Paper: arXiv 2306.03620
Authors: Reza Nematirad, Amin Ahmadisharaf, Ali Lashgari
Abstract
The US stock market experienced instability following the recession (2007-2009). COVID-19 poses a significant challenge to US stock traders and investors. Traders and investors should keep up with the stock market. This is to mitigate risks and improve profits by using forecasting models that account for the effects of the pandemic. With consideration of the COVID-19 pandemic after the recession, two machine learning models, including Random Forest and LSTM are used to forecast two major US stock market indices. Data on historical prices after the big recession is used for developing machine learning models and forecasting index returns. To evaluate the model performance during training, cross-validation is used. Additionally, hyperparameter optimizing, regularization, such as dropouts and weight decays, and preprocessing improve the performances of Machine Learning techniques. Using high-accuracy machine learning techniques, traders and investors can forecast stock market behavior, stay ahead of their competition, and improve profitability. Keywords: COVID-19, LSTM, S&P500, Random Forest, Russell 2000, Forecasting, Machine Learning, Time Series JEL Code: C6, C8, G4.
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 (Random Forest, LSTM) to financial time series forecasting with moderate mathematical complexity, but it is highly empirical, focusing on data preprocessing, hyperparameter tuning, and backtest-ready methodology for trading signals.
Research Flowchart
flowchart TD
A["Research Goal: Forecast US Stock Indices<br>S&P500 & Russell 2000<br>During COVID-19"] --> B["Data Collection<br>Historical Prices Post-2008 Recession"]
B --> C["Preprocessing & Feature Engineering<br>Normalization, Technical Indicators, Split Data"]
C --> D{"Model Training & Optimization"}
D --> D1["Random Forest (RF)"]
D --> D2["LSTM Neural Network<br>Dropout & Weight Decay"]
D1 & D2 --> E["Model Evaluation<br>Cross-Validation & Performance Metrics"]
E --> F["Key Findings<br>LSTM outperforms RF<br>Validated for Risk Mitigation & Profitability"]
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 Random Forest and LSTM models to forecast the S&P 500 and Russell 2000 indices, incorporating COVID‑19 effects and hyperparameter tuning.
Source sentence
In this paper, we focus on the use of random forest and LSTM models to predict the behavior of the S&P 500 and Russell 2000 indices after the big recession.
Assumes
No explicit assumptions found in the text.
Data
- S&P 500 dataset (D1, D2, D3) market-prices unknown
Source sentence
The first, second, and third datasets are named D1, D2, D3 and are illustrated on Fig. 2.
- Russell 2000 dataset (D1, D2, D3) market-prices unknown
Source sentence
The results obtained from the random forest model for the RUT dataset are presented in table I as for different sample sizes (D1, D2, and D3).
Costs
Not applicable
Finds 1
- LSTM outperforms Random Forest for both indices across all sample sizes. performance
Source sentence
Furthermore, based on the results, LSTM outperforms random forest in all cases.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 11 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.