Paper: arXiv 2406.02604
Authors: Bivas Dinda
Abstract
The recent advancement of deep learning architectures, neural networks, and the combination of abundant financial data and powerful computers are transforming finance, leading us to develop an advanced method for predicting future stock prices. However, the accessibility of investment and trading at everyone’s fingertips made the stock markets increasingly intricate and prone to volatility. The increased complexity and volatility of the stock market have driven demand for more models, which would effectively capture high volatility and non-linear behavior of the different stock prices. This study explored gated recurrent neural network (GRNN) algorithms such as LSTM (long short-term memory), GRU (gated recurrent unit), and hybrid models like GRU-LSTM, LSTM-GRU, with Tree-structured Parzen Estimator (TPE) Bayesian optimization for hyperparameter optimization (TPE-GRNN). The aim is to improve the prediction accuracy of the next day’s closing price of the NIFTY 50 index, a prominent Indian stock market index, using TPE-GRNN. A combination of eight influential factors is carefully chosen from fundamental stock data, technical indicators, crude oil price, and macroeconomic data to train the models for capturing the changes in the price of the index with the factors of the broader economy. Single-layer and multi-layer TPE-GRNN models have been developed. The models’ performance is evaluated using standard matrices like R2, MAPE, and RMSE. The analysis of models’ performance reveals the impact of feature selection and hyperparameter optimization (HPO) in enhancing stock index price prediction accuracy. The results show that the MAPE of our proposed TPE-LSTM method is the lowest (best) with respect to all the previous models for stock index price prediction.
Complexity vs Empirical Score
- Math Complexity: 4.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Street Traders — empirical and applied, lower math complexity
Why this score: The paper employs standard deep learning architectures (LSTM, GRU) with Bayesian optimization, involving moderate mathematical complexity focused on model structure and hyperparameter tuning, but demonstrates strong empirical rigor through detailed data sourcing, preprocessing, and performance evaluation on real market data.
Research Flowchart
flowchart TD A["Research Goal: Enhance stock index prediction accuracy for NIFTY 50"] --> B["Data Preparation & Feature Selection"] B --> C["Developing TPE-GRNN Models <br> (LSTM, GRU, GRU-LSTM, LSTM-GRU)"] C --> D["Applying TPE Bayesian Optimization for Hyperparameter Tuning"] D --> E["Model Evaluation <br> Metrics: R2, MAPE, RMSE"] E --> F["Key Finding: TPE-LSTM achieves the lowest MAPE <br> (Highest accuracy among all tested models)"]
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 gated recurrent neural network models optimized with Tree‑structured Parzen Estimator Bayesian optimization (TPE‑GRNN) to predict the next day’s closing price of the NIFTY 50 index.
Source sentence
The aim is to improve the prediction accuracy of the next day's closing price of the NIFTY 50 index, a prominent Indian stock market index, using TPE‑GRNN.
Assumes 5
- The selected eight macro‑economic, fundamental, technical and crude‑oil features sufficiently capture the drivers of the NIFTY 50 index price. data
Source sentence
A combination of eight influential factors is carefully chosen from fundamental stock data, technical indicators, crude oil price, and macroeconomic data to train the models
- Future daily closing price can be predicted from past data without accounting for market microstructure effects. market
Source sentence
This study focuses on a computational framework that aims to predict the next day's closing price of the NIFTY 50 index using LSTM, GRU, GRU‑LSTM, and LSTM‑GRU models.
- The historical data collected from Yahoo Finance and Investing.com are accurate and free of errors after cleaning. data
Source sentence
The dataset is cleaned after carefully selecting features and collecting data from Yahoo Finance, and Investing.com.
- Hyperparameter optimization with TPE Bayesian optimization yields the best configuration for each GRNN architecture. model
Source sentence
TPE Bayesian optimization for the HPO technique is now applied. Once the optimum hyperparameters are found, the normalized dataset is executed in GRNN models with optimum hyperparameters.
Show 1 more
- Model performance can be evaluated solely with statistical error metrics (R2, MAPE, RMSE) without incorporating transaction costs. execution
Source sentence
The prediction accuracy of the proposed models is evaluated using the error metrics of
Data
- NIFTY 50 index price data market-prices public
Source sentence
The dataset is cleaned after carefully selecting features and collecting data from Yahoo Finance, and Investing.com.
- Macro‑economic and crude oil data macro public
Source sentence
A combination of eight influential factors is carefully chosen from fundamental stock data, technical indicators, crude oil price, and macroeconomic data
Costs
Unclear - sources disagree
Source sentence (the authors' words)
Investors can use this model as an extra tool to make better trading decisions. However, the trading
Finds 2
- The single‑layer TPE‑LSTM model achieved MAPE = 0.0137 %, R2 = 0.9645 and RMSE(ND) = 0.0154, outperforming all other tested GRNN variants. predictive-accuracy
Source sentence
It is found that the single-layer TPE‑LSTM model (MAPE= 0.0137%, R2 =0.9645, and RMSE(ND)=0.0154) provides superior prediction accuracy among single and multilayer TPE‑GRNN models.
- The proposed TPE‑LSTM method yields the lowest MAPE among all compared methods for stock index prediction. performance
Source sentence
The results show that the MAPE of our proposed TPE‑LSTM method is the lowest (best) with respect to all the previous models for stock index price prediction.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 10 related papers). Coverage is partial, so this is neither replication nor refutation.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.