Predicting Stock Prices: A Hands-On Guide with ARIMA
Welcome to the world of stock price predictions! This guide will demystify the process using ARIMA models—a blend of technology, math, and finance. Whether you’re a finance expert, data scientist, or just intrigued by financial markets, this tutorial is tailored for you.
We focused on AMD, a leading NYSE-listed company, collecting daily stock prices from January 1, 2021, to June 15, 2023. Key metrics like Open, High, Low, and Close prices formed our robust dataset.
A lower p-value indicates stronger evidence for stationarity.
If Test Statistic < Critical Values, the data is likely stationary.
At first, our data didn’t show the consistency we had hoped for. We tried to stabilise the series by subtracting the rolling mean to deal with this. By doing so, we were able to determine how much prior values affected the current values in the series, allowing for a more accurate evaluation.
After subtracting the rolling mean, we saw a considerable decrease in noise, which effectively eliminated the time series’s temporal dependencies over the preceding 12 days. We verified this by examining the resulting series for stationarity using the Dickey-Fuller test:
0 Comments