Paper: arXiv 2306.03763
Authors: Zihan Chen, Lei Nico Zheng, Cheng Lu, Jialu Yuan, Di Zhu
Abstract
ChatGPT has demonstrated remarkable capabilities across various natural language processing (NLP) tasks. However, its potential for inferring dynamic network structures from temporal textual data, specifically financial news, remains an unexplored frontier. In this research, we introduce a novel framework that leverages ChatGPT’s graph inference capabilities to enhance Graph Neural Networks (GNN). Our framework adeptly extracts evolving network structures from textual data, and incorporates these networks into graph neural networks for subsequent predictive tasks. The experimental results from stock movement forecasting indicate our model has consistently outperformed the state-of-the-art Deep Learning-based benchmarks. Furthermore, the portfolios constructed based on our model’s outputs demonstrate higher annualized cumulative returns, alongside reduced volatility and maximum drawdown. This superior performance highlights the potential of ChatGPT for text-based network inferences and underscores its promising implications for the financial sector.
Complexity vs Empirical Score
- Math Complexity: 4.0/10
- Empirical Rigor: 6.0/10 (capped from 7.5: transaction costs not modelled; see penalty flags)
- Quadrant: Street Traders — empirical and applied, lower math complexity (effective empirical rigor capped from 7.5)
Why this score: The paper leverages advanced deep learning architectures (GNNs, LSTMs) and NLP techniques, but the mathematics is primarily application-oriented rather than heavily theoretical, lacking dense derivations or new theoretical proofs. Empirical rigor is high due to the use of a real-world dataset (DOW 30), detailed backtesting with portfolio construction, and reporting of specific metrics (F1 scores, annualized returns, volatility, drawdown).
Research Flowchart
flowchart TD
A["Research Goal<br>& Question"] --> B["Data & Inputs<br>Temporal Financial News"]
B --> C["Methodology Steps<br>Extract Evolving Network Structures using ChatGPT"]
C --> D["Computational Processes<br>Integrate Networks into GNN Model"]
D --> E["Key Findings & Outcomes<br>Superior Stock Movement Prediction"]
E --> F["Portfolio Construction<br>Higher Returns & Reduced Risk"]
subgraph Key_Research_Goal
A
end
subgraph Methodology
C
end
subgraph Computation
D
end
subgraph Outcomes
E
F
end
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 novel framework that uses ChatGPT to infer dynamic network structures from financial news and integrates these graphs into a Graph Neural Network for stock movement prediction.
Source sentence
we introduce a novel framework that leverages ChatGPT’s graph inference capabilities to enhance Graph Neural Networks (GNN).
Assumes 5
- Stock prices reflect all relevant market information (Efficient Market Hypothesis). market
Source sentence
Owing to the efficient market hypothesis, stock prices are assumed to encapsulate all relevant market information
- ChatGPT was trained only on data up to September 2021, so test period starts after that to avoid data leakage. data
Source sentence
To prevent potential data leakage issues, considering that the ChatGPT model was trained on data available only up to September 2021, we designate the test period to begin from October 1, 2021.
- The inferred network structure from ChatGPT is sufficient as input for the GNN without additional edge attributes. model
Source sentence
our study solely utilizes the network structure inferred by ChatGPT as input for GNN.
- Daily news headlines mentioning at least one DOW 30 company provide relevant textual data for graph inference. data
Source sentence
We then extract news that not only originates from reputable media outlets but also explicitly mentions at least one DOW 30 company.
Show 1 more
- The stock movement labels are defined by price return thresholds of ±1%. statistical
Source sentence
defined thresholds (rup = 0.01, rdown = −0.01) as follows
Data
- Dow Industrial Average 30 Companies (DOW 30) market-prices public
Source sentence
real-world dataset, setting the DOW 30 companies as our targets.
- CRSP daily stock data market-prices commercial
Source sentence
we acquire daily numerical variables of each DOW 30 company from the CRSP Databases as stock market data.
- Financial news headlines text-news commercial
Source sentence
We collect 2,713,233 and 3,717,666 unique headlines for the training and test periods respectively, gleaned from 5,489 unique providers.
Costs
Not modeled
Finds 2
- The ChatGPT‑informed GNN model consistently outperforms all baseline models in weighted F1, Micro, and Macro F1 metrics with a minimum improvement of 1.8%. predictive-accuracy
Source sentence
the experimental results demonstrate that our model consistently surpasses all baseline models in weighted F1, Micro, and Macro F1 metrics with a minimum improvement of 1.8%.
- Portfolios constructed from the model’s outputs achieve higher cumulative returns, lower annualized volatility, and reduced maximum drawdown compared to benchmarks. performance
Source sentence
Our model consistently exceeds benchmarks in terms of cumulative returns during the out-of-sample period. Our model also manifests a lower annualized volatility and a reduced maximum drawdown.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 25 related papers). Coverage is partial, so this is neither replication nor refutation.
11 papers in the archive : Can Large Language Models Trade? Testing Financial Theories … ; Temporal Relational Reasoning of Large Language Models for … ; A Survey of Large Language Models for Financial …
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.