All Seminars are at 12 noon (CDT) unless otherwise noted *
January 29, 2021
Speaker: Margie Rosenberg, University of Wisconsin - Madison
Title: A Cluster Analysis Application Using only Social Determinant Variables to Predict Profiles of US Adults having the Highest Health Expenditures
Abstract: Social determinants of health are defined as the social and physical conditions in which people are born, grow, live, work and age that impact health outcomes. In the late 1960s, Andersen developed a behavioral health framework to help shape a discussion of the impact of social determinants on medical services and other outcomes. Andersen and Newman acknowledged that some populations were not receiving, nor having access to, the same level of medical care as other populations. Our work focuses on social determinants and examining their impact on health expenditures of working aged US adults (20 – 59). We use longitudinal data that are nationally representative of the US adult working‐age civilian non‐institutionalized population. Our study includes Individuals who participated in the National Health Interview Study (NHIS), and who are included in the following two years of the Medical Expenditure Panel Study (MEPS). We form clusters based on the 2010 NHIS demographic, economic, and health‐related characteristics that are commonly used in studies of health care utilization. We use data from the 2010 NHIS cohort to create clusters using a clustering algorithm called Partitioning Around Medoids. Health expenditure distributions for this cohort are examined over the following two years. We validate the approach by applying the centers of the clusters to the 2008 and 2009 NHIS cohorts. Finally, we examine the effectiveness of these clusters in representing the top 5% of health care utilizers. Our findings show that these clusters can provide health care organizations a sampling approach to perform a first‐stage audit using a small segment of the population that can help identify the highest of the utilizers. The approach also identifies those who do not have health expenditures that could signal underutilization. While the profiles designed are representative of US adults, the approach can be applied to any population to reveal the impact of the profiles on utilization. Clusters formed using the data without comorbidities can profile new insureds to allow prospective management of certain individuals. The same group profiles can be used in multiple studies with different outcomes, such as inpatient, outpatient, or drug expenditures.
Bio: Margie Rosenberg, PhD, FSA is the Assurant Health Professor of Actuarial Science Professor at the University of Wisconsin-Madison. Margie’s research interests are in the application of statistical methods to health care, and applying her actuarial expertise to cost and policy issues in health care. Her recent research involves linking social determinants to outcomes such as (i) assessing the impact of delayed attention to oral health issues on emergency department visits and (ii) assessing the impact of unhealthy behaviors on perceived health status and predicting individuals with persistent high expenditures. Prior to her starting on her academic career, Margie worked as a life actuary for Allstate Life Insurance Company in Northbrook, IL.
February 5, 2021
Speaker: Sandra Paterlini
Title: Sorting out your investments: sparse portfolio selection via the sorted l1-norm
Abstract: We introduce a financial portfolio optimization framework that allows us to automatically select the relevant assets and estimate their weights by relying on a sorted l1-Norm penalization, henceforth SLOPE. To solve the optimization problem, we develop a new efficient algorithm, based on the Alternating Direction Method of Multipliers. SLOPE is able to group constituents with similar correlation properties, and with the same underlying risk factor exposures. Depending on the choice of the penalty sequence, our approach can span the entire set of optimal portfolios on the risk-diversification frontier, from minimum variance to the equally weighted. Our empirical analysis shows that SLOPE yields optimal portfolios with good out-of-sample risk and return performance properties, by reducing the overall turnover, through more stable asset weight estimates. Moreover, using the automatic grouping property of SLOPE, new portfolio strategies, such as sparse equally weighted portfolios, can be developed to exploit the data-driven detected similarities across assets.
Bio: Sandra Paterlini is full professor at the University of Trento, Italy. From 2013 to 2018, she held the Chair of Financial Econometrics and Asset Management at EBS Universität für Wirtschaft und Recht, Germany. Before joining EBS, she was assistant professor in statistics at the Faculty of Economics at the University of Modena and Reggio E., Italy. From 2008 to 2012, she has been a long-term visiting professor at the School of Mathematics, University of Minnesota. Her research on financial econometrics, statistics, operational research and machine learning have been predominantly interdisciplinary and often with an applied angle. Her work experience as a business consultant in finance and as a collaborator of central banks, such as for European Central Bank, Deutsche Bundesbank and the Fed Cleveland, has given her valuable input to guide and validate her research. Furthermore, she spent many years abroad (US, Germany, UK, and Denmark) to broaden and improve her skills further and to establish an international network of collaborators. She has been a consultant on business projects related to style analysis, portfolio optimization and risk management.
Her latest research interests are on machine learning methods for asset allocation, network analysis, risk management and ESG.
February 12, 2021
Speaker: Indranil SenGupta
Title: A machine learning-driven crude oil data analysis, with applications in continuous-time quadratic hedging
Abstract: In this presentation, a refined Barndorff-Nielsen and Shephard (BN-S) model is implemented to find an optimal hedging strategy for commodity markets. The refinement of the BN-S model is obtained through various machine and deep learning algorithms. The refinement leads to the extraction of a deterministic parameter from the empirical data set. The analysis is implemented to the Bakken crude oil data and the aforementioned deterministic parameter is obtained for a wide range of data sets. With the implementation of this parameter in the refined model, it is shown that the resulting model performs much better than the classical stochastic models.
Bio: Indranil SenGupta is an Associate Professor at the Department of Mathematics at North Dakota State University (NDSU). He is currently the mathematics graduate program director at NDSU. He received his Ph.D. in mathematics from Texas A&M University in 2010. His research interests include mathematical finance, stochastic processes, and data-science. He was the Associate Editor-in-Chief of the journal Mathematics, 2014-2019. Currently, he is an associate editor in the area of finance and risk management for the Journal of Modelling in Management. He is in the editorial board for several other journals.
February 19, 2021
Speaker: Sergei Levendorskii
Title: Static and semi-static hedging as contrarian or conformist bets
Abstract: Once the costs of maintaining the hedging portfolio are properly takeninto account, semi-static portfolios should more properly be thought of as separate classes of derivatives, with non-trivial, model-dependent payoff structures. We derive new integral representations for payoffs of exotic European options in terms of payoffs of vanillas, different from the Carr-Madan representation, and suggest approximations of the idealized static hedging/replicating portfolio using vanillas available in the market. We study the dependence of the hedging error on a model used for pricing and show that the variance of the hedging errors of static hedging portfolios can be sizably larger than the errors of variance-minimizing portfolios. We explain why the exact semi-static hedging of barrier options is impossible for processes with
jumps, and derive general formulas for variance-minimizing semi-static portfolios. We show that hedging using vanillas only leads to larger errors than hedging using vanillas and first touch digitals. In all cases, efficient calculations of the weights of the hedging portfolios are in the dual space using new efficient numerical methods for calculation of the Wiener-Hopf factors and Laplace-Fourier inversion.
Bio: Dr. Levendorskii is a founding partner at Calico Science Consulting in Austin TX. Dr. Levendorskii has developed several models and methods used by the financial services industry. His areas of expertise are Lévy processes with heavy and semi-heavy tails, Financial Mathematics, Real Options, Stochastic Optimization, Applied Fourier Analysis, Spectral Theory, Degenerate Elliptic Equations, Pseudo-differential operators, Numerical methods, Insurance, Quantum Groups, and Fractional Differential Equations. Prior to Calico, he was Chair in Financial Mathematics and Actuarial Sciences, Department of Mathematics and Deputy Director of Institute of Finance, University of Leicester, United Kingdom. He holds a Doctor of Sciences in Mathematics from Academy of Sciences of the Ukraine and he also earned a PhD in Mathematics from Rostov State University."
February 26, 2021 * 9 am CDT
Speaker: Justin Sirigano
Title: Deep Learning Models of High-Frequency Financial Data
Abstract: We develop and evaluate deep learning models for predicting price movements in high-frequency data. Deep recurrent networks are trained on a large limit order book dataset from hundreds of stocks across multiple years. Several data augmentation methods to reduce overfitting are analyzed. We also develop and evaluate deep reinforcement learning models for optimal execution problems with limit order book data. "Optimal execution" is the problem of formulating, given an a priori determined order direction (buy or sell) and order size, the optimal adaptive submission strategy to complete the order at the best possible price(s).The performance of deep recurrent models is compared against other types of models trained with reinforcement learning, such as linear VAR models and feedforward neural networks.
Bio: Justin Sirignano is an Associate Professor at the Mathematical Institute at the University of Oxford, where he is a member of the Mathematical & Computational Finance and Data Science groups. He received his PhD from Stanford University and was a Chapman Fellow at the Department of Mathematics at Imperial College London. His research interests are in the areas of applied mathematics, machine learning, and computational methods.
March 5, 2021
Speaker: Xiaobai Zhu
Title: Cyclical Design for Target Benefit Pension Plan
Abstract: In this paper, we derived the optimal cyclical design of Target Benefit (TB) pension plan. We focused on the stability of the benefit payment, and formulated an optimal control problem using a regime-switching model. We drew a number of remarks to improve the readability of our explicit solution, and made simplifications to enhance the transparency of the risk sharing design. We provided a new yet natural interpretation for a commonly used parameter under the TB context. We highlighted that cautions must be made when studying TB design using optimal control theory. Our numerical result suggested that a 100/0 investment strategies is preferred for the robustness of TB design, and the risk sharing mechanism should include both counter- and pro-cyclical components.
Bio: For my personal information, my full name is Xiaobai Zhu, I am assistant professor at School of Insurance, Southwestern University of Finance and Economics, China, my research interest is on hybrid pension plans and longevity modelling.
Speaker: David Matteson, Affiliation: Cornell
Title: Dynamic Shrinkage ProcessesAbstract: We propose a novel class of dynamic shrinkage processes for Bayesian time series and regression analysis. Building on a global–local framework of prior construction, in which continuous scale mixtures of Gaussian distributions are employed for both desirable shrinkage properties and computational tractability, we model dependence between the local scale parameters. The resulting processes inherit the desirable shrinkage behaviour of popular global–local priors, such as the horseshoe prior, but provide additional localized adaptivity, which is important for modelling time series data or regression functions with local features. We construct a computationally efficient Gibbs sampling algorithm based on a Pólya–gamma scale mixture representation of the process proposed. Using dynamic shrinkage processes, we develop a Bayesian trend filtering model that produces more accurate estimates and tighter posterior credible intervals than do competing methods, and we apply the model for irregular curve fitting of minute‐by‐minute Twitter central processor unit usage data. In addition, we develop an adaptive time varying parameter regression model to assess the efficacy of the Fama–French five‐factor asset pricing model with momentum added as a sixth factor. Our dynamic analysis of manufacturing and healthcare industry data shows that, with the exception of the market risk, no other risk factors are significant except for brief periods. If time permits, we will also highlight extensions to change point analysis and adaptive outlier detection.
Youtube link to Presentation
Speaker: John Dodson, Options Clearing Corporation
Title: Trends in applied mathematics and its adoption in the finance industry, or why you should pass on blockchains and big data
Abstract: Over the course of the twentieth century, applied mathematics has gradually assimilated and standardized the subjects of probability, statistics, control, and information. While an outside observer of decadal trends in STEM in finance might instead focus on the industry's embrace of computing technology during the Moore's Law era, I claim these quieter developments are ultimately more impactful because they help firms to organize information technology and financial innovation to create lasting value for clients. I will demonstrate this through a survey of the changing role of quants, and make an attempt to describe current opportunities.
Speaker: Shae Armstrong, Optum
Title: Quantifying the Impact of the Social Determinants of Health in the Covid-19 Era
Abstract: The Social Determinants of Health (SDoH) are key factors in each person’s environment and life that influence clinical outcomes of their health and wellbeing. These factors include, but are not limited to, income, housing, food security, education, and geography. In the age of Covid-19, understanding these factors and how they correlate to each other is more important than ever. Once we as industry gain insight on these clinical and financial impacts, we need to translate that insight into policy to mitigate root cause issues to better serve patients across the country.
During this lecture we lay the foundation by defining what the Social Determinants of Health are and the various categories they fall into. We will also examine what data sources feed various SDoH models and limitations of said data sources. Next we will conduct a deep-dive examination on a variety of case studies and models aimed at quantifying the short-term and long-term clinical and financial impact of Covid-19. From there we will touch on the future and impact of healthcare data analytics within the healthcare industry and as human beings navigating an unprecedented pandemic.
Speaker: Zhiguang (Gerald) Wang, South Dakota State University
Title: Multi-Step Forecast of Implied Volatility Surface using Deep Learning
Abstract: Modeling implied volatility surface (IVS) is of paramount importance to price and hedge an option. We contribute to the literature by modeling the entire IVS using recurrent neural network architectures, namely Convolutional Long Short Term Memory Neural Network (ConvLSTM) to produce multivariate and multi-step forecasts of the S&P 500 implied volatility surface. Using the daily S&P 500 index options from 2002 to 2019, we benchmark the ConvLSTM model against traditional multivariate time series VAR model, VEC model, and LSTM neural network. We find that both LSTM and ConvLSTM can fit the training data extremely well with mean absolute percentage error (MAPE) being 3.56% and 3.88%, respectively. As for out-of-sample data, the ConvLSTM (8.26% ) model significantly outperforms traditional time series models as well as the LSTM model for a 1-day, 30-day, and 90-day horizon, for all moneyness groups and contract months of both calls and puts.
Speaker: Liban Mohammed, University of Wisconsin -Madison
Title: Efficient Risk-sensitivity Estimation for Equity-Linked Insurance Benefits
Abstract: For an organization with billions of dollars in assets, precise risk management is necessary to safeguard those assets. However, when the risks these assets are exposed to depend on the future performance of equities in complex ways, directly estimating them in real-time to the necessary precision can be prohibitively expensive. This talk discusses some approaches to resolving this tension via metamodeling techniques.
YouTube Link to Presentation
Speaker: Max Rudolph, Rudolph Financial
Title: Actuarial Implications of COVID-19
Abstract: COVID-19 has had a material impact on all practice areas of the actuarial profession, ranging widely include traditional areas like health and mortality claims, assets and economic activity, but also risk management and strategic planning. This session assumes you know many of the basic statistics and provides observations about how analysis of the virus is evolving. Bio: MAX J. RUDOLPH, FSA CFA CERA MAAA. Max Rudolph is a credentialed actuary, active in the Asset-Liability Management and Enterprise Risk Management space for many years. He was named a thought leader in ERM within the actuarial profession, chaired the ERM Symposium, the SOA Investment Section Council and the SOAs Investment Actuary Symposium. He is a past SOA board member and received a Presidential Award for his role developing the CERA credential. He was the subject matter expert for the original Investment and ERM modules, wrote the ERM courseware document and has been involved with the actuarial professions climate change and pandemic efforts. He is a frequent speaker at actuarial seminars and universities, and an award-winning author. For the past 14 years Max has led Rudolph Financial Consulting, LLC, an independent consulting practice, focusing its insurance practice on ERM and ALM consulting. He has completed projects relating to life, health, annuity, and casualty insurers. He is an adjunct professor for Creighton Universitys Heider School of Business, where he focuses on ERM and investment topics.Max has completed a number of well received research reports covering topics such as emerging risks, low growth, low interest rates, investments, systemic risk and ERM. Other topics he has written about include pandemics, ALM and value investing. Many of his papers can be found at www.rudolph-financial.com. He comments on a variety of risk topics from @maxrudolph on twitter.
YouTube Link to Presentation
Speaker: Kaisa Taipale, C.H. Robinson
Title: Pricing in Contractual Freight Compared to Finance
Abstract: In this talk, I’ll discuss the contractual freight business, in which a large shipper makes a contract with a company like CH Robinson to procure carriers (trucks) for their goods over the course of a year for a given rate, as opposed to using the volatile “spot” or transactional market. Because these year-long contracts aren’t legally binding, some shippers treat them more like an American option on the underlying price of freight — but this has game-theoretic economic consequences for the shipper! Dr. Taipale, Data Scientist at C.H. Robinson will also talk about the data science and mathematical skills that are important for her job at C.H. Robinson