Paper: arXiv 2307.07868
Authors: Varun Sangwan, Vishesh Kumar Singh, Bibin Christopher
Abstract
Our research aims to find the best model that uses companies projections and sector performances and how the given company fares accordingly to correctly predict equity share prices for both short and long term goals.
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced LSTM architectures (Bidirectional, Seq2Seq, Two-Path) and sentiment analysis, indicating high mathematical complexity. It demonstrates strong empirical rigor through backtesting on real historical data (12 months) for multiple stocks, comparing multiple model architectures and including quantitative error metrics (RMSE, MAE).
Research Flowchart
flowchart TD
A["Research Goal: Find optimal LSTM model for equity price prediction using sector performance & sentiment"] --> B{"Data Collection & Processing"}
B --> C["Stock Price Data<br>Historical Prices"]
B --> D["Sector & Company Fundamentals<br>Projections & Performance"]
B --> E["Sentiment Data<br>News & Social Media"]
C & D & E --> F{"Computational Models"}
F --> G["LSTM Variants<br>GRU, BiLSTM, Stacked LSTM"]
F --> H["Feature Engineering<br>Technical Indicators"]
G & H --> I["Model Training & Evaluation"]
I --> J["Key Findings<br>Optimal Model Selected<br>Sentiment significantly improves accuracy<br>Short vs Long-term prediction variance"]
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 that the LSTM 2-Path model is the best performing algorithm for stock price prediction.
Source sentence
Our research pointed to the conclusion that the LSTM 2-Path model was the best performing algorithm followed by Bidirectional LSTM and SEQ2SEQ LSTM.
Assumes 2
- A sector decline relative to projections implies a future price correction. market
Source sentence
The existing share of a stock; assuming Reliance declines in the Oil Sector as compared to its projections signalling a down quarter, but its price value has not fluctuated much, leading to the assumption that a correction is in order which will result in large volumes of shares being sold and the stock price taking a
- The model is better suited for short‑term predictions than long‑term horizons. model
Source sentence
suggesting that the model may be better suited for short-term predictions
Data
No dataset named in the text.
Costs
Not applicable
Finds 5
- The LSTM 2-Path model achieved the lowest MSE (0.00035) and RMSE (0.019) among all tested models. performance
Source sentence
The LSTM 2-Path model had the lowest MSE and RMSE scores, with values of 0.00035 and 0.019 respectively.
- Bidirectional LSTM and SEQ2SEQ LSTM had higher MSE and RMSE (0.00049/0.022 and 0.00056/0.023). performance
Source sentence
The Bidirectional LSTM and SEQ2SEQ LSTM models had MSE and RMSE scores of 0.00049 and 0.022, and 0.00056 and 0.023, respectively.
- The LSTM with sentiment analysis performed worst with MSE 0.00208 and RMSE 0.046. performance
Source sentence
The LSTM with sentiment analysis had the highest MSE and RMSE scores, with values of 0.00208 and 0.046, respectively.
- Performance differences between LSTM 2-Path and other models were statistically significant (p < 0.01). statistical
Source sentence
We also found that the performance differences between the LSTM 2-Path model and the other models were statistically significant with a p-value of less than 0.01.
- Model performance improves with longer input sequences but degrades as prediction horizon increases. other
Source sentence
We found that the model's performance improved with longer input sequences, but the improvement leveled off after a certain point. We also learned that the model's performance decreased as the prediction horizon increased
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 32 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.