Forecasting India S&P BSE SENSEX and USA S&P-500 Using ARIMA and Prophet

1. Introduction

• The goal of Time Series Analysis and Forecasting is to deal with the class of data science problems where the primary values of interest are a sequence of data points measured over a period of time or recorded over regular intervals of time or which are equally spaced in time and we want to forecast either a single measurement (univariate) or multiple measurements (multi-variate) using the time-series models.

• This study forecasts the index close values and volatility dynamics of S&P BSE SENSEX of Bombay Stock Exchange (BSE) and S&P-500 of New York Stock Exchange (NYSE).

• To achieve the objectives, the study uses descriptive statistics, statistical tests including Augmented Dickey-Fuller for checking the stationarity of the underlying time series data before modeling. The developed models forecasts daily close values for the S&P BSE SENSEX and S&P-500 financial time-series data using the ARIMA model and a suite of time series models provided by a Prophet library developed by Facebook especially for forecasting time series data in the Python programming language.

• The project is inspired by the Kaggle Notebook titled A Modern Time Series Tutorial by Rohan Rao.

2. Datasets

• The financial time series datasets are collected using either the libraries such as quandl, investpy or by Yahoo Finance website.

Sr. No	Dataset Name	Time Period	Source of Dataset
1	S&P BSE SENSEX	2000-2020	Refer sensex_data_2001_2020.csv file in ./Datasets/ directory
2	S&P-500	2000-2020	Refer S&P-500-Price-Action-Historical-Prices-Dataset.csv file in ./Datasets/ directory
3	India VIX	2008-2020	Refer india_vix_data_2008_2020.csv file in ./Datasets/ directory
4	CBOE VIX	1990-2020	Refer cboe_data_1990_2020.csv file in the ./Datasets/ directory

• Although the *.csv files present in the ./Datasets/ directory consists of the corresponding datasets, those ./Datasets/*.csv files are exported from pandas dataframes which are created as a result of accessing the data from quandle, yahoo, and investpy websites/libraries using DataFrame.to_csv() method.

• investpy is a really good library which most people don't know about and it provides financial time series data for almost all the indexed financial products/assets which belong to different asset classes which are covered by https://www.investing.com/ website.

3. How to Execute

• 3.1 Dependencies
  The project has dependencies on the following libraries so make sure you have installed those libraries using pip package manager using command python3 -m pip install -r requirements.txt.
  1. quandl
  2. investpy
  3. yfinance
  4. pmdarima
  5. fbprophet
  If you get any error regarding auto_arima() method of pmdadrima.arima submodule then first uninstall the older version of statsmodels library and current installation of pmdarima and then again install pmdarima, no need to install statsmodels library as pmdarima has a dependency on statsmodels which gets automatically installed during the pmdarima installation process.

• 3.2 Execution Steps

  • To execute the jupyter notebook click on badge/shield and do remember that the runtime is around 3 Hrs if you want to run all the cells again.

  • If you have some problems while executing the Google Colab notebook then consider executing it directly through the following Google Drive link.
    Google Drive Link: Forecasting S&P BSE SENSEX and S&P-500 Using ARIMA and Prophet.

NOTE: For project explaination, refer the presentation.

4. Publications Based on the Work

1. S. P. Sharma, J. R. and K. Deepa, "Forecasting India S&P BSE SENSEX and USA S&P-500 Benchmark Indices Using SARIMAX and Facebook Prophet Library," 2022 6th International Conference on Intelligent Computing and Control Systems (ICICCS), 2022, pp. 1523-1530, doi: 10.1109/ICICCS53718.2022.9788127.

@inproceedings{9788127,
	title        = {Forecasting India S&P BSE SENSEX and USA S&P-500 Benchmark Indices Using SARIMAX and Facebook Prophet Library},
	author       = {Sharma, Suraj Prakash and R., Jeyanthi and Deepa, K.},
	year         = 2022,
	booktitle    = {2022 6th International Conference on Intelligent Computing and Control Systems (ICICCS)},
	volume       = {},
	number       = {},
	pages        = {1523--1530},
	doi          = {10.1109/ICICCS53718.2022.9788127}
}

NOTE: To cite our work, please add the above block of code into your *.bib file in the latex environment and if you want to cite this repository, then please copy the respective style information (APA or BibTex) provided under cite this repository option as shown in the tutorial: https://github.blog/wp-content/uploads/2021/08/GitHub-citation-demo.gif

GNU General Public License v3.0