Forecasting Stock Returns Using Google Trends

Abstract

Stock price changes reflect decisions made by market participants and participants begin their decision-making process by gathering information, which in today’s world largely consists of searching online sources (i.e. Google). In study I examine the correlation between Google Search volumes and stock returns to determine if it is possible to predict market participant behavior, and subsequently stock returns, based on online search volumes for company names along with the word “stock”. I control for market capitalization, market sector, and pre and post COVID states in my analysis. The results indicate that there is no correlation between search volume and stock returns when search volume is the only explanatory variable in the model, and market capitalization nor market sector had a significant effect. When excess market return and credit spread are added to the regression model, search volume has a negative effect on stock returns at a confidence level of 99 percent in the period 2018-2023, and a negative effect on stock returns at a confidence level of 95 percent both pre and post COVID. This phenomenon can be explained in part by the negative nature of the news, as negative news stories tend to drive increased attention and search volume and are therefore associated with a decrease in stock prices. If search volume for a company stock is a proxy for new information about that company, which I believe it to be, then the lag in time between investors’ search interest and subsequent investment decisions indicates that investors do not always immediately act upon new information. This is further proven by the increase in R-squared value when search volume is added as an explanatory variable in the CAPM. This disproves the Efficient Market Hypothesis, which states that stock prices reflect all available information.