ARIMA and R - Stock Price Forecasting Made Easy!
Tech Know How Tech Know How
7.84K subscribers
34,782 views
582

 Published On Sep 15, 2019

This video tutorial is a complete walkthrough on how to do quick stock price forecasting with ARIMA models in R. We will forecast the future values of SPY (the S&P 500 ETF) with daily close price data from Yahoo Finance. I will show you how to get the daily data in XTS (time series) format, explain XTS formatting and then walk you through a quick and easy forecasting process that uses 4 models (auto.arima, 2 custom arimas, and a typical default arima) for comparison.

Get a free stock like Apple, Ford, or Facebook worth up to $500 just by signing up and joining Robinhood using my link: https://join.robinhood.com/davidm16707

As you will see in the video the stock price forecasts are very high accuracy (determined by MAPE and numerous other indicators). An important note that I discuss and make clear in the video is to use this forecasted stock price data in a directional manner. Don't try and forecast the exact price each day or similar. Also, don't try forecasting exact gains on this time series. There are far too many external and internal influencers that are highly volatile and can really give you wrong forecasting data. Instead look at a range and the general direction. Is it positive, negative or flat? With the SPY stock price forecast we see that the models all point to a positive or flat trend for the next 100 days (period I selected in the term variable).

Join in the discussions and more on my FaceBook Data Science group. It's packed full of great free resources, code, cheatsheets, expert advice and much more!   / 2558540907733212  

This same arima forecasting process can be used for stock price forecasting with any publicly listed stock on Yahoo Finance. You can use IBM, AAPL, XOM, etc... Try some different ETFs (better diversification) and mutual funds or stocks and have fun seeing where they are trending over the next few months. Also feel free to change from past 5 years of time series data to 10 or 20 years to include some recessions and see how or if that changes the forecasted trends.

I hope you found this both informational and entertaining.

Please take a moment to subscribe, like and share. Be sure and click the bell so that you will be notified every time I create and publish a great video like this one!

Thanks again and God bless!

show more

Share/Embed