Paper: arXiv 2402.06689

Authors: Himanshu Gupta, Aditya Jaiswal

Abstract

Predicting a fast and accurate model for stock price forecasting is been a challenging task and this is an active area of research where it is yet to be found which is the best way to forecast the stock price. Machine learning, deep learning and statistical analysis techniques are used here to get the accurate result so the investors can see the future trend and maximize the return of investment in stock trading. This paper will review many deep learning algorithms for stock price forecasting. We use a record of s&p 500 index data for training and testing. The survey motive is to check various deep learning and statistical model techniques for stock price forecasting that are Moving Averages, ARIMA which are statistical techniques and LSTM, RNN, CNN, and FULL CNN which are deep learning models. It will discuss various models, including the Auto regression integration moving average model, the Recurrent neural network model, the long short-term model which is the type of RNN used for long dependency for data, the convolutional neural network model, and the full convolutional neural network model, in terms of error calculation or percentage of accuracy that how much it is accurate which measures by the function like Root mean square error, mean absolute error, mean squared error. The model can be used to predict the stock price by checking the low MAE value as lower the MAE value the difference between the predicting and the actual value will be less and this model will predict the price more accurately than other models.

Complexity vs Empirical Score

  • Math Complexity: 3.5/10
  • Empirical Rigor: 4.0/10
  • Quadrant: Philosophers — conceptual discussion, lower math and empirical scores

Why this score: The paper primarily reviews and applies standard deep learning and statistical models (ARIMA, LSTM, RNN, CNN) with a focus on common error metrics (RMSE, MAE), showing limited advanced mathematical derivations or novel statistical theory. While it uses a specific S&P 500 dataset and describes an experimental setup, the methodology is introductory, lacks detailed backtesting protocols, code, or robust implementation specifics, and serves more as a conceptual review than a backtest-ready empirical study.

Research Flowchart

  flowchart TD
  A["Research Goal: Determine the best model<br>for accurate stock price forecasting"] --> B["Input Data: S&P 500 Index Time Series"]
  B --> C["Methodology: Compare Models"]
  C --> D["Statistical Models<br>ARIMA, Moving Averages"]
  C --> E["Deep Learning Models<br>LSTM, RNN, CNN, Full CNN"]
  D --> F["Model Evaluation<br>Metrics: RMSE, MAE, MSE"]
  E --> F
  F --> G{"Outcome: Model Selection"}
  G --> H["Best Model: Lowest MAE<br>indicates highest prediction accuracy"]

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 to evaluate and compare several deep learning and statistical models for forecasting S&P 500 index closing prices, aiming to identify the most accurate approach.

Source sentence
Our main motive is to build an accurate predictive framework. We have used a spy 500 ETF data for building and testing the model.

Assumes 3

  • The time series data must be made stationary before applying ARIMA. data
    Source sentence
    to make it stationary we calculate the difference.
  • Only the closing price of each stock is used for forecasting. data
    Source sentence
    In this survey we have taken the closed price for each stock.
  • The ARIMA model requires the Augmented Dickey‑Fuller test to confirm stationarity. model
    Source sentence
    Augmented Dickey-Fuller and the Null Hypothesis Test is used to check for data non-stationarity so if the NH fails, there is a probability that the data will be stationary

Data

  • SPY 500 ETF data · Yahoo Finance market-prices public
    Dec 2015 to Jan 2018 · S&P 500 constituents
    Source sentence
    We have used a spy 500 ETF data for building and testing the model. This data has been obtained from the yahoo finance.

Recognised sources: Yahoo Finance

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • The LSTM model with a 30‑day window achieved the lowest mean absolute error (MAE) of 1.18, outperforming all other models. predictive-accuracy
    Source sentence
    The MAE value of 30‑day window size is 1.18 which is lower than any other model which we have used till now.
  • The Full CNN model achieved an MAE of 7.98, better than the pre‑process CNN (MAE 12.65). predictive-accuracy
    Source sentence
    the result of preprocess CNN that we got is 12.65 MAE value and for the Full CNN the MAE value is 7.98 which is much lower than the pre process CNN.
  • The ARIMA model produced an MAE of 2.8. predictive-accuracy
    Source sentence
    the MAE value of Arima model is 2.8.
  • The RNN seq‑to‑seq configuration achieved an MAE of 4.19, better than the seq‑to‑vector configuration (MAE 23.07). predictive-accuracy
    Source sentence
    seq-to-vector have the MAE value 23.07 which is higher than the MAE value of
  • The naive forecast and moving‑average baselines yielded higher MAE values than the LSTM model. performance
    Source sentence
    ... the MAE value of 20 day came higher than the 5 day moving average this means that the price is a recent up trend

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 28 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.