Three essays on the informational efficiency of financial markets through the use of Big Data Analytics.

Summary The massive increase in the availability of data generated everyday by individuals on the Internet has made it possible to address the predictability of financial markets from a different perspective. Without making the claim of offering a definitive answer to a debate that has persisted for forty years between partisans of the efficient market hypothesis and behavioral finance academics, this dissertation aims to improve our understanding of the price formation process in financial markets through the use of Big Data analytics. More precisely, it analyzes: (1) how to measure intraday investor sentiment and determine the relation between investor sentiment and aggregate market returns, (2) how to measure investor attention to news in real time, and identify the relation between investor attention and the price dynamics of large capitalization stocks, and (3) how to detect suspicious behaviors that could undermine the in-formational role of financial markets, and determine the relation between the level of posting activity on social media and small-capitalization stock returns. The first essay proposes a methodology to construct a novel indicator of investor sentiment by analyzing an extensive dataset of user-generated content published on the social media platform Stock-Twits. Examining users’ self-reported trading characteristics, the essay provides empirical evidence of sentiment-driven noise trading at the intraday level, consistent with behavioral finance theories. The second essay proposes a methodology to measure investor attention to news in real-time by combining data from traditional newswires with the content published by experts on the social media platform Twitter. The essay demonstrates that news that garners high attention leads to large and persistent change in trading activity, volatility, and price jumps. It also demonstrates that the pre-announcement effect is reduced when corrected newswire timestamps are considered. The third essay provides new insights into the empirical literature on small capitalization stocks market manipulation by examining a novel dataset of messages published on the social media plat-form Twitter. The essay proposes a novel methodology to identify suspicious behaviors by analyzing interactions between users and provide empirical evidence of suspicious stock recommendations on social media that could be related to market manipulation. The conclusion of the essay should rein-force regulators’ efforts to better control social media and highlights the need for a better education of individual investors.