The Department of Industrial Engineering and Operations Research
(IEOR) at Columbia University in New York City invites applications
for a post-doctoral position. The primary responsibility of the
candidate is to research on novel methodologies for the analysis of
robo-advising systems. The objective of this research is to leverage
and extend existing techniques from the machine learning and
statistical literature to design automated investment systems that
exhibit superior performance to human-only and machine-only driven
systems. The position is funded by JP Morgan, and the candidate will
work closely with Prof. Agostino Capponi and with the artificial
intelligence team at JP Morgan, with ample opportunities to leverage
their technology and in-house data.
Machines have the ability to process hard information, i.e., to make
complex reasoning based on the gigantic amounts of market information.
However, they can only approximate human behavior up to a quantitative
model. In contrast, humans can assess complex environments and make
judgments based on a holistic perspective that incorporates soft
information sources. Humans’ internalized risk preferences and
objectives, however, are difficult to quantify and communicate to a
machine. It is thus central to design a mechanism through which the
machine progressively learns and acts according to the investor that
it serves. Investors need to trust that the machine understands the
dynamics of their risk preferences, their objectives, and their market
beliefs, before delegating higher autonomy to the machine.
The candidate will have the unique opportunity to work on a cutting
edge framework that departs from the current approaches currently used
in the wealth management industry. These approaches are typically
based on a one-shot interaction: the investor communicates once and
for all his risk attitude or objectives to the machine, and the latter
executes autonomously the investment. Unlike these approaches, in the
proposed approach, the machine actively elicits information about the
investor’s risk preference. Through the establishment of a
communication protocol, the human-machine system selects portfolio
instruments and strategies that uniquely reflect the investor’s
current taste for risk and reward.
Candidates must have a PhD degree in Computer Science, Statistics,
Electrical Engineering, Operations Research, or Applied Mathematics.
Candidates are expected to be familiar with reinforcement learning,
active learning, collaborative filtering, game theory, computational
economics, convex and stochastic optimization. Experience with
programming languages such as Python is highly desirable.
The candidate will also benefit from interactions with various faculty
from the IEOR Department, the school of Engineering and Applied
Sciences, the Business School, and the School of International and
Public Affairs. This is a one-year position with possibility of
renewal depending on the progress achieved.
Candidates should submit electronically to Prof. Agostino Capponi
(email@example.com) the following: curriculum-vitae, a copy of their
official degree transcript, and a representative research paper. The
applicants should also arrange to have at least one letter of
recommendation submitted electronically to the same address. The
position will remain open until filled and applications will be
reviewed as they are submitted.
Applicants can consult https://ieor.columbia.edu/ for more information
about the department.
Columbia University is an Equal Opportunity/Affirmative Action