Proceedings

David Willingham, MathWorks

In this keynote presentation, we explore the transformative impact of MATLAB^® in the field of quantitative finance. We focus on four critical areas that have revolutionized workflows for finance professionals, leading to increased productivity and innovation.

First, MATLAB's advanced automation capabilities have streamlined manual processes, saving valuable time for financial experts. By automating repetitive tasks, professionals can now focus on strategic decision-making and in-depth research, resulting in more insightful and data-driven outcomes.

Second, MATLAB has democratized model development and deployment by integrating low-code and no-code workflows. This shift allows non-programmers to actively participate in the quantitative finance domain, fostering a more inclusive and collaborative environment. The diverse expertise of individuals from various backgrounds contributes to novel financial solutions.

Next, the seamless integration of AI and quantum technologies in MATLAB enables finance engineers to develop intelligent algorithms and solve complex financial problems rapidly. Leveraging these transformative technologies empowers professionals to achieve breakthroughs in risk assessment, asset management, and predictive analytics, paving the way for new dimensions of financial success.

Finally, the incorporation of ModelOps within MATLAB has revolutionized model deployment and monitoring. Continuous monitoring and adaptive management ensure that financial models remain dynamic and effective, even in changing market conditions, bolstering overall financial resilience.

These transformational technologies have ushered in a new era of productivity and innovation in quantitative finance, with MATLAB playing a central role as an enabler of progress. As the financial world continues to evolve, these advancements will empower finance professionals, shaping the future of the industry and unlocking new possibilities for success.

Valerio Sperandeo, MathWorks

Determining the asset allocation of a fund is not a trivial task, particularly over longer time horizons and in a multiperiod portfolio rebalancing framework. Traditional approaches used to solve the optimal asset allocation are often based on strong model assumptions. Reinforcement learning is an area of machine learning where the portfolio problem can be defined with greater flexibility, and that may help produce better trading and asset allocation strategies.

This talk demonstrates how reinforcement learning can be used to solve a multiperiod asset allocation problem. We show how MATLAB^® can be used to:

• Define portfolio wealth states and reinforcement learning actions from a frontier of portfolio strategies
• Model reward functions and train Q-learning and custom agents
• Simulate portfolios with the trained agents within specified environments

Killian Pluzanski, Amundi Asset Management

The Amundi Convexity Solutions team is a leader in derivative management specializing in offering personalized option-based protective overlays to investors. These unique strategies focused on risk management and downside protection challenge the conventional performance metrics typically used to assess investments, such as return on investment or Sharpe ratio.

To address this challenge, the team has developed a robust performance analysis framework. This framework accounts for various dimensions, including downside risk protection, drawdown management, and stress-testing scenarios. By evaluating protective overlays under diverse market conditions, the team gains valuable insights into the effectiveness of these strategies.

A key aspect of these protective overlays is their high level of customization, which is tailored to each investor's specific needs. As a result, each analysis requires the generation of dedicated reports. These reports provide comprehensive insights into the efficiency of the protective overlays by considering the cost-benefit ratio and aligning with broader investment objectives.

To facilitate this process, the team has integrated the project into an application using MATLAB^® App Designer. This tool offers a complete environment, encompassing application development and deployment. By utilizing this application, the team can efficiently generate, customize, and present performance analysis reports, enabling clear communication with both internal and external clients.

Marcus Veltum, Helaba Invest
Thomas Nitschke, Helaba Invest

Helaba Invest was founded in 1991 as a wholly owned subsidiary of Helaba. Since then, it has been responsible for the professional management of assets of institutional investors within the Helaba Group. Helaba Invest’s business strategy is based on the three business areas of liquid assets, illiquid assets, and administration. With a volume of assets under management of around €223 billion (EUR), Helaba Invest is one of the leading capital management companies in the segment of institutional asset management.

Helaba Invest has been using MATLAB^® in its risk management processes for many years and has established scalable derivative valuation and key indicator calculation methods within the company. In 2020, Helaba embarked on the development of an in-house front-office platform based on existing architecture for its risk overlay management system using products such as MATLAB, MATLAB^® Production Server™, and MATLAB^® Web App Server™.

Hear the advantages and the extensive scope of application that the newly developed front-office platform offers, from back testing to detailed portfolio views. This presentation also addresses current development ideas and--in collaboration with MathWorks—demonstrates how the efficient integration between MATLAB Production Server, MATLAB Web App Server, and Microsoft Excel works in practice.

Sofia Ma, MathWorks

In this session, we navigate the rapidly evolving intersection of quantum computing and finance, unveiling its potential to reshape industry paradigms. The focus centers on two prominent areas: quantum finance portfolio optimization and quantum Monte Carlo simulations.

We commence our exploration with quantum finance portfolio optimization. The talk illuminates how quantum algorithms, with their inherent computational advantages, have the potential to revolutionize portfolio management by processing voluminous financial datasets with more efficacy and identifying optimal portfolio mixes with unprecedented methodology.

Following this, we plunge into the realm of quantum Monte Carlo simulations. These stochastic simulations, while integral to financial risk assessment, often suffer from computational sluggishness due to the large number of variables and random outcomes involved. Our talk examines how quantum Monte Carlo methods can turbocharge these computations, leading to considerable enhancements in risk analytics and decision-making procedures.

Through the course of this talk, we provide a detailed exposition of the expanding role of quantum computing in finance. We showcase how these advanced computational methodologies can reengineer traditional financial practices by solving intricate problems previously deemed computationally intractable. We also consider the practical hurdles, current advancements, and outlook of quantum computing's integration into the financial industry.

Sudeep Lahiri, Morgan Stanley

This presentation provides an overview of artificial intelligence and machine learning techniques within the world of finance, spanning across trading, banking, wealth and asset management, and other financial applications. The exuberance of employing AI and machine learning techniques in finance driven by the success in software engineering and social media applications has led to rapid deployment of AI and machine learning in certain areas of finance like client outreach, language processing, and utilization of large and alternate data. Organizations should tread with caution around the challenges of using such techniques and increased interest from regulators. Other conventional areas like pricing and risk management have deliberately lagged in deployment of AI and machine learning techniques, driven by market conventions of pricing, which may change in the future thanks to increased access to data and more dynamic pricing and risk-management capabilities. This presentation also mentions expectations from diverse regulators and progress made in academia and the upcoming generation of finance professionals.

Léa Boittin, Caisse Centrale de Réassurance

Caisse Centrale de Réassurance (CCR) is a public reinsurer operating in France. CCR provides coverage against earthquakes, drought (clay shrinkage and swelling), floods, storm surge, and tropical storms, as well as terrorism and civil nuclear liability. Within CCR, the R&D Modeling department develops hazard, vulnerability, and damage models that allow CCR to understand the risk exposure of the French territory and estimate the insured losses from catastrophic events.

In this talk, learn how CCR uses climate simulations provided by Météo-France along with its own catastrophe models to assess how hazards and insured losses in France will evolve in 2050. CCR and Météo-France use constant climate simulations: for a chosen target year (2000 or 2050) and a chosen IPCC scenario (current climate, RCP 4.5, or RCP 8.5), they compute 400 repetitions of the target year at an eight kilometer spatial resolution and an hourly temporal resolution. This original methodology allows CCR to estimate return periods for extreme events for a given target year and compare their annual probability of occurrence in a changing climate.

Michael Robbins, Columbia University
Arpit Narain, MathWorks

In this conference presentation, we examine a market-based methodology to assess banks' resilience to climate-related risks. This methodology is based on the work done by the economists at the Federal Reserve Bank of New York and other academics, including the Nobel Prize winner Bob Engle, and has been published in the Fed staff working paper. Additionally, a large Asian central bank replicated the methodology for their geography and published the results.

The methodology examines the climate-related risk exposure of large global banks using a novel metric called CRISK, which represents the expected capital shortfall in a climate stress scenario. It also introduces climate risk factors and measures banks' stock return sensitivity—referred to as climate beta—towards these factors.

This presentation highlights the significance of this approach in assessing and managing climate-related risks at the financial institution level and emphasizes its potential in enhancing the financial sector's resilience to climate change.

Sébastien Villemot, CEPREMAP

Dynare is a macroeconomic modeling tool that runs on top of MATLAB^®. It offers a user-friendly and intuitive way of describing these models through a domain-specific language. It is able to perform simulations of those models given a calibration of the model parametersand estimate these parameters given a data set. A large panel of applied mathematics and computer science techniques are internally employed by Dynare: multivariate nonlinear solving and optimization, matrix factorizations, local functional approximation, Kalman filters and smoothers, MCMC techniques for Bayesian estimation, graph algorithms, and optimal control.

Private financial institutions and various public bodies including central banks, ministries of economy and finance, and international organizations use Dynare for performing policy analysis exercises and as a support tool for economic forecasting exercises. In the academic world, Dynare is used for research and for teaching purposes in postgraduate macroeconomics courses.

Dynare is free/libre/open-source software. Over the years, a large community has formed around it, through online forums and dedicated physical events such as scientific conferences and training workshops.

This presentation gives an overview of the project, in its scientific, technical, and social dimensions.

Alistair Dieppe, European Central Bank

The Bayesian Estimation, Analysis and Regression (BEAR) toolbox is a comprehensive Bayesian (Panel) vector autoregression toolbox for forecasting and policy analysis. It is based on MATLAB^® and widely used by central banks, academia, and finance. This presentation includes an overview of the toolbox and the latest developments.

Kadir Tanyeri, International Monetary Fund

The global dynamic stochastic general equilibrium model for forecasting main macroeconomic variables like GDP, inflation, and unemployment is nonlinear. There is a crucial need to compute confidence bands around the projections in order to establish the uncertainty about them, detect escalated up/down risks, and calculate useful statistics like recession and deflation probabilities.

Confidence intervals are calculated by drawing samples from the estimated distributions of exogenous shock terms and each time solving the system of equations using a nonlinear solver in MATLAB^®. The standard method, Monte Carlo sampling, is not practical due to the enormous number of drawings needed. We opted for a more structured way of drawing the shocks in order to more evenly sweep the high dimensional space—Latin hypercube sampling–which we have implemented in MATLAB. This sampling technique implies a faster convergence; in other words, a smaller number of simulations is needed to obtain good estimates of the confidence bands.

Furthermore, we do use distributed computing in MATLAB over a cluster of servers to speed up the process even further. System solution and calculations are sent to more than 100 workers on a cluster of several servers, then results are collected and compiled, which enables the whole calculation to be completed overnight instead of taking months if none of these methods were involved.

Mohammad Jahan-Parvar, Federal Reserve Board

This talk highlights how we document the predictive ability and economic significance of foreign economic policy uncertainty for US equity returns. After orthogonalizing global economic policy uncertainty (EPU) with respect to the US EPU, we find that it has significant predictive power for aggregate stock returns and returns of portfolios constructed on size, investment, capital expenditure, and foreign sales when forecasting 6 to 12 months out. For individual companies, we show that foreign EPU commands an economically significant and negative-valued premium in the cross-section of returns. The stocks of firms that are highly sensitive to foreign EPU, whether positive or negative, outperform those that are less sensitive to this measure.