Paper: arXiv 2501.16697
Authors: Suke Li
Abstract
This study analyzes historical data from five agricultural commodities in the Chinese futures market to explore the correlation, cointegration, and Granger causality between Peanut futures and related futures. Multivariate linear regression models are constructed for prices and logarithmic returns, while dynamic relationships are examined using VAR and DCC-EGARCH models. The results reveal a significant dynamic linkage between Peanut and Soybean Oil futures through DCC-EGARCH, whereas the VAR model suggests limited influence from other futures. Additionally, the application of MLP, CNN, and LSTM neural networks for price prediction highlights the critical role of time step configurations in forecasting accuracy. These findings provide valuable insights into the interconnectedness of agricultural futures markets and the efficacy of advanced modeling techniques in financial analysis.
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 4.0/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper employs advanced econometric and deep learning models (DCC-EGARCH, CNN/LSTM) but relies on historical data analysis without providing executable code, backtest performance metrics, or robustness checks needed for trading implementation.
Research Flowchart
flowchart TD
A["Research Goal:<br>Analyze Interdependence of<br>Agricultural Commodities in China's<br>Futures Market"] --> B["Data Input:<br>5 Commodities Futures Data<br>(Peanut, Soybean Oil, etc.)"]
B --> C{"Methodology Steps"}
C --> D["Statistical Analysis:<br>Correlation, Cointegration,<br>Granger Causality"]
C --> E["Time-Series Modeling:<br>VAR, DCC-EGARCH"]
C --> F["Machine Learning:<br>MLP, CNN, LSTM"]
D & E & F --> G["Key Findings & Outcomes"]
G --> H["Peanut & Soybean Oil<br>show significant dynamic linkage<br>(DCC-EGARCH)"]
G --> I["VAR indicates limited<br>influence from other futures"]
G --> J["Neural Networks:<br>Time step configuration critical<br>for price prediction accuracy"]
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 using advanced neural network architectures—MLP, CNN, and LSTM—for forecasting Peanut futures prices and analyzes interdependencies with other agricultural futures.
Source sentence
This study employs advanced neural network architectures—Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM)—for Peanut
Assumes 4
- Error term εt is independently and identically distributed (i.i.d.). statistical
Source sentence
ϵt: Error term at time t, assumed to be independently and identically distributed (i.i.d.).
- Historical data split into training (80%) and test (20%) sets for neural network modeling. data
Source sentence
The historical data is divided into a training dataset (80%) and a test dataset (20%)
- VAR model uses first lag of logarithmic returns as sufficient dynamics. model
Source sentence
The VAR(1) model is formulated as follows for a system of five agricultural commodities’ logarithmic returns of futures prices
- Residuals are assumed normal for OLS inference. statistical
Source sentence
rejects the null hypothesis of normally distributed residuals
Data
- Five agricultural commodities futures data market-prices unknown
Source sentence
This study analyzes historical data from five agricultural commodities in the Chinese futures market
Costs
Not applicable
Finds 5
- A significant dynamic linkage between Peanut and Soybean Oil futures is identified using DCC‑EGARCH. statistical
Source sentence
The results reveal a significant dynamic linkage between Peanut and Soybean Oil futures through DCC-EGARCH
- The VAR model suggests limited influence of other futures on Peanut prices. statistical
Source sentence
whereas the VAR model suggests limited influence from other futures
- Soybean Oil has a significant positive effect on Peanut prices (β = 0.746, p < 0.001). statistical
Source sentence
Oil demonstrates a significant positive effect (β = 0.746, p < 0.001)
- LSTM achieves the best predictive performance with 5 features and a time step of 10, e.g., lowest MSE 23,777.70 and MAE 122.35 at Step 2. predictive-accuracy
Source sentence
LSTM demonstrated the best performance, particularly in Step 2, where it achieved the lowest MSE (23,777.70) and MAE (122.35)
- With 4 features (without Peanut) and time step 10, MLP consistently yields the lowest MSE and MAE across all steps. performance
Source sentence
MLP consistently achieved the lowest MSE and MAE values across all steps
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 54 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 (9)
- Peanut Futures (2021)
- Macroeconomics and reality (1980)
- Dynamic conditional correlation: A simple class of multivariate generalized autoregressive conditional heteroskedasticity models (2002)
- Conditional heteroskedasticity in asset returns: A new approach (1991)
- Gradient-based learning applied to document recognition (1998)
- Long short-term memory (1997)
- Learning representations by back-propagating errors (1986)
- GARCH Models: Structure, Statistical Inference, and Financial Applications (2019)
and 1 more
- Analysis of Financial Time Series (2010)
0 of 9 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.