Paper: arXiv 2407.01572
Authors: Jaydip Sen, Hetvi Waghela, Sneha Rakshit
Abstract
This paper explores using a deep learning Long Short-Term Memory (LSTM) model for accurate stock price prediction and its implications for portfolio design. Despite the efficient market hypothesis suggesting that predicting stock prices is impossible, recent research has shown the potential of advanced algorithms and predictive models. The study builds upon existing literature on stock price prediction methods, emphasizing the shift toward machine learning and deep learning approaches. Using historical stock prices of 180 stocks across 18 sectors listed on the NSE, India, the LSTM model predicts future prices. These predictions guide buy/sell decisions for each stock and analyze sector profitability. The study’s main contributions are threefold: introducing an optimized LSTM model for robust portfolio design, utilizing LSTM predictions for buy/sell transactions, and insights into sector profitability and volatility. Results demonstrate the efficacy of the LSTM model in accurately predicting stock prices and informing investment decisions. By comparing sector profitability and prediction accuracy, the work provides valuable insights into the dynamics of the current financial markets in India.
Complexity vs Empirical Score
- Math Complexity: 4.0/10
- Empirical Rigor: 6.0/10 (capped from 7.0: transaction costs not modelled; see penalty flags)
- Quadrant: Street Traders — empirical and applied, lower math complexity (effective empirical rigor capped from 7.0)
Why this score: The paper uses established deep learning architectures (LSTM) with minimal novel theoretical derivations, placing it on the lower end of math complexity. However, it demonstrates high empirical rigor by detailing data acquisition, model training, and backtesting results on 180 stocks across Indian sectors, with concrete profitability metrics.
Research Flowchart
flowchart TD
A["Research Goal<br>Optimize Portfolio Design via<br>Accurate Stock Price Prediction"] --> B["Data Collection<br>Historical Prices of 180 Stocks<br>Across 18 NSE Sectors"]
B --> C["Methodology<br>Build Optimized LSTM<br>Deep Learning Model"]
C --> D["Computational Process<br>Predict Future Stock Prices"]
D --> E{"Decision Logic"}
E -->|"Predicted Increase"| F["Buy Signal"]
E -->|"Predicted Decrease"| G["Sell Signal"]
F & G --> H["Outcome & Analysis<br>Sector Profitability &<br>Volatility Insights"]
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 introduces an optimized LSTM model for robust portfolio design and uses its predictions for buy/sell transactions across 180 Indian stocks.
Source sentence
introducing an optimized LSTM model for robust portfolio design
Assumes 4
- A fictional investor trades every day, buying if the next‑day forecast is higher and selling otherwise. execution
Source sentence
In this study, it is assumed that a fictional investor engages in mandatory trading every day. If the forecasted price for the next day is higher than the current day’s price, the investor must buy the stock. Conversely, if the forecasted price is lower or equal to the current day’s price, the investor must sell the st
- Buy/sell profits are computed assuming the trading of a single equity share. execution
Source sentence
Buy/sell profits are computed assuming the trading of a single equity share.
- Returns for stocks and sectors are presented as percentage figures and are independent of the number of shares traded. execution
Source sentence
the returns for stocks and sectors are independent of the number of equity shares traded, as they are presented as percentage figures.
- Historical price data are retrieved from Yahoo Finance using ticker names, start date Jan 1 2005 and end date Apr 23 2024. data
Source sentence
The start date for all stocks is set as January 1, 2005, and the end date as April 23, 2024.
Data
- Historical stock prices of 180 NSE stocks market-prices public
Source sentence
Using historical stock prices of 180 stocks across 18 sectors listed on the NSE, India
Costs
Not modeled
Finds 3
- The LSTM model achieves high directional accuracy, with the lowest accuracy score of 0.958315 (95.83% correct predictions). predictive-accuracy
Source sentence
the lowest value of the accuracy score is found to be 0.958315. In other words, in the worst case, the model correctly predicted the direction of movement of the price (upward or downward) of the stock the next day in 95.83% of cases.
- Sector profitability index shows ESG as the most profitable sector and Realty as the least profitable. performance
Source sentence
It is observed that ESG, Services, Pharma, and FMCG are the fur sectors that have yielded the highest profitability index. On the other hand, Realty, Media, PSU Banks, and Energy, are the four least profitable sectors.
- Across sectors, the LSTM model’s Huber loss, MAE and accuracy scores are generally low loss and high accuracy, e.g., Banking HL 0.000077, MAE 14.56, Acc 0.980422. performance
Source sentence
Table 20. The average performance level of the LSTM model for the sectors (Period: January 1, 2005 – April 23, 2024)
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 24 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.