.134, outperforming HIST’s 0.131 and significantly surpassing GRU and Transformer
- CSI100 market-prices unknown
Source sentence
The experimental results based on the CSI100 and CSI300 datasets demonstrate a significant advantage of our model over previous models
Costs
Not applicable
Finds 2
- GRU-PFG achieves an IC of 0.134 on the CSI300 dataset, surpassing HIST’s 0.131 and significantly outperforming GRU and Transformer models. performance
Source sentence
The experimental results show that on the CSI300 dataset, the IC of GRU-PFG is 0.134, outperforming HIST’s 0.131 and significantly surpassing GRU and Transformer
- The model performs better on the larger CSI300 universe than on CSI100, indicating that capturing more inter‑stock relationships improves prediction. performance
Source sentence
It is worth mentioning that our model exhibits a greater advantage on the CSI300 dataset compared to the CSI100 dataset. One possible explanation is that our model effectively captures the relationships between stocks.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 17 related papers). Coverage is partial, so this is neither replication nor refutation.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (41)
- Deep learning for stock prediction using numerical and textual information
- Stock market prediction and portfolio composition using a hybrid approach combined with self-adaptive evolutionary algorithm (2022) — link
- The importance of financial ratios in predicting stock price trends: A case study in emerging markets (2016)
- Hedging inflation with individual US stocks: A long-run portfolio analysis (2016) — link
- Agent based-stock flow consistent macroeconomics: Towards a benchmark model (2016) — link
- Firms’ information security investment decisions: Stock market evidence of investors’ behavior (2011) — link
- Financial reporting of good news and bad news: evidence from accounting narratives (2003) — link
- Design and Implementation of a Comprehensive Intelligent E-Learning Systems Integrated Web App Solution For Dyslexic Students
and 33 more
- Gate-variants of Gated Recurrent Unit (GRU) neural networks
- DTSMLA: A dynamic task scheduling multi-level attention model for stock ranking (2024) — link
- Does Equity Ownership Matter for Corporate Social Responsibility? A Literature Review of Theories and Recent Empirical Findings (2018) — link
- StockNet—GRU based stock index prediction (2022) — link
- How financial literacy and impatience shape retirement wealth and investment behaviors (2020) — link
- FinGAT: Financial Graph Attention Networks for Recommending Top-$K$K Profitable Stocks (2023)
- Sentiment-aware stock market prediction: A deep learning method
- Applications of deep learning in stock market prediction: Recent progress (2021) — link
- A hybrid model integrating deep learning with investor sentiment analysis for stock price prediction (2021) — link
- Combining Support Vector Machine with Genetic Algorithms to optimize investments in Forex markets with high leverage (2018) — link
- Evaluating and understanding text-based stock price prediction models (2014) — link
- 101 Formulaic Alphas (2015)
- Performance versus turnover: a story by 4000 alphas (2016)
- Cosine similarity to determine similarity measure: Study case in online essay assessment
- Management of risks associated with the disclosure of future-oriented information in integrated reports (2021) — link
- Discovering public sentiment in social media for predicting stock movement of publicly listed companies (2017) — link
- Learning Multiple Stock Trading Patterns with Temporal Routing Adaptor and Optimal Transport (2021)
- Predicting shareholder litigation on insider trading from financial text: An interpretable deep learning approach (2020) — link
- Stock market value prediction using neural networks
- Can institutions and macroeconomic factors predict stock returns in emerging markets? (2014) — link
- Stock market’s price movement prediction with LSTM neural networks
- Predicting stock and stock price index movement using trend deterministic data preparation and machine learning techniques (2015)
- MDGNN: Multi-Relational Dynamic Graph Neural Network for Comprehensive and Dynamic Stock Investment Prediction — → 2402.06633
- Determinants of Stock Market Investors’ Behavior in COVID-19: A Study on the Pakistan Stock Exchange (2020)
- Attention Is All You Need (2017)
- Stock market index prediction using deep Transformer model (2022) — link
- ALSTM: An attention-based long short-term memory framework for knowledge base reasoning (2020) — link
- Forecasting Model of Shanghai Stock Index based on ALSTM
- HIST: A Graph-based Framework for Stock Trend Forecasting via Mining Concept-Oriented Shared Information (2021)
- Qlib: An AI-oriented Quantitative Investment Platform (2020)
- Multi-Graph Convolutional Network for Relationship-Driven Stock Movement Prediction
- A Review of Recurrent Neural Networks: LSTM Cells and Network Architectures (2019) — link
- Stock Trend Prediction with Machine Learning: Incorporating Inter-Stock Correlation Information through Laplacian Matrix (2024)
1 of 41 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.