Applying PCA to Stocks

This blog post is a summary of a data science project I worked on a few months ago. This was another attempt at trying to understand hidden trends in the stock market. Hopefully results will show you how "non-linear", complex, and unpredictable the market can be. Introduction: A stock's time series can be thought of as some realization of an underlying trend with added stochasticity or "noise". So surely, for an appropriate window of time, one can group bunches of stocks that are moving with an underlying trend.