Financial Machine Learning & Natural Language Processing as it Applies to ESG Investing

 

 

ESG stands for environmental, social, and governance. Today, over 20 trillion dollars in investments are allocated in an ESG conscious manner. Evaluating corporations and entities based on these criterion is primarily performed in manual ad-hoc manners such as checklists. However, natural language processing and other machine learning algorithms are particularly well-poised to overcome the current methodologies’ weaknesses. This project is a data-driven search for an algorithmic approach to ESG evaluation, scoring, and investing.

Optimal Cross-Currency Hedging for Differing Asset Classes

 

 

 

Consider the example of an institutional investor buying an asset in a foreign market. To protect against FX risk, the current methods to hedge this risk are swaps and futures. However, oftentimes there exists a correlation between the foreign asset and currency. The goal of this project is to better understand the relationship between forex returns and foreign markets.

 

Our preliminary investigation combines empirical methods with a methodology relying primarily on factor models and regime-switching models.

Mathematical Theory of Exotic Volatile & Fragmented Financial Markets – Inspired by Cryptocurrency

 

 

 

The cryptocurrency market can be thought of an additional example of a fragmented and volatile up-and-coming market with a particularly high number of speculating investors. The purpose of this project is to further develop and innovate a new mathematical theory to address and evaluate financial decisions in markets which share the aforementioned characteristics. Tailored theories of portfolio construction and risk management for these types markets is the ultimate goal.

Analyzing Market Crisis and the Asset Allocation Methodologies

 

 

 

 

In a 2013 RiskLab paper, we studied minimum variance, equal-risk contribution, and 1/N portfolio allocation strategies to find that under different market conditions, particularly crisis states, optimality changes. It was thus proven that switching between portfolios according to said market conditions produced better Sharpe and Omega ratios. The purpose of this project is to generalize these results to more complicated back-testing methodologies, more allocation strategies, and more refined dynamic portfolio constructions.

 Financial Category Theory

 

 

 

Our research in this area is focused on generalizing a new research area - financial category theory - introduced in 2016. In particular, by looking at categories for generalized portfolio theories, we seek to precisely determine if some empirically cited flaws of portfolios under different investment objectives are avoidable or not.