Predicting Financial Markets with Google Trends and Not so Random Keywords

Abstract

We discuss the claims that data from Google Trends contain enough information to predict future financial index returns. We first review the many subtle (and less subtle) biases that may affect the backtest of a trading strategy, particularly when based on such data. Expectedly, the choice of keywords is crucial: by using an industry-grade backtest system, we verify that random financerelated keywords do not to contain more exploitable predictive information than random keywords related to illnesses, classic cars and arcade games. However, other keywords applied on suitable assets yield robustly profitable strategies, thereby confirming the intuition of [24].

When Can Social Media Lead Financial Markets?

“…We evaluate Δ Tweet volume against Δ Price (the hourly returns) for each financial-instrument/Twitter-Filter combination to evaluate the extent to which hourly changes in Tweet volumes can lead the securities' hourly returns using our methodology as an echo of past studies which compare social media89 and search engine10111213 message volumes with financial market performance. We then also repeat this experiment to consider Δ Tweet volume against |Δ Price | (the absolute hourly returns) to further explore the ability of hourly changes in Tweet volumes to lead securities' hourly returns.…”

“…We perform our analysis on hourly changes in Tweet sentiments vs. the hourly returns of forty-four financial instruments, showing that Twitter sentiment leads securities' returns in a statistically-significant manner for twelve instruments. We then perform identical analyses on the hourly changes in Twitter message volumes vs. the hourly returns and the absolute hourly returns of the same forty-four financial instruments, to echo recent studies which compare social media89 and search engine10111213 message volumes with financial market performance. We demonstrate that the Tweet sentiments result in proportionally larger maximum information surplus values compared to the maximum information surplus values seen from our Tweet volume (rather than Tweet sentiment ) experiments.…”

“…Social media data have been used to measure and predict real-world phenomena such as brand popularity3, motion picture box office returns4 and election outcomes5. Of present focus is the prediction of financial markets via the analysis of Tweets6789 and other comparable data sources such as Google Trends101112, Yahoo! search engine data13 and Wikipedia articles14.…”

“…The first centres on the evaluation of the volumes of social media message89, search engine queries10111213, and Wikipedia views & edits14. However, such studies do not quantitatively evaluate the contents of social media and Internet text-strings – a valuable source of data – and instead consider just their volumes .…”

Can We Predict the Financial Markets Based on Google's Search Queries?

“…When a new maximum is found in the dataset, it changes all the previous values of normalized search frequency. Unfortunately, this is a permanent problem when working with this dataset and it may jeopardize the reproducibility of the research (Challet and Ayed, ). We expect, however, that given the scale of our research, our main results are not be biased by this property.…”