Modeling Multifactor Event Driven Trading Strategy

Abstract

This research paper investigates the effects of corporate events such as guidance, stock buybacks, dividend announcements, etc. on stocks in the Russell 1000 index. Impact is measured via market relative returns over a multi-day period, thereby making the resulting research viable for inter-day trading. On the long-side, our model portfolio exceeded the Russell 1000 index by over 5762 bps, generating a 21.36% annualized return from 2013 to 2019. The short-only portfolio beat its benchmark, ProShares Short S&P500, by 3023 bps over the same 7 year period. The content herein outlines a process for extracting predictive features from the Corporate event data. Algorithmic models trained on these features exhibit excellent behavior, outperforming their respective benchmark indices. For this reason, these models are a suitable base from which to create Model Portfolios. CCS Concepts • Applied Computing methodologies➝Machine Learning.