Paper: arXiv 2308.08549
Authors: S. Srinivas, R. Gadela, R. Sabu, A. Das, G. Nath, V. Datla
Abstract
Predicting future prices of a stock is an arduous task to perform. However, incorporating additional elements can significantly improve our predictions, rather than relying solely on a stock’s historical price data to forecast its future price. Studies have demonstrated that investor sentiment, which is impacted by daily news about the company, can have a significant impact on stock price swings. There are numerous sources from which we can get this information, but they are cluttered with a lot of noise, making it difficult to accurately extract the sentiments from them. Hence the focus of our research is to design an efficient system to capture the sentiments from the news about the NITY50 stocks and investigate how much the financial news sentiment of these stocks are affecting their prices over a period of time. This paper presents a robust data collection and preprocessing framework to create a news database for a timeline of around 3.7 years, consisting of almost half a million news articles. We also capture the stock price information for this timeline and create multiple time series data, that include the sentiment scores from various sections of the article, calculated using different sentiment libraries. Based on this, we fit several LSTM models to forecast the stock prices, with and without using the sentiment scores as features and compare their performances.
Complexity vs Empirical Score
- Math Complexity: 6.5/10
- Empirical Rigor: 7.5/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper uses advanced deep learning (LSTM) and multiple sentiment scoring formulas, but focuses heavily on a concrete, data-heavy implementation framework with scraping, preprocessing, and model validation on a large dataset.
Research Flowchart
flowchart TD A["Research Goal<br>Does news sentiment improve stock price<br>forecasting beyond historical data alone?"] --> B["Data Collection & Preprocessing"] B --> C["Data Inputs<br>1. ~500k Financial News Articles<br>2. Nifty50 Stock Prices<br>3. 3.7-Year Timeline"] C --> D["Computational Processes<br>1. Extract sentiment scores from articles<br>2. Generate time series data<br>3. Fit LSTM models<br>4. Compare performance"] D --> E["Model Variants<br>1. LSTM without sentiment features<br>2. LSTM with sentiment features"] E --> F["Key Findings<br>1. Sentiment significantly impacts price volatility<br>2. Multi-feature models outperform historical-only models<br>3. Efficient framework for sentiment extraction established"]
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 sentiment scores extracted from daily news as additional features in LSTM models to improve stock price forecasting for NIFTY50 stocks.
Source sentence
Based on this, we fit several LSTM models to forecast the stock prices, with and without using the sentiment scores as features and compare their performances.
Assumes 3
- Stock price movements can be influenced by investor sentiment. market
Source sentence
Based on the assumption that stock price movements can be influenced by investor sentiment, the current study seeks to investigate how financial news-based sentiment can improve stock price prediction in the Indian market.
- Economics Times provides representative daily national business news for Indian stocks. data
Source sentence
We believe that the most appropriate choice for a national newspaper is Economics Times (ET). ET is India’s largest business daily that publishes more than 300 news articles daily.
- Investors are sentimental and emotionally driven, contrary to rational homogeneous assumptions. market
Source sentence
Modern financial theories assume investors are homogeneous and rational, which is erroneous. Behavioural finance, in contrast, explores investor behaviour in the stock market and its possible effects on stock price movements.
Data
- NIFTY50 news database text-news public
Source sentence
This paper presents a robust data collection and preprocessing framework to create a news database for a timeline of around 3.7 years, consisting of almost half a million news articles.
- NIFTY50 stock price series market-prices unknown
Source sentence
We also capture the stock price information for this timeline and create multiple time series data, that include the sentiment scores from various sections of the article, calculated using different sentiment libraries.
Costs
Not applicable
Finds 4
- For at least eighty percent of the NIFTY50 stocks, LSTM models that include sentiment scores outperform models without sentiment features. performance
Source sentence
we have clearly seen that, for at least eighty percent of the NIFTY50 stocks, these models outperform the models without any sentiment scores.
- Sentiment scores derived from the full-text of articles give the best prediction performance across stocks. predictive-accuracy
Source sentence
Among the sections of the article, we found that sentiment scores from full-text to give the best predictions
- Harvard IV4 (HIV4) sentiment library yields the best overall performance (14 stocks), closely followed by VADER (13 stocks) and Loughran‑McDonald (13 stocks). performance
Source sentence
Comparing only the variants with sentiment scores, across the three libraries, we could see that, for 14 of the stocks, variants with HIV4 sentiment score columns outperformed other variants, Vader for 13 of the stocks and LM for 13 of the stocks.
- Variants using full-text sentiment scores outperform other variants for 12 stocks; using both heading and synopsis scores also for 12 stocks; heading only for 9 stocks; synopsis only for 7 stocks. performance
Source sentence
the variant that uses both Heading and Synopsis sentiment scores as features, outperformed other variants for 12 of the stocks. The variant using only Heading sentiment score as a feature, outperformed other variants for 9 of the stocks, while the v
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 14 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
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.