CSML Alumni Profile: Ujjwal Dahuja uses data science to make sense of complex, big data in finance

Aug. 24, 2022

Ujjwal Dahuja, 28, Class of 2017

Background:

Dahuja earned his bachelor’s degree in economics from Princeton University, while also fulfilling requirements for the undergraduate certificate at the Center for Statistics and Machine Learning (CSML), and additional certificates in finance, engineering and management systems, and South Asian studies.

Current Work:

Dahuja is currently an associate at Blackstone Alternative Asset Management (BAAM), a leading alternative investment firm. At BAAM, Dahuja is in a team responsible for making allocation recommendations on quantitative hedge funds, private hedge funds that use systematic processes for asset selection. In his role, Dahuja evaluates various quantitative strategies and how they may fit within client portfolios. This involves using data science and machine learning methods such as principal component analysis, hierarchical clustering, and ensembling methods to analyze and combine insights from several financial datasets.

Prior to Blackstone, Dahuja was a quantitative research analyst at Weiss Multi-Strategy Advisers and at AQR Capital Management. Earlier in his career, he also interned at Deutsche Bank and Princeton University Investment Company (PRINCO).

Dahuja has been using machine learning techniques in personal projects too. Most recently, he has been working with neural networks and reinforcement learning models to build short term prediction rules in crypto markets.

Undergraduate Work:

For his undergraduate senior thesis in economics, which also fulfilled the requirements for the CSML certificate, Dahuja was advised by Mark Watson, the Howard Harrison and Gabrielle Snyder Beck Professor of Economics and Public Affairs. His thesis focused on why firms choose accelerated share repurchases over open market share purchases. Public companies usually return capital to shareholders in the form of dividends and share buybacks. There are many different ways of doing share buybacks, one of which was the focus of his thesis.

The machine learning component of his thesis included principal component analysis and natural language processing which were used to both construct and analyze data sets.

Dahuja said the CSML certificate gave him a good foundation that helped him in his thesis and subsequent work.

“Understanding the mathematical foundations, assumptions and mechanics of a machine learning model are very important,” he said. “Given how difficult it can be to gain insight into why a model behaves the way it does, understanding the assumptions and mathematics are key to successful implementation. Equally important is the quality of data being used. Particularly in finance, where signal to noise ratios are often low and time series are often not stationary, properly cleaning and normalizing features are equally as important. Although it has come at the cost of iterating through fewer ideas, I have leaned towards spending more time to ensure data inputs are appropriate for the model setup, that the model is suitable for the research question and hyperparameters are well thought through prior to testing.”

Dahuja believes that ORF 350: Analysis of Big Data, was a particularly helpful class because it helped him gain an understanding of both machine learning theory and its applications. “A lot of machine learning classes often emphasize applications over theory, but ORF 350 strikes a good balance”, Dahuja said.

Future Goals:

Dahuja is passionate about quantitative investing and hopes to continue building his skill set in the alternative investment industry for the foreseeable future. Eventually, he would like to consider creating his own financial services company that is driven by empirical evidence and uses data science to make decisions.

Extracurricular Activities:

While at Princeton, Dahuja was a member of the University’s Model UN and Tower Club. Outside of work, Dahuja enjoys playing chess, cricket, and board games.