MATLAB Computational Finance Virtual Conference

Abstracts

Embracing Complexity

Complexity is neither good nor bad. It is a reality of life. It is a condition that must be addressed whether you want to model financial systems or understand the world around you. So, how can we, as financial professionals, improve our ability to deal with complex challenges and perform increasingly complex tasks? In this presentation, MathWorks Fellow Jim Tung discusses how financial companies are creating and adopting new ways to master the development of complex systems and the analysis of complex phenomena, using MATLAB.

Jim Tung, MathWorks


Algorithmic Trading Strategies with MATLAB Examples

The traditional paradigm of applying nonlinear machine learning techniques to algorithmic trading strategies typically suffers massive data snooping bias. On the other hand, linear techniques, inspired and constrained by in-depth domain knowledge, have proven to be valuable. This presentation describes the application of the Kalman filter, a quintessentially linear technique, in two different ways to algorithmic trading.

Ernest Chan, QTS Capital Management LLC


Aspects of Risk Management for Variable Annuity Contracts

This presentation focuses on elements of PnL volatility arising from dynamically hedging VA (GMxB) contracts with a particular focus on risk-managed subaccounts. Historically, variable annuity contracts have been criticized as being underpriced in the marketplace given that the benefits and guarantees they offer are very rich. In this presentation, we show that the cost of hedging can be controlled if variable annuity contracts are written on risk-managed subaccounts.

Eoin Elliffe, Lincoln Financial Group


Embedding MATLAB Production Server into the OpenGamma Risk Analytics Platform

Taking MATLAB algorithms directly into a risk analytics environment, when done well, offers significant value for a financial organization while mitigating operational and model risks incurred from recoding. In this presentation, Joan demonstrates how you can rapidly and directly embed MATLAB components into the OpenGamma Platform. He shows how you can incorporate a MATLAB function into a MATLAB Production Server™ application, declare the Java interface, create the metadata to express input requirements, register the function in the function repository, and visualize the results in OpenGamma’s risk viewer. While easy to demonstrate and extremely agile on paper, this approach could disrupt a firm’s development process. The presentation concludes with some thoughts, based on Joan’s experiences as development manager at a European bank, about mitigating risk through incorporating into automated testing frameworks and ensuring resilient performance.

Joan Puig, OpenGamma


Evaluating Systematic Trading Strategies: Using MATLAB to Accelerate Quantitative Research

The key to developing successful investment strategies is ongoing research, and this depends on the pace of investigating and understanding new ideas. Two variables can influence this speed of research progress: the organization of the research team and the tools they have available. In this presentation, Ben demonstrates how Fischer Francis Trees & Watts, the global fixed-income investment manager within BNP Paribas, used MATLAB to build such tools. They were built using standard functionality from MATLAB products, user-written toolboxes, community content from MATLAB Central File Exchange, and more advanced integration with native Java™. These tools have allowed greater team collaboration and increased research bandwidth, to enable faster research progress.

Ben Steiner, Fischer Francis Trees & Watts


I Can Show You Some Great Derivatives Models, But Can You Crunch Them Fast Enough?

Most valuation and risk management models for derivatives are minor tweaks of Black-Scholes. Black-Scholes is a great model, whereas some of the tweaks are a step in the wrong direction. However, there are a few models based on Black-Scholes that are closer to how derivatives are used in practice and that capture essential features previously ignored. These are great models, better than all the more classical tweaks. So far so good. The main problem stopping these models from becoming more popular is that they are just too difficult to solve numerically when you have a typical portfolio consisting of many underlyings and many different types of derivatives. Can this number-crunching problem be solved?

Paul Wilmott, Wilmott Associates


Intuitive Analytics Grows Revenue with Rapid Development

Financial analysts develop algorithms, perform backtesting, and visualize key metrics to interpret data and refine customer portfolio strategies.

Peter Orr, Intuitive Analytics


Macroprudential Models: New Techniques to Support New Policies

In the aftermath of the global financial crisis, many central banks and financial regulators are considering new macroprudential policies to deal with so-called tail-risk events, the unlikely episodes of high financial distress with devastating impact on the economy. Such events have traditionally been outside the domain of mainstream macroeconomic research. To support the new policies, an entirely different family of models and modeling techniques need to be developed. In this presentation, we discuss some of the progress made in the International Monetary Fund and design simulation experiments that use MATLAB to illustrate the main issues.

Jaromir Benes, International Monetary Fund


A Real-Time Trading System in MATLAB

MATLAB has traditionally been used for analyzing data offline, presenting analytic recommendations that were then acted upon manually. However, MATLAB supports a direct interface with data feeds and online brokers, as well as the ability to present sophisticated graphics and user interfaces—all in real time.

This presentation demonstrates an end-to-end demo trading system in MATLAB, highlighting its potential as a platform of choice. Interactive Brokers is used to demonstrate live market data feed and account/portfolio input, as well as for sending trading orders to the market. The system’s user interface showcases the hidden visualization and interactivity potential of MATLAB for tracking order executions and charting financial time-series in real time. Some best practices for improving real-time performance are also discussed.

Yair Altman, Undocumented MATLAB


Speeding Up Algorithms: When Parallel Computing and GPUs Do and Don’t Accelerate

Most of us have multiple cores as well as a graphics processing unit (GPU) in our desktop computer, with grids and clouds readily available at arm’s length. While popular hype purports that it’s easy to use such resources to facilitate massive speed-up of numerical computing at practically no cost and effort, the reality is of course somewhat different. In this presentation, Aly and Michael use MATLAB valuation and backtesting examples to look at the sorts of calculations that can be sped up by CPUs, GPUs, and server-based solutions, and they suggest a common-sense framework to help answer the questions:

  • When does it make sense to go to a GPU or cluster?
  • What changes are reasonable or unreasonable to make in your program and workflow?
  • What are the costs and risks of speed-up, and how can we mitigate against them?

Aly Kassam, QuantSupport
Michael Weidman, QuantSupport


Enfoque VAR-Bayesiano para Proyecciones en MATLAB

Las autoridades monetarias buscan anticiparse a eventos que afecten el correcto desempeño de la economía. Así, los ejercicios de proyección de las principales variables macroeconómicas son fundamentales. Consecuentemente, los bancos centrales vienen desarrollando modelos con el objetivo de anticipar la evolución de algunas variables de interés, los cuales suelen requerir cierta sofisticación analítica. Además, es también de interés de un banco central que los distintos agentes tengan una lectura similar respecto de su percepción del entorno económico. De esta forma se fomenta la estabilidad macroeconómica y se facilita la administración monetaria.

No obstante, con frecuencia los modelos desarrollados por los bancos centrales no son fáciles de replicar por el público, por lo tanto se requiere de modelos que busquen cubrir esta dificultad. Por este motivo, en esta presentación se discutirán algunos modelos estadísticos de vectores autoregresivos de fácil implementación, cuya capacidad de pronóstico se puede mejorar significativamente con técnicas bayesianas.

Será evidente en esta sesión que MATLAB constituye una herramienta bastante versátil en la implementación y evaluación de modelos con estas características. Más aún, además de facilitar este tipo de ejercicios, MATLAB es sumamente útil para reportar resultados. Debido a la facilidad con la que MATLAB interactúa con plataformas para edición de texto como LaTeX y archivos PostScripts encapsulados para gráficos.

Alan Ledesma, Banco Central de Reserva del Perú


Modelización del Riesgo de Crédito con MATLAB

El contenido de esta sesión muestra cómo departamentos de riesgos pueden construir una infraestructura ágil de gestión de riesgos de crédito. Si está interesado en desarrollar y distribuir análisis de riesgos, esta sesión será de su interés.

Lo más destacado:

  • Clasificación de Credit Rating
  • Matrices de Transición y Probabilidades de impago
  • Análisis de riesgo de crédito

Paula Poza, MathWorks


Pronóstico de Probabilidades de Incumplimiento de Empresas con MATLAB

En esta sesión mostramos cómo construir y trabajar con un modelo de pronósticos de probabilidades de incumplimiento de empresas en MATLAB.

Los temas a tratar son:

  • Cómo trabajar con datos históricos de migración de crédito para construir ciertas series de tiempo de interés y visualizar la dinámica de las tasas de impago
  • Cómo construir, usando nuestras herramientas de Estadística y Econometría, un modelo de pronósticos para las probabilidades de incumplimiento de empresas
  • Cómo validar el desempeño del modelo (backtesting)
  • Cómo visualizar regiones de riesgo y cómo utilizarlas para pruebas de estrés, o tensión (stress testing)
  • Cómo pronosticar matrices de transición

Gabo Lopez-Calva, MathWorks