Paper: arXiv 2405.10584

Authors: Huiyu Li, Junhua Hu

Abstract

Stock price prediction has always been a difficult task for forecasters. Using cutting-edge deep learning techniques, stock price prediction based on investor sentiment extracted from online forums has become feasible. We propose a novel hybrid deep learning framework for predicting stock prices. The framework leverages the XLNET model to analyze the sentiment conveyed in user posts on online forums, combines these sentiments with the post popularity factor to compute daily group sentiments, and integrates this information with stock technical indicators into an improved BiLSTM-highway model for stock price prediction. Through a series of comparative experiments involving four stocks on the Chinese stock market, it is demonstrated that the hybrid framework effectively predicts stock prices. This study reveals the necessity of analyzing investors’ textual views for stock price prediction.

Complexity vs Empirical Score

  • Math Complexity: 6.0/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced deep learning models (XLNet, BiLSTM with highway mechanisms) requiring substantial mathematical understanding, but its primary focus is on a practical, data-heavy implementation with specific experimental details and stock data, making it geared toward application.

Research Flowchart

  flowchart TD
  A["Research Goal: Predict stock prices using investor sentiment from online forums"] --> B["Data Collection"]
  
  subgraph B ["Data Collection"]
      B1["Stock Price Data"]
      B2["Online Forum Posts"]
      B3["Technical Indicators"]
  end

  B --> C["Sentiment Analysis Pipeline"]
  
  subgraph C ["Sentiment Analysis Pipeline"]
      C1["XLNET Model"]
      C2["Post Popularity Factor"]
      C3["Daily Group Sentiment Calculation"]
  end

  C --> D["Hybrid Deep Learning Framework"]
  
  subgraph D ["Hybrid Deep Learning Framework"]
      D1["Input Layer<br>Sentiment + Technical Indicators"]
      D2["Improved BiLSTM-Highway Model"]
  end

  D --> E["Prediction & Validation"]
  
  subgraph E ["Prediction & Validation"]
      E1["Stock Price Prediction"]
      E2["Comparative Experiments<br>4 Chinese Stocks"]
  end

  E --> F["Key Findings"]
  
  subgraph F ["Key Findings"]
      F1["Effective stock price prediction achieved"]
      F2["Sentiment analysis crucial for forecasting"]
      F3["Hybrid framework outperforms baselines"]
  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 hybrid deep learning framework called XSI-BiLSTM that combines XLNET‑derived sentiment, post popularity, and technical indicators in an improved BiLSTM‑highway model for stock price prediction.

Source sentence
We propose a novel hybrid deep learning framework for predicting stock prices.

Assumes

No explicit assumptions found in the text.

Data

  • GitHub community sentiment datasets · GitHub text-news public
    Source sentence
    We collected four relevant datasets from the GitHub community1 to form the first text data type.
  • Four Chinese stocks market-prices unknown
    four stocks on the Chinese stock market
    Source sentence
    Through a series of comparative experiments involving four stocks on the Chinese stock market, it is demonstrated that the hybrid framework effectively predicts stock prices.

Costs

Not applicable The paper reports no trading or execution result.

Finds 3

  • XSI‑BiLSTM outperforms other prediction models on most performance metrics, with lower absolute prediction errors across the four stocks. performance
    Source sentence
    The prediction results of XSI‑BiLSTM outperform those of the other prediction models in most of the performance metrics, and the actual stock closing prices differ less from the predicted stock closing prices.
  • Incorporating investor sentiment and the highway mechanism improves prediction accuracy, especially for short input windows. predictive-accuracy
    Source sentence
    The model's ability to mine the data is enhanced by adding investor sentiment and highway methods, which help extract time-series features.
  • The average error of XSI‑BiLSTM is the lowest among all four stocks, indicating superior forecasting performance. statistical
    Source sentence
    the forecast error decreases overall, as reflected by the fact that its average error (i.e., the black square in the figure) is the lowest among all four stocks.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 16 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.