Research Project Title:
Predicting Investor Behavior During Economic Crises
abstract:Predicting investor behavior is critical to understanding economic activity and the movement of money. Thus far, studies have only performed empirical studies with little data. In contrast, we aim to use machine learning techniques to analyze a dataset of portfolio snapshots from 600,000 investors spanning 2003 to 2015. Naturally, we must look into data-efficient frameworks and techniques to process our data. Then, we plan to extract relevant features from past trades and portfolios and derive a robust model that will help us predict investor behavior over time. After pinpointing key predictors of investor behavior, we hope to feed these features into a neural network that will construct a predictive model for investor behavior during economic crises.
From writing papers to presenting my findings, I am excited to gain insight into a wide variety of topics during guest lectures and experience the research process firsthand throughout this SuperUROP. Moreover, I hope to increase my understanding of financial economics, develop my skills in data analytics, and apply machine learning concepts from class to the real world.