Beat The Stock Market With Statistics

Wall Street traders make and lose a lot of money by buying and selling thousands of stocks in high volumes and at ever faster rates. How do they decide which stocks to sell or buy? Once upon a time, this was done by looking at individual stocks and trying to determine the company’s underlying value through painstakingly careful research. But in the computer era, things have changed, and traders are increasingly making decisions about what stocks to buy, sell, or hold on the basis of computer models. These models mine vast records of stock sales and purchases in order to determine which way the market is headed.  

In this talk, I will discuss some basic statistical concepts that arise in the automated computational analysis of stock market data. I’ll talk about how we can determine which stocks are correlated with each other, how we can predict which way the market is going, and more. I’ll also discuss the risks involved in misspecifying the statistical model, and how a little mistake in modeling the data can lead to a big bust!