Paper: arXiv 2410.19291
Authors: Zhiyuan Pei, Jianqi Yan, Jin Yan, Bailing Yang, Ziyuan Li, Lin Zhang, Xin Liu, Yang Zhang
Abstract
Recently, deep learning in stock prediction has become an important branch. Image-based methods show potential by capturing complex visual patterns and spatial correlations, offering advantages in interpretability over time series models. However, image-based approaches are more prone to overfitting, hindering robust predictive performance. To improve accuracy, this paper proposes a novel method, named Sequence-based Multi-scale Fusion Regression Convolutional Neural Network (SMSFR-CNN), for predicting stock price movements in the China A-share market. By utilizing CNN to learn sequential features and combining them with image features, we improve the accuracy of stock trend prediction on the A-share market stock dataset. This approach reduces the search space for image features, stabilizes, and accelerates the training process. Extensive comparative experiments on 4,454 A-share stocks show that the model achieves a 61.15% positive predictive value and a 63.37% negative predictive value for the next 5 days, resulting in a total profit of 165.09%.
Complexity vs Empirical Score
- Math Complexity: 4.5/10
- Empirical Rigor: 8.0/10
- Quadrant: Street Traders — empirical and applied, lower math complexity
Why this score: The paper’s mathematics is moderate, relying primarily on standard CNN architectures and time series decomposition rather than advanced theoretical derivations. However, it demonstrates high empirical rigor through extensive experiments on 4,454 A-share stocks, reporting specific metrics like 61.15% positive predictive value and 165.09% total profit, indicating backtest-ready implementation and substantial data/effort.
Research Flowchart
flowchart TD A["Research Goal<br>Improve Stock Price Prediction<br>Accuracy for China A-Share Market"] --> B["Data Preparation<br>4,454 A-Share Stocks<br>Open/High/Low/Close/Time Images"] B --> C["Methodology: SMSFR-CNN<br>Sequence-based Multi-scale Fusion<br>Regression CNN"] C --> D["Computational Process<br>1. CNN Feature Extraction<br>2. Multi-scale Fusion<br>3. Regression Prediction"] D --> E["Key Findings<br>PPV: 61.15% | NPV: 63.37%<br>Total Profit: 165.09%<br>Reduced Overfitting"]
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 Sequence-based Multi-scale Fusion Regression Convolutional Neural Network (SMSFR-CNN) for stock price movement prediction.
Source sentence
this paper proposes a novel method, named Sequence-based Multi-scale Fusion Regression Convolutional Neural Network (SMSFR-CNN)
Assumes 3
- The weekend effect influences price dynamics in the China A‑share market. market
Source sentence
Within the China A-shares stock dataset, there is a notable phenomenon known as the weekend effect
- Samples with an upper‑limit price movement in the final candlestick are excluded. data
Source sentence
we exclude the samples where the final candlestick within the specified window exhibited an upper limit price movement
- Turnover rate is used instead of trading volume as a more comparable liquidity metric. data
Source sentence
We give the main reason for using the turnover rate feature to replace the volume feature
Data
2022-01 – 2024-06 · China A-share stocks · equities · daily
- China A‑share stock market dataset market-prices commercial
Source sentence
Daily trading data for these 4,454 stocks were collected from January 1, 2012, to June 30, 2024, using the Tonghuashun API
Costs
Unclear
Finds 3
total return 165.1% · max drawdown 27.9%
- The model achieves a 61.15% positive predictive value and a 63.37% negative predictive value for the next 5 days, with a total profit of 165.09%. performance
Source sentence
the model achieves a 61.15% positive predictive value and a 63.37% negative predictive value for the next 5 days, resulting in a total profit of 165.09%
- SMSFR method yields an absolute return of 165.09% from January 2022 to June 30, 2024, surpassing major A‑share market indices. performance
Source sentence
the SMSFR method achieved the optimal results, with an absolute return of 165.09% during the period from January 2022 to June 30, 2024, far surpassing the performance of the three major A-share market indices
- SMSFR-CNN outperforms other advanced methods in positive and negative predictive values. performance
Source sentence
Our proposed method, SMSFR-CNN, outperforms other advanced methods in terms of positive predictive value (PPV) and negative predictive value (NPV)
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 19 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 (51)
- Moving Averages. (2011)
- A deep learning based stock trading model with 2-D CNN trend detection (2017)
- Using deep learning neural networks and candlestick chart representation to predict stock market (2019) — link
- Learned Data-aware Image Representations of Line Charts for Similarity Search (2023)
- Encoding candlesticks as images for pattern classification using convolutional neural networks (2020)
- Can Stock Price Image Predict Future Return? Evidence from Chinese Stock Market (2022)
- (Re-) Imag (in) ing Price Trends (2023)
- Empirical asset pricing via machine learning (2020)
and 43 more
- Understanding the difficulty of training deep feedforward neural networks (2010)
- Batch normalization: Accelerating deep network training by reducing internal covariate shift (2015)
- Adam: A Method for Stochastic Optimization (2014)
- Dropout: a simple way to prevent neural networks from overfitting (2014)
- Analysis of integration of financial series classification and constrained portfolio optimization (2021)
- The real value of China’s stock market (2021)
- Deep residual learning for image recognition (2016)
- Quantitative trading on stock market based on deep reinforcement learning (2019)
- Forecasting the equity risk premium: the role of technical indicators (2014)
- ModAugNet: A new forecasting framework for stock market index value with an overfitting prevention LSTM module and a prediction LSTM module (2018)
- CNNpred: CNN-based stock market prediction using a diverse set of variables (2019)
- A CNN-BiLSTM-AM method for stock price prediction (2021)
- Stock Price Prediction Using Convolutional Neural Networks on a Multivariate Time Series (2020)
- Intraday prediction of Borsa Istanbul using convolutional neural networks and feature correlations (2017)
- A novel graph convolutional feature based convolutional neural network for stock trend prediction (2021)
- An analysis of decision making in the stock investment (2017)
- The cross section of expected stock returns in the Chinese A-share market (2007)
- Green Investing and Financial Education: The Case of China’s A-Share Market (2024)
- Implementation of Artificial Intelligence in Investment Decision-making in the Chinese A-share Market (2024)
- Machine learning approaches in stock market prediction: A systematic literature review (2023)
- Prediction And Analysis Using A Hybrid Model For Stock Market (2023)
- Forecasting of Indian stock market using time-series ARIMA model (2014)
- Comparison of Linear, Non-Linear and Frequency Domain Models in Forecasting Stock Returns of Emerging Markets. (2019)
- Predicting stock and stock price index movement using trend deterministic data preparation and machine learning techniques (2015)
- Stock market forecasting using machine learning algorithms (2012)
- A machine learning approach for stock price prediction (2014)
- Machine learning in the Chinese stock market (2022)
- Forecasting stock market prices using machine learning and deep learning models: A systematic review, performance analysis and discussion of implications (2023)
- McVCsB: A new hybrid deep learning network for stock index prediction (2023)
- Forecasting volatility of stock indices with ARCH model (2013)
- Modelling and estimation of volatility using ARCH/GARCH models in Jordan’s stock market (2016)
- Trader-Company Method: A Metaheuristics for Interpretable Stock Price Prediction (2021)
- TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting (2024)
- TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis (2023)
- Are transformers effective for time series forecasting ? (2023)
- A Time Series is Worth 64 Words: Long-term Forecasting with Transformers (2023)
- Predicting Stock Market Time-Series Data using CNN-LSTM Neural Network Model — → 2305.14378
- U-CNNpred: A Universal CNN-based Predictor for Stock Markets (2019) — link
- Forecasting stock prices with a feature fusion LSTM-CNN model using different representations of the same data (2019)
- Deep stock representation learning: From candlestick charts to investment decisions (2018)
- The short-selling hypothesis of weekend effect and $T+ 1$ Trading mechanism (2021)
- Stock prices prediction using deep learning models (2019) — link
- A parallel multi-module deep reinforcement learning algorithm for stock trading (2021)
1 of 51 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.