Risk, Return and Deal Success: An Empirical Study of Merger Arbitrage Performance and the Use of Prediction Models to Enhance Profitability

Abstract

This is an extensive empirical study where we examine the risk and return characteristics of the merger arbitrage strategy. We have analysed 4987 deals in the period 1996 to 2015 from the US market. In contrast to earlier findings, we conclude that merger arbitrage possess linear dependency with the market. Additionally our findings suggest that a merger arbitrage strategy outperforms the stock market both in terms of Sharpe ratio and alpha. Further we evaluate the possibility to enhance the performance by building a model predicting deal success. The model discovers both new and previously documented predictors of deal outcome. Using online machine learning techniques, we create an algorithm that invest in a sub-sample of the available deals, given predictions by the model. This algorithm successfully improves the annual CAPM alpha from 8.4% to 12.0% and the Sharpe ratio from 0.76 to 1.15 for a merger arbitrage portfolio from 2002 to 2015. Consequently, we conclude that factor predictability is not sufficiently priced in.