Workshop: Data Science and Machine Learning in Finance

Jin Zheng -- University of Bristol

Research Talks on Fintech. This event is hybrid, and a Zoom invitation will be sent later for online attendance.

Wills Reception Room, Clifton Hill House

<h1>Agenda</h1><p>10:00 am - 10:20 am: Arrival, tea &amp; coffee</p><p>10:20 am - 11:00 am: Dr. Emmanouil Platanakis, University of Bath</p><p>11:00 am - 11:40 am: Dr. Ramin Okhrati, University College London</p><p>11:40 am - 12:00 pm: Tea/coffee and mid-morning snack</p><p>12:00 pm - 12:40 pm: Prof. Gilles Chemla, Imperial College London</p><p>12:40 pm - 2:00 pm: Lunch</p><p>2:00 pm - 2:40 pm: Prof. Ram Gopal, University of Warwick</p><p>2:40 pm - 3:20 pm: Dr. Yan Ge, University of Bristol</p><p>3:20 pm - 3:40 pm: Tea/coffee and fruits</p><p>3:40 pm: Discussion</p><p></p><h1>Invited talks</h1><h3>When Bayes-Stein Meets Machine Learning: A Generalized Approach for Portfolio Optimization --- Emmanouil Platanakis</h3><p>The Bayes-Stein model provides a framework for remedying parameter uncertainty in the Markowitz mean-variance portfolio optimization. The classical version, however, suffers from estimation errors of model components and fails to consistently outperform the naive 1/N asset allocation rule. We comprehensively investigate the drawbacks of the traditional Bayes-Stein model and develop a generalized counterpart by refining model components with various well-tailored machine learning techniques, expanding the scope and applicability of the original Bayes-Stein model. Specifically, we propose a time-dependent weighted Elastic Net (TW-ENet) approach predicting expected asset returns, a hybrid double selective clustering combination (HDS-CC) strategy calibrating shrinkage factors, and a graphical adaptive Elastic Net (GA-ENet) algorithm estimating the inverse covariance matrix. Empirical studies demonstrate that our generalized Bayes-Stein framework can always offer better out-of-sample performance than the 1/N strategy. Importantly, our study tailors existing machine learning methods considering the specifics of financial issues, illustrating appealing directions for solving challenging financial problems with machine learning.</p><h1></h1><h3>Offline Deep Reinforcement Learning for Dynamic Pricing of Consumer Credit --- Ramin Okhrati</h3><p>We introduce a method for pricing consumer credit using recent advances in offline deep reinforcement learning. This approach relies on a static dataset and requires no assumptions on the functional form of demand. Using both real and synthetic data on consumer credit applications, we demonstrate that our approach using the conservative Q-Learning algorithm is capable of learning an effective personalized pricing policy without any online interaction. Our experimental analysis indicates a significant increase in the expected profit using this approach.</p><p></p><h3>Discretionary Choice, Motivation, and the SUTVA ---Gilles Chemla</h3><p>We illustrate how a rational agent&#x27;s motivation changes as one moves from RCTs to settings with discretionary choice. A distinguishing characteristic of discretion is due diligence: Before choosing treatment status, agents acquire information (signals). This alters motivation, with due diligence putting self-esteem at risk: Outcomes consistent with due diligence signals suggest good decision-making ability. Self-esteem risk exposure has instrumental value since learning about decision-making ability improves future decision-making. This channel increases interim incentives for undertaking actions that are informative about due diligence quality. Under random assignment without due diligence: nothing is revealed about decision-making ability; the self-esteem component of motivation is absent; and SUTVA is violated. Thus, RCTs may have limited portability. Nevertheless, RCTs with high treatment probabilities and selective trials with small reassignment probabilities can increase due diligence incentives, improving portability. To illustrate, we consider implications for: randomised micro-credit; military draft lotteries; and addiction treatment trials.</p><p></p><h3>Portfolio Choice for Online Loans and Implications for Platforms---Ram Gopal</h3><p>Online lending&#x27;s growth is limited by risk, regulations, and low awareness. To realize its potential, platforms must offer attractive investments. We introduce General Characteristics-based Portfolio Policies (GCPP), a framework for loan portfolio construction without estimating loan distributional properties. Using LendingClub data from 2013-2020, GCPP portfolios achieve 8.86% to 13.08% annualized IRR, outperforming equal-weight portfolios. Comparing to traditional assets, online loans yield competitive or higher returns, offering diversification benefits. GCPP can help platforms better serve borrowers and lenders, fostering business growth.</p><p></p><h3>Graph Representation Learning in Financial Technology---Yan Ge</h3><p>Financial technology seeks to improve and automate the delivery and use of financial services. It is an emerging industry that uses machine learning to help finance professionals make more informed decisions and create new opportunities for growth and innovation. Graphs are important data structures that abstract relations between discrete objects, such as social networks, transaction networks and stock networks.  Graph representation learning (GRL) aims to map graph data into a low-dimensional dense vector space in which the graph information is maximally preserved. This talk focuses on developing and applying powerful GRL to finance, such as risk management and asset pricing. </p><p></p><p></p>

Our {communityGuidelinesLink} describe the sort of content we prohibit on Eventbrite. If you suspect an event may be in violation, you can report it to us so we can investigate.

To report other categories of prohibited or illegal content, submit {link}

Workshop: Data Science and Machine Learning in Finance

More events from Jin Zheng -- University of Bristol

Discover more events from Jin Zheng -- University of Bristol, from Science & Tech to other experiences you might love.

Still looking for the right event?

Explore all events in Clifton and filter by date, category, and more to find the perfect fit.

Workshop: Data Science and Machine Learning in Finance

More events from Jin Zheng -- University of Bristol

Discover more events from Jin Zheng -- University of Bristol, from Science & Tech to other experiences you might love.

You might also like...

Browse more events with different dates, prices, and formats to find your next great experience.

Still looking for the right event?

Explore all events in Clifton and filter by date, category, and more to find the perfect fit.