How I tried to Predict the S&P500 Stock Price with Python and Machine Learning
Predicting stock prices with machine learning and sentiment analysis tutorial.
Part1 : “Stock News Sentiment Analysis with FinBERT Transformer.”
In a previous post, I wrote a tutorial on how to conduct sentiment analysis on news headlines using the FinBERT Deep Learning Transformer model. I suggest you spend 5 minutes to read it before you continue with the current article. I used the Hugging Face library and provided the relevant code to automate it.
In this article, I use the sentiment analysis scores extracted, in order to make price predictions on the S&P500 index. Our aim is to correctly predict, whether the stock price will move upwards or downwards in a given day. Machine Learning (ML) models will be used.
1. The Strategy
S&P500 index consists of around 500 companies. However, the biggest 35 companies represent about 48% of the index’s value. If we conduct a sentiment analysis on the headlines referring to those 35 companies, we can speculate that the average sentiment score represents the overall sentiment of the S&P500 (which in turn represents the US stock market).
Hence, I gathered headlines for those 35 companies, over a 10 year period ranging from 2010 to 2020…
