Quant Interview with Morgan Stanley: Don’t Make This Critical Error

Before you step into a quant interview with Vishwanath Tirupattur, a director of quant research and the head of U.S. fixed income research at Morgan Stanley, you should probably know where his phobias lie.

Tirupattur is ultra-averse to data mining, or the process by which usable data is extracted from larger sets of raw data in an effort to discover hidden meanings. “We are acutely conscious that data mining is something we don’t want to do,” said Tirupattur, speaking at the recent Quant Conference Digital. “If you torture the data enough, it will tell you something you want to hear.”

Instead of saying you’re an expert in data mining during an interview with Tirupattur, you should describe your talent for combining scientific research with empirical data to create strategies that generate returns

“Complexity in itself is not a virtue,” Tirupattur said. “The position of travel for Morgan Staney’s quant research now is to rediscover simple scientific quant strategies… rooted in empirical research and designed to harvest persistent returns.” 

Tirupattur added that a good quant strategy is one that allows several questions to be answered: Is the source of the returns understood? How can it be cross-validated? And what could be wrong?

If you want a quant research job at Morgan Stanley, Tirupattur said, you’ll need to combine common sense with an understanding of cutting-edge technologies and a willingness to collaborate. “A lot of the work that we do in quant research involves taking the fundamental insights of our analysts,” he said. It’s a question of creating a “simple set of rules” from qualitative analysis provided by credit strategists, equity analysts and mortgage analysts, for example. 

Morgan Stanley is increasingly applying new technologies and quantitative techniques to different areas, and Tirupattur said the distinction between quant and non-quant investing is being eroded as a result. In Municipal Bond Investing, for example, he said the bank has begun using natural language processing (NLP) to analyze 120,000-word bond prospectuses for clues to future ratings actions and the likelihood of defaults.