Invest in PTOs or die trying: Achieving abnormal returns by trading takeover likeliness

Abstract

In this Master’s Thesis, we examine the potential to achieve abnormal returns by predicting takeover targets within the US market. This study extends on previous literature by aiming to address known issues that dilute the returns of takeover prediction strategies, such as prediction errors, timing issues, and the misclassification of distressed firms as potential targets. By utilizing advanced machine learning techniques, including random forest and gradient boosting, along with a monthly forecasting framework and new variables, we are able to notably enhance predictive accuracy as well as generate abnormal returns.

Our approach introduces a framework to differentiate between various levels of firm risk, using measures such as Distance to default and Safety, which improves the model’s performance and provides deeper insights into the characteristics of potential takeover targets. Additionally, we apply subcomponents of the Quality factor developed by Asness et al. (2018) to enhance our model’s predictive power. Doing so we improve model accuracy but also find that firms with higher risk profiles, albeit not in distress, are more likely to be takeover targets.

Our best models beat random selection by a magnitude of nearly three times – a clear improvement to previous literature. Correspondingly, we find that all our models overperform the market and generate significant abnormal returns. We also introduce a long-short strategy to isolate pure PTO- driven returns. Our long-short strategies yield significant abnormal returns and achieve near factor- neutrality in the Fama-French three and momentum framework.

In conclusion, while predicting takeovers remains challenging, our research demonstrates that it is possible to achieve abnormal returns with takeover prediction models. This study not only timely revisits the topic of takeover prediction but also offers practical applications for investment strategies based on takeover prediction, as well as new insights on the characteristics of PTO targets.