Paper: arXiv 2611.23057

Authors: Ahmet Akusta

Abstract

This study investigates the integration of sentiment analysis with machine learning models to forecast stock price movements using the Nvidia Corporation as a case study. Sentiment scores were derived from Nvidia-related financial news headlines using two sentiment analysis tools: FinBERT, a domain-specific tool, and TextBlob, a general-purpose tool. These scores were integrated into predictive frameworks based on bidirectional long short-term memory (Bi-LSTM) networks and multilayer perceptrons (MLPs) developed alongside historical stock price data. This study assesses the predictive performance over the entire observation period and across distinct market phases: bullish, stagnation, and strong bullish conditions. The findings indicate that sentiment features enhance predictive accuracy in specific contexts, particularly during stagnation phases, with TextBlob demonstrating superior performance to FinBERT in specific scenarios. In addition, Bi-LSTM models exhibit consistently superior performance in capturing temporal dependencies compared to MLPs. However, the impact of sentiment features diminished during strongly directional trends, such as those observed in strong bullish markets. The combination of FinBERT and TextBlob in the same dataset allows for a dual-perspective approach to sentiment analysis, thereby providing new insights into the dynamic relationship between market sentiment and stock price behavior. This research contributes to the existing literature on applying sentiment analysis to financial forecasting by advancing the integration of complementary sentiment tools and phase-specific evaluations.

Complexity vs Empirical Score

  • Math Complexity: 4.0/10
  • Empirical Rigor: 6.0/10
  • Quadrant: Street Traders — empirical and applied, lower math complexity

Why this score: The paper applies established machine learning models (Bi-LSTM, MLP) and sentiment analysis tools (FinBERT, TextBlob) to a specific financial forecasting problem. While the mathematical underpinnings of these models are complex, the paper itself does not delve into novel mathematical derivations. The empirical setup is reasonably rigorous, involving comparative studies across market phases and different sentiment tools.

Research Flowchart

  flowchart TD
    A["Research Goal: Integrate Market Sentiments for Stock Price Prediction"] --> B{"Methodology: Comparative Study"};
    B --> C["Data/Inputs: Nvidia Stock Prices + Financial News Headlines"];
    C --> D{"Sentiment Analysis: FinBERT & TextBlob"};
    D --> E{"ML Models: Bi-LSTM & MLP"};
    E --> F["Performance Evaluation: Overall & Market Phases (Bullish, Stagnation, Strong Bullish)"];
    F --> G["Key Findings: Sentiment enhances accuracy in stagnation; TextBlob superior to FinBERT; Bi-LSTM consistently outperforms MLP; Sentiment impact diminishes in strong trends."];

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 integrating sentiment scores from financial news headlines, derived using FinBERT and TextBlob, into predictive frameworks based on bidirectional LSTM and multilayer perceptron models for Nvidia stock price forecasting.

Source sentence
This study investigates the integration of sentiment analysis with machine learning models to forecast stock price movements using the Nvidia Corporation as a case study. These scores were integrated into predictive frameworks based on bidirectional long short-term memory (Bi-LSTM) networks and multilayer perceptrons (MLPs) developed alongside historical stock price data.

Assumes 6

  • An 80/20 train‑test split is used for model evaluation. data
    Source sentence
    The train,test split was performed using an 80/20 ratio.
  • Shuffling is applied to the time‑series data while preserving temporal relationships. model
    Source sentence
    Because the data is a time series dataset, shuffling was applied to ensure randomness while preserving the integrity of the temporal relationships within the dataset.
  • Market conditions can be categorized into bullish, stagnation, and strong bullish phases for phase‑specific analysis. market
    Source sentence
    This study assesses the predictive performance over the entire observation period and across distinct market phases: bullish, stagnation, and strong bullish conditions. Akusta, 2025 Bi,LSTM (Market Phases): Bi,LSTM models are independently trained on data from each defined phase to investigate the impact of different market conditions.
  • Sentiment scores derived from news headlines are relevant predictors of stock price movements. statistical
    Source sentence
    Feature engineering In this study, we derived sentiment scores from financial news headlines to integrate market sentiments into Nvidia’s stock price predictions. Sentiment analysis of the collected news headlines was conducted using two distinct approaches: TextBlob and FinBERT.
Show 2 more
  • Historical price data and sentiment features are sufficient inputs for the Bi‑LSTM and MLP models without additional macro or order‑book information. data
    Source sentence
    The first dataset, obtained from Yahoo Finance, comprises 390 daily historical stock price records for Nvidia Corporation, encompassing metrics, such as opening, closing, highest, and lowest prices, and trading volume. The second dataset includes 9,400 news headlines about Nvidia during the same timeframe, measuring market sentiment toward the company.
  • The models are evaluated on a single stock (Nvidia) and results are assumed to reflect the effect of sentiment integration for that asset. other
    Source sentence
    This study investigates the integration of sentiment analysis with machine learning models to forecast stock price movements using the Nvidia Corporation as a case study. The first dataset, obtained from Yahoo Finance, comprises 390 daily historical stock price records for Nvidia Corporation, encompassing metrics, such as opening, closing, highest, and lowest prices, and trading volume.

Data

  • Nvidia daily historical stock price records · Yahoo Finance market-prices public
    Nvidia Corporation
    Source sentence
    The first dataset, obtained from Yahoo Finance, comprises 390 daily historical stock price records for Nvidia Corporation, encompassing metrics, such as opening, closing, highest, and lowest prices, and trading volume.
  • Nvidia news headlines · Nasdaq.com text-news public
    Nvidia Corporation
    Source sentence
    The second dataset includes 9,400 news headlines about Nvidia during the same timeframe, measuring market sentiment toward the company. These data were obtained from Nasdaq.com.

Recognised sources: Yahoo Finance

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • Sentiment features improve predictive accuracy during stagnation phases, with TextBlob outperforming FinBERT in those contexts. predictive-accuracy
    Source sentence
    The findings indicate that sentiment features enhance predictive accuracy in specific contexts, particularly during stagnation phases, with TextBlob demonstrating superior performance to FinBERT in specific scenarios. In contrast, sentiment features gained prominence during low,volatility or stagnation phases, indicating the context,dependent value of both specialized (FinBERT) and general,purpose (TextBlob)
  • Bi‑LSTM models consistently achieve superior performance over MLPs in capturing temporal dependencies. performance
    Source sentence
    In addition, Bi-LSTM models exhibit consistently superior performance in capturing temporal dependencies compared to MLPs. In this study, Bi,LSTM models generally outperformed MLPs in terms of capturing temporal dependencies and sequential patterns, particularly during the Stagnation Phase and the overall period.
  • TextBlob sentiment scores yield the lowest MAE, MSE, and RMSE among all model configurations. performance
    Source sentence
    TextBlob scores led to the lowest MAE, MSE, and RMSE, along with a higher R² Score.
  • Feature importance analysis shows that the daily closing price is the dominant predictor across all market conditions. statistical
    Source sentence
    Additionally, the feature importance findings reveal that “Daily Close” consistently remains the single most influential predictor, especially when market momentum is strong. Feature importance analysis revealed that “Daily Close” remained the dominant predictor across all market condi, tions, particularly during strong directional trends.
  • A large train‑test MAE gap (e.g., Model 3: 1.60 vs. 30.19) indicates potential overfitting, though test R² remains high (0.93). negative
    Source sentence
    The substantial discrepancy between the train and test MAE (e.g., Model 3: 1.60 vs. 30.19) raises concerns about potential overfitting; however, the maintained high test R² (0.93) suggests that while generalization may be limited, the model retains explanatory power on unseen data.

Checked by others

Supported by

  • FNSPID: A Comprehensive Financial News Dataset in Time Series (2024)
    On the question: Does incorporating sentiment scores improve stock price prediction performance?
    Compare the findings

    This paper: TextBlob sentiment scores yield the lowest MAE, MSE, and RMSE among all model configurations.

    That paper: Adding sentiment scores modestly enhances performance on the transformer-based model.

    Finding A reports a modest performance boost for a transformer model, and Finding B shows TextBlob sentiment yields the lowest MAE, MSE, and RMSE, supporting A.

Code

No repository by the authors found in the paper.

Extracted 2026-11-18 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (27)

  1. Ambiguities inherent in sums-of-squares-based error statistics (2008) — link
  2. Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance (2005) — link
  3. Stock Trend Prediction Using News Sentiment Analysis (2016) — link
  4. FinBERT: Financial Sentiment Analysis with Pre-trained Language Models (2019) — link
  5. Augmented Textual Features-Based Stock Market Prediction (2020) — link
  6. FinBERT: A Pretrained Language Model for Financial Communications (2020) — link
  7. Machine learning sentiment analysis, COVID-19 news and stock market reactions (2023) — link
  8. Opinion Mining for the Customer Feedback using TextBlob (2020) — link
and 19 more
  1. Financial Context News Sentiment Analysis for the Lithuanian Language (2021) — link
  2. A Comprehensive Review of Investor Sentiment Analysis in Stock Price Forecasting (2021) — link
  3. Performance Evaluation of Recurrent Neural Networks Applied to Indoor Camera Localization (2022) — link
  4. Synergy between stock prices and investor sentiment in social media (2022) — link
  5. A Sentiment Analysis Method Based on Bidirectional Long Short-Term Memory Networks (2022) — link
  6. Decomposition of the mean absolute error (MAE) into systematic and unsystematic components (2023) — link
  7. Stock trend prediction using sentiment analysis (2023) — link
  8. Predicting Exchange Rate with FinBERT-Based Sentiment Analysis of Online News (2022) — link
  9. Research on stock price prediction using TextRank based text summarization technology and sentiment analysis (2022) — link
  10. Prediction and Sentiment Analysis of Stock using Machine Learning (2023) — link
  11. Stock Prediction by Polyglot Sentiment Analysis on Twitter (2023) — link
  12. Deep learning approaches for MIMO time-series analysis (2023) — link
  13. Domain-Specific Sentiment Analysis: An Optimized Deep Learning Approach for the Financial Markets (2023) — link
  14. Short‐term electricity load forecasting based on improved sparrow search algorithm with optimized BiLSTM (2023) — link
  15. PRIME: Posterior Reconstruction of the Input for Model Explanations (2023) — link
  16. Predicting Price Trends Using Sentiment Analysis: A Study of StepN’s SocialFi and GameFi Cryptocurrencies (2023) — link
  17. FinBERT-FOMC: Fine-Tuned FinBERT Model with Sentiment Focus Method for Enhancing Sentiment Analysis of FOMC Minutes (2023) — link
  18. Predictive Learning and Career Path Using Artificial Intelligence (2024) — link
  19. SUPPORT FOR STOCK TREND PREDICTION USING TRANSFORMERS AND SENTIMENT ANALYSIS — → 2305.14368

1 of 27 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.