Web Search Queries Can Predict Stock Market Volumes
The study of queries is gaining more and more attention as an important tool for the understanding of social and financial systems. Users perform web searches in order to collect news or browse e-newspaper sites. In particular local or global events such as natural disasters can generate local or global waves of searches through the web. As a result, the logs of these search-engines’ queries are an unprecedented source of anonymized information about human activities. In this paper we provide a detailed analysis on a particular application of these ideas; that is, the anticipation of market activity from user queries. This picture graphically summarizes our procedure. In particular, we investigate which is the relationship between web searches and market movements and whether web searches anticipate market activity. While we can expect that large fluctuations in markets, produce spreading of news or rumors or government’s actions and therefore induce web searches (solid green arrow in panel ), we would like to check if web searches affect or even anticipate financial activity (broken violet arrow in panel ). In detail we investigate if today’s query volumes about financial stocks somehow anticipate financial indicators of tomorrow such as trading volumes, daily returns, volatility, etc, (panels and ) and we find a significant anticipation for trading volumes.