Quantitative Finance Program Webinars
Zoom Link and Password: Request by e-mailing to Mrs. Laurie Dalessio (email@example.com)
Fall 2022 Semester
Ruodu Wang | 2 September | Model Aggregation for Risk Evaluation and Robust Optimization
Speaker: Ruodu Wang
Title: Model Aggregation for Risk Evaluation and Robust Optimization
Date/Time: 2 Sep, 2022, 10:00-11:00 AM
Abstract: We introduce a new approach for prudent risk evaluation based on stochastic dominance, which will be called the model aggregation (MA) approach. In contrast to the classic worst-case risk (WR) approach, the MA approach produces not only a robust value of risk evaluation but also a robust distributional model which is useful for modeling, analysis and simulation, independent of any specific risk measure. The MA approach is easy to implement even if the uncertainty set is non-convex or the risk measure is computationally complicated, and it provides great tractability in distributionally robust optimization. Via an equivalence property between the MA and the WR approaches, new axiomatic characterizations are obtained for a few classes of popular risk measures. In particular, the Expected Shortfall (ES, also known as CVaR) is the unique risk measure satisfying the equivalence property for convex uncertainty sets among a very large class. The MA approach for Wasserstein and mean-variance uncertainty sets admits explicit formulas for the obtained robust models, and the new approach is illustrated with various risk measures and examples from portfolio optimization.
Speaker Bio: Dr. Ruodu Wang is University Research Chair, Sun Life Fellow, and Professor of Actuarial Science and Quantitative Finance at the University of Waterloo in Canada. He received his PhD in Mathematics (2012) from the Georgia Institute of Technology, after completing his Bachelor (2006) and Master’s (2009) degrees at Peking University. He holds editorial positions in leading journals in actuarial science and mathematical economics, including Co-Editor of the European Actuarial Journal, and Co-Editor of ASTIN Bulletin - The Journal of the International Actuarial Association. His scientific work has appeared in academic journals and conferences in various other fields, such as Management Science, Operations Research, The Annals of Statistics, Journal of the Royal Statistical Society Series B, and NeurIPS. He is an affiliated member of RiskLab at ETH Zurich. Among other international awards and recognitions, he is the inaugural winner of the SOA Actuarial Science Early Career Award (2021) from the Society of Actuaries, and a Fellow of the Institute of Mathematical Statistics (elected 2022).
Stan Uryasev | 16 September | Factor Model of Mixtures and Linear Regression
Speaker: Stan Uryasev
Title: Factor Model of Mixtures and Linear Regression
Date/Time: 16 Sep, 2022, 10:00-11:00 AM
Abstract: Joint talk with Cheg Peng (PhD student at the AMS Dept. of Stony Brook University). This paper considers the problem of estimating conditional distribution conditioned on observing some factors. The quantile function of the conditional distribution is modeled by a mixture (linear combination) of some basis quantile functions. The weight of each basis quantile function is a nonlinear function of the factors. We use the spline function as a primary example. The spline function is adaptive to the data, and can be estimated with the model in one shot. The model calibration is formulated as a linear regression problem similar to quantile regression. It can be efficiently solved by convex and linear programming. Various types of constraints, such as cardinality of the factors in the model, and penalties can be included in the optimization problem. The quantile function has a flexible shape in both tail and body, and has an analytic expression. The calibration method can focus on the tail of the distribution by assigning higher weights to the tail confidence levels. We prove the asymptotic normality of the estimator in a special case. We also prove the equivalence of our calibration method to minimization of Continuous Probability Ranked Score (CRPS). The proposed approach is generalized to conditional distributions defined by CVaR (also known as Expected Shortfall, superquantile), expectile and other functions. The considered approach is based on Risk Quadrangle theory. We conducted numerical experiments and demonstrated the efficiency of the approach.
Speaker Bio: Stan Uryasev received his M.S. in Applied Mathematics from the Moscow Institute of Physics and Technology (MIPT), Russia, in 1979 and Ph.D. in Applied Mathematics from the Glushkov Institute of Cybernetics, Kiev, Ukraine in 1983. From 1979 to 1987 he held a research position at the Glushkov Institute. From 1988 to 1992 he was a Research Scholar at the International Institute for Applied System Analysis, Luxenburg, Austria. From 1992 to 1998 he held the Scientist position at the Risk and Reliability Group, Brookhaven National Laboratory, Upton, NY. From 1998 to 2019 he was the George and Rolande Willis Endowed Professor at the University of Florida, and the director of the Risk Management and Financial Engineering Lab. His research and teaching interests include quantitative finance, risk management, stochastic optimization, machine learning, and military operations research. His joint paper with Prof. Rockafellar on Optimization of Conditional Value-At-Risk in The Journal of Risk, Vol. 2, No. 3, 2000 is among the 100 most cited papers in Finance. Many risk management/optimization packages implemented the approach suggested in this paper (MATLAB implemented a toolbox).
Harvey Stein | 30 September | Model Invariants and Functional Regularization
Speaker: Harvey Stein
Title: Model Invariants and Functional Regularization
Date/Time: 30 Sep, 2022, 10:00-11:00 AM
Abstract:When modeling data, we would like to know that our models are extracting facts about the data itself, and not about something arbitrary, like the order of the factors used in the modeling. Formally speaking, this means we want the model to be invariant with respect to certain transformations. Here we look at different models and the nature of their invariants. We find that regression, MLE and Bayesian estimation all are invariant with respect to linear transformations, whereas regularized regressions have a far more limited set of invariants. As a result, regularized regressions produce results that are less about the data itself and more about how it is parameterized. To correct this, we propose an alternative expression of regularization which we call functional regularization. Ridge regression and lasso can be recast in terms of functional regularization, as can Bayesian estimation. But functional regularization preserves model invariance, whereas ridge and lasso do not. It is also more flexible, easier to understand, and can even be applied to non-parametric models.
Speaker Bio: Harvey Stein is a senior VP in the Labs group in Two Sigma. From 1993 to 2022, Dr. Stein was at Bloomberg, where he served as the head of several departments including Quantitative Risk Analytics, Counterparty and Credit Risk, Interest Rates Derivatives, and Quantitative Finance R&D. Harvey is well known in the industry, having published and lectured on credit risk modeling, financial regulation, interest rate and FX modeling, CVA calculations, mortgage backed security valuation, COVID-19 data analysis, and other subjects. Dr. Stein is on the board of directors of the IAQF, a board member of the Rutgers University Mathematical Finance program, an adjunct professor at Columbia University, and organizer of the IAQF/Thalesians financial seminar series. He's also worked as a quant researcher on the Bloomberg for President campaign. Harvey holds a Ph.D. in Mathematics from the University of California, Berkeley (1991) and a B.S. in Mathematics from Worcester Polytechnic Institute (1982).
Raja Velu | 14 October | Efficient Modeling of Cross-Sectional Asset Retursn (via Reduced-Rank Regression)
Speaker: Raja Velu
Title: Efficient Modeling of Cross-Sectional Asset Returns (via Reduced-Rank Regression)
Date/Time: 14 Oct, 2022, 10:00-11:00 AM
Abstract: Kelly, Pruitt and Su(2019) introduced a new modeling approach, Instrumented Principal Component Analysis(IPCA) , where factor loadings are related to asset characteristics. It was shown that five factors resulting from IPCA model explain the cross-sectional variation better than other widely used factor models through only ten asset characteristics. In this paper, we first show that the IPCA model is in the same framework as Seemingly Unrelated Reduced-Rank Regression (RRSUR). By incorporating the cross-sectional dependence in the error term, we demonstrate that higher Sharpe ratios are obtained. The RRSUR model can accommodate serial dependence and its sparse version can produce parsimonious models. The talk will be balanced between theory and empirical results.
Speaker Bio: Raja Velu graduated from University of Wisconsin-Madison in 1983; he taught in the UW system and at Syracuse University. He has been a visiting professor at Stanford University (2005-2016) in the statistics department. He has co-authored two books: Multivariate Reduced-Rank Regression ( Springer-Verlag) and Algorithmic Trading and Quantitative Strategies(Taylor & Francis). Raja is also affiliated with leading technology and financial companies; he served as the Forecasting lead at Yahoo! in their sponsored search and competitive intelligence areas. At Microsoft Research, he collaborated with researchers in the Search Labs to develop a Forecasting and recommendation system for high-end products called Prodcast. He also worked at IBM-Almaden and Google at Silicon Valley. Raja served as JPMC Faculty Fellow and spent a year with JPMC’s Electronic Client Solutions group in NYC. His recent position was with Google as a Visiting Researcher working in the Forecasting area. He is now a Professor Emeritus at Syracuse University.
Marco C. Campi | 28 October | Universal Distributions for Empirical CVaR and Other Risk Measures
Speaker: Marco C. Campi
Title: Univeral Distributions for Empirical CVaR and Other Risk Measures
Date/Time: 28 Oct, 2022, 10:00-11:00 AM
Abstract: We consider Conditional Value at Risk (CVaR) and its optimization based on a record of historical data (empirical CVaR). It is shown that the probability distribution of the excess of threshold is universal, i.e., it does not depend on the distribution of the rates-of-return. This has important practical implications, for example in portfolio selection one can obtain quantitative risk evaluations that hold independently of the validity of any prior. Further, we show that the result for CVaR is just the point of the iceberg, and universal evaluations can be applied to various risk measures. This research unveils deep-seated connections between two concepts: risk and complexity, and opens new avenues to highlight the double role of the data (that of designing and that of assessing the quality of the design) beyond what traditional statistics is able to do.
Speaker Bio: Marco Claudio Campi is a professor of control and inductive methods at the University of Brescia, Italy, where he has taught topics related to data-driven methods, systems theory and inductive reasoning for more than twenty-five years. He has been a distinguished lecturer of the Control Systems Society and, until recently, he has served as the chair of the Technical Committee IFAC on Modeling, Identification, and Signal Processing. In 2008, he received the IEEE CSS George S. Axelby award for his article "The Scenario Approach to Robust Control Design". He has delivered plenary addresses at major conferences, including CDC, MTNS, and SYSID. Marco C. Campi is a Fellow of IEEE and a Fellow of IFAC. His research interests include: data-driven decision making, stochastic control, epistemology, and the foundations and interpretation of probability theory.
Valery Manokhin | 11 November | Machine Learning for Probabilistic Prediction
Speaker: Valery Manokhin
Title: Machine Learning for Probabilistic Prediction
Date/Time: 11 Nov, 2022, 10:00-11:00 AM
Abstract: Abstract: Prediction is the key objective of many machine learning applications. Accurate, reliable, and robust predictions are essential for optimal and fair decisions by downstream components of artificial intelligence systems, especially in high-stakes applications, such as personalised health, self-driving cars, finance, new drug development, forecasting of election outcomes and pandemics. Many modern machinelearning algorithms output overconfident predictions, resulting in incorrect decisions and technology acceptance issues. Classical calibration methods rely on artificial assumptions and often result in overfitting, whilst modern calibration methods attempt to solve calibration issues by modifying components of blackbox deeplearning systems. While this provides a partial solution, such modifications do not provide mathematical guarantees of predictions validity, are intrusive, complex, and costly to implement. Conformal prediction is a powerful and flexible framework that provides solutions to many of the Uncertainty Quantification problems regardless of the data distribution whilst providing robust mathematical guarantees of prediction validity.
Speaker Bio: Dr. Valery Manokhin is the Head of Data Science at ‘’Open Predictive Technologies”. He received PhD in Machine Learning (2022) from Royal Holloway, University of London having conducted research specialising in probabilistic and conformal prediction supervised by Prof. Vladimir Vovk. He holds a number of advanced MSc degrees including from UCL (Computational Statistics and Machine Learning), University of Sussex (Quant Finance) and an MBA from the University of Warwick. He has published in the leading machine learning journals, including ‘Neurocomputing’, ‘Journal of Machine Learning Research’ and ‘Machine Learning Journal’. He is the creator of 'Awesome Conformal Prediction' - the most comprehensive professionally curated resource on Conformal Prediction that has over 700 stars that has been featured in Kevin Murphy's bestselling book 'Probabilistic Machine Learning: An Introduction.'
Fabio Mercurio | 2 December | Libor Transition: Looking Forward to Backward-Looking Rates
Speaker: Fabio Mercurio
Title: Libor Transition: Looking Forward to Backward-Looking Rates
Date/Time: 2 Dec, 2022, 10:00-11:00 AM
Abstract: In this talk, we define and model forward risk-free term rates, which appear in the payoff definition of derivatives and cash instruments, based on the new interest-rate benchmarks that are replacing IBORs globally. We show that the classic interest-rate modeling framework can be naturally extended to describe the evolution of both the forward-looking (IBOR-like) and backward-looking (setting-in-arrears) term rates using the same stochastic process. We then introduce an extension of the most popular interest rate model, that is the LIBOR Market Model (LMM), to backward-looking rates. This extension, which we call generalized forward market model (FMM), completes the LMM by providing additional information about the rate dynamics between fixing/payment times, and by implying dynamics of forward rates under the classic money-market measure.
Speaker Bio: Fabio Mercurio is the global head of Quantitative Analytics at Bloomberg LP, New York. His team is responsible for the research on and implementation of cross-asset analytics for derivatives pricing, XVA valuations and credit and market risk. Fabio is also an adjunct professor at NYU. He has co-authored the book "Interest rate models: theory and practice" and published extensively in books and international journals, including 20 cutting-edge articles in Risk Magazine. Fabio is the recipient of the 2020 Risk quant of the year award.
Spring 2023 Semester
Johannes Royset | 3 February | Rockafellian functions: The most important concept in optimization that you haven't heard of
Speaker: Johannes Royset
Title: Rockafellian functions: The most important concept in optimization that you haven’t heard of
Date/Time: 3 Feb, 2022, 10:00-11:00 AM
Abstract: Rockafellian functions are central to sensitivity analysis, optimality conditions, algorithmic developments, and duality. They encode an embedding of an actual problem of interest within a family of problems and lead to broad insights and computational possibilities. We introduce Rockafellians and illustrate their application in stochastic optimization,machine learning, and outlier detection. Through Rockafellian relaxation, we are able to explore a decision space broadly and discover solutions that remain hidden for more conservative approaches to decision making under uncertainty such as distributional robust optimization.
Speaker Bio: Dr. Johannes O. Royset is a Professor of Operations Research at the Naval Postgraduate School. His research focuses on formulating and solving stochastic and deterministic optimization problems arising in data analytics, sensor management, and reliability engineering. Dr. Royset is the co-inventor of epi-splines, a functional approximation tool with wide applications in data fitting and forecasting, and of superquantile regression, second-order superquantile risk, and buffered probability. He was awarded a National Research Council postdoctoral fellowship in 2003, a Young Investigator Award from the Air Force Office of Scientific Research in 2007, and the Barchi Prize as well as the MOR Journal Award from the Military Operations Research Society in 2009. He received the Carl E. and Jessie W. Menneken Faculty Award for Excellence in Scientific Research in 2010 and the Goodeve Medal from the Operational Research Society in 2019. Dr. Royset was a plenary speaker at the International Conference on Stochastic Programming (2016), the SIAM Conference on Uncertainty Quantification (2018), and the INFORMS Security Conference (2022). He has a Doctor of Philosophy degree from the University of California at Berkeley (2002). Dr. Royset has been an associate or guest editor of SIAM Journal on Optimization, Operations Research, Mathematical Programming, Journal of Optimization Theory and Applications, Naval Research Logistics, Journal of Convex Analysis, Set-Valued and Variational Analysis, and Computational Optimization and Applications. He has published more than 100 papers and two books.
Ali Hirsa | 17 February | TBD
Speaker: Ali Hirsa
Date/Time: 17 Feb, 2023, 10:00-11:00 AM
Rita D'Ecclesia | 3 March | Firms’ profitability and ESG score: a machine learning approach
Speaker: Rita D'Ecclesia
Title: Firms’ profitability and ESG score: a machine learning approach
Date/Time: 3 March, 2023, 10:00-11:00 AM
Abstract: Existing literature pointed out that corporate social responsibility (CSR) has a potential impact on the performance of firms. However, the literature provides limited evidence of the relationship between non-financial indicators, such as the ESG score, and the firm’s profitability, often measured by the earnings before interest and taxes (EBIT). We aim to study the relationship between ESG scores and the firm’s profitability by analyzing a sample of about 400 companies listed in the EuroStoxx-600 from 2011 to 2021 using machine learning techniques. The novelty of our contribution lies in assessing whether the ESG score significantly impacts the firms’ profitability. The virtuous circle between ESG investments and the firms’ success is not trivial. We show that only massive investment in sustainability and ESG criteria, implying higher ESG scores, lead to successful objectives by enhancing the strength of a company’s balance sheet. Poor commitments to bind ESG elements into an investment strategy do not create extra profit. We could codify a sort of ESG sentiment that contributes to the profitability of the investments. Currently, the higher ESG score of the financial investment could contribute to determining the profitability of the firm business. For instance, the role of the banks and other institutions as transmitters of political-economic impulses on environmental issues is crucial, and they implement an adequate set of incentives to support lending to green projects. Our outcomes are in light of the new theories on the market participants’ expectations about implementing the climate policies (climate sentiments). The relationship between ESG score and EBIT is deepened using interpretability techniques, as the SHAP, the partial dependence plots, individual conditional expectation and LIME which help to visualize the functional relationship between the predicted response and one or more features. Our findings show that the model can reach high levels of accuracy in detecting the relationship between EBIT and ESG score. We find that the ESG score has a promising predictive ability, higher than other traditional accounting variables.
Speaker Bio: Rita Laura D’Ecclesia is Professor of Quantitative Methods at the Sapienza University of Rome, Chair of the Euro Working Group for Commodities and Financial Modelling, Director of the Ph.D. programme in Finance and Institutions at Sapienza and Chair of the International Summer School on Risk Measurement and Control. She teaches Financial Math, Asset Pricing, and Risk Management at undergraduate, graduate, and Ph.D. levels. She has broad expertise and a deep understanding of financial markets, price modeling, optimization techniques, and energy market dynamics. She has served as independent expert on the Board of major Italian Banks since 2016, Igea Banca, and Banco BPM. Currently, she is Deputy Chairman of the Board of Banco Monte Dei Paschi di Siena. Her research interests include portfolio optimization, pricing financial securities and commodities, responsible investment, and using AI to model financial markets.
Markus Pelger | 24 March | TBD
Speaker: Markus Pelger
Date/Time: 24 March, 2023, 10:00-11:00 AM
TBD | 7 April | TBD
TBD | 21 April | TBD
Spring 2022 Semester
Vladimir Bugera | 4 February | Analytics and Big Data in Real Estate
Title: Analytics and Big Data in Real Estate
Date/Time: 4 Feb, 2022, 10:00-11:00 AM
Abstract: The real estate industry accounts for a significant part of Gross Domestic Product for any country. The United States has the most developed and most dynamic real estate market in the world. The real estate industry is naturally data-driven. Significant technological advancements happen today in leveraging analytics and big data for real estate. We will review the process of how data is used in real estate transactions today and will discuss several problem formulations for data science and mathematical programming applications.
Speaker Bio: Dr. Bugera is the founder and Data Scientist of Big Data Realty, an analytics real estate company in Miami, FL. Additionally, Vladimir is a faculty member at College of Communications at the University of Miami, where he teaches data analytics for communications and journalism students, and the Commissioner of Public Utilities Commission in the City of North Miami Beach. Prior to his career in real estate analytics, Dr. Bugera spent 15+ years with financial companies in various managerial roles leading development of business intelligence analytics and risk management capabilities. Vladimir holds Ph.D. in Quantitative Finance and M.Sc. in Industrial Engineering from the University of Florida, as well as M.Sc. and B.Sc. in Applied Mathematics and Physics from Moscow Institute of Physics and Technology.
Dirk Tasche | 18 February | Calibrating Sufficiently
Speaker: Dirk Tasche
Title: Calibrating Sufficiently
Date/Time: 18 Feb, 2022, 10:00-11:00 AM
Abstract: When probabilistic classifiers are trained and calibrated, the so-called grouping loss component of the calibration loss can easily be overlooked. Grouping loss refers to the gap between observable information and information actually exploited in the calibration exercise. We investigate the relation between grouping loss and the concept of suffi ciency, identifying comonotonicity as a useful criterion for su fficiency. We revisit the probing reduction approach of Langford and Zadrozny (2005, 'Estimating Class Membership Probabilities using Classifier Learners') and find that it produces an estimator of probabilistic classifiers that reduces grouping loss. Finally, we discuss Brier curves as tools to support training and 'sufficient' calibration of probabilistic classifiers.
Speaker Bio Dirk Tasche is a senior risk manager in the banks division of the Swiss Financial Market Authority FINMA. His duties primarily consist of reviews of banks’ internal rating systems and the regulatory recognition of rating agencies. Before joining FINMA, he worked for the Prudential Regulation Authority (PRA), Lloyds Banking Group and Fitch Ratings in London and the Deutsche Bundesbank in Frankfurt. Dirk holds a doctorate in probability theory from Berlin University of Technology. He has published a number of papers on quantitative risk management and machine learning.
Vadim Omelchenko | 4 March | Optimal Balancing of Wind Parks with Virtual Power Plants
Speaker: Vadim Omelchenko
Title: Optimal Balancing of Wind Parks with Virtual Power Plants
Date/Time: 4 Mar, 2022, 10:00-11:00 AM
Abstract: In this paper, we explore the optimization of virtual power plants (VPP), consisting of a portfolio of biogas power plants and a battery whose goal is to balance a wind park while maximizing their revenues. We operate under price and wind production uncertainty and in order to handle it, methods of machine learning are employed. For price modeling, we take into account the latest trends in the field and the most up-to-date events affecting the day-ahead and intra-day prices. The performance of our price models is demonstrated by both statistical methods and improvements in the profits of the virtual power plant. Optimization methods will take price and imbalance forecasts as input and conduct parallelization, decomposition, and splitting methods in order to handle sufficiently large numbers of assets in a VPP. The main focus is on the speed of computing optimal solutions of large-scale mixed-integer linear programming problems, and the best speed-up is in two orders of magnitude enabled by our method which we called Gradual Increase.
Speaker Bio Vadim works as a senior optimizer at Alpiq, where I develop optimization models. I also used to work in consulting and bank sector. He has a Ph.D. from Charles University in Prague (Faculty of Mathematics and Physics). His research area covers the following areas: stochastic optimization, mixed integer programming, and quantitative analysis.
Giovanni Faonte | 1 April | New Frontiers in Deep Learning and Quantitative Finance: An Overview
Speaker: Giovanni Faonte
Title: New Frontiers in Deep Learning and Quantitative Finance: An Overview
Date/Time: 1 Apr, 2022, 10:00-11:00 AM
Abstract: The field of artificial intelligence has been revolutionized in the recent years by the staggering successes of deep and reinforcement learning in computer vision, natural language processing and strategies-learning. The impact on several industries has been substantial and has introduced a new paradigm around automation based on data and learned systems. Recently, a substantial amount of research has been dedicated to automation and optimization problems in quantitative finance with the promises of revolutionizing its computational and fidelity aspects. In my talk I will give an overview of some the most promising applications of deep and reinforcement learning to finance from the practitioner perspective, highlighting the current status of progress, discussing remaining limitations and bottlenecks and connecting the classical and formal approaches of quantitative finance with the ones brought by artificial intelligence.
Speaker Bio Giovanni Faonte is a lead researcher in AI at Goldman Sachs focused on applications of deep neural architectures to Quantitative Finance, NLP and Anomaly Detection. He received a PhD in Mathematics from Yale University and held (post-graduate) research positions at Kavli IPMU and Max Planck Institute for Mathematics. Prior to Goldman, he has worked as a quantitative risk-manager and as a machine learning research engineer developing perception stack for autonomous driving systems based on deep neural networks.
Leonard MacLean and Yonggan Zhao | 15 April | A Generalized-Entropy Approach to Portfolio Selection under a Hidden Markov Model
Speaker: Leonard MacLean and Yonggan Zhao
Title: A Generalized-Entropy Approach to Portfolio Selection under a Hidden Markov Model
Date/Time: 15 Apr, 2022, 10:00-11:00 AM
Abstract: This paper develops a dynamic portfolio selection model incorporating economic uncertainty for business cycles. Assuming market risk is characterized by the overall equity market and volatility indices, we develop an entropy-based dynamic portfolio selection model constrained with wealth surplus greater than or equal to shortfall over an expected target return. In the empirical analysis, we use the Select Sector ETFs to test the asset pricing model and examine the portfolio performance. Asset returns are projected by a regime-switching regression model on the two market risk factors. Weekly economic data from January 1999-December 2017 is employed for the estimation of economic state transitions, while the out-of-sample period from January 2018-February 2022 is used for portfolio performance test. It is found that, under both the empirical Sharpe and return-to-entropy ratios, the dynamic portfolio under the proposed strategy is much improved in contrast with the mean-variance models.
Speaker Bio Dr. Leonard MacLean is Professor Emeritus in the Rowe School of Business at Dalhousie University in Halifax, Canada. Dr. MacLean has held visiting appointments at Cambridge University, University of Bergamo, University of British Columbia, Simon Fraser University, Royal Roads University, University of Zimbabwe, and University of Indonesia. From 1989 to 1995 he served as Director of the School of Business Administration at Dalhousie University. Professor MacLean’s research focuses on stochastic models in finance, systems reliability in transportation, and sports analytics. He has authored more than 100 papers and 6 books. This work has been funded by grants from the Natural Sciences and Engineering Council of Canada and the Herbert Lamb Trust. Dr. MacLean teaches in the areas of statistics and operations management. Dr. Yonggan Zhao is Professor of Finance in the Rowe School of Business. In 2001, he received his PhD degree from University of British Columbia. He worked at Nanyang Technological University in Singapore from 2001 to 2006 and joined Dalhousie University as Canada Research Chair (Tier II) in Risk Management in 2006. He has published research papers in the areas of Economics, Finance, Management Science, and Mathematical Finance. His research has been supported by NSERC and SSHRC grants. He is Associate Editors of Journal of Management Mathematics and Journal of Economics and Administrative Sciences. His current research interests include optimal capital allocation, derivatives pricing, and risk management.
Kiseop Lee | 29 April | Optimal Execution with Liquidity Risk in a Diffusive Order Book Market
Speaker: Kiseop Lee
Title: Optimal Execution with Liquidity Risk in a Diffusive Order Book Market
Date/Time: 29 Apr, 2022, 10:00-11:00 AM
Abstract: We present an optimal order placement strategy with the presence of a liquidity cost. In this problem, a stock trader wishes to clear her large inventory by a predetermined time horizon T. A trade can be placed using both limit and market orders, and a large market order faces an adverse price movement caused by the liquidity risk. First, we study a single period model where the trader places limit orders and/or market orders at the beginning. We show the behavior of optimal amount of market order, m∗, and optimal placement of limit order,y∗, under different market conditions. Next, we extend to a multi-period model, where the trader can cancel the non-executed limit orders and resubmit market or limit orders to clear the remaining inventory at multiple time steps.
Speaker Bio: Professor Kiseop Lee is the director of Data Science in Finance program at Purdue University, and an associate professor of statistics. Kiseop Lee's research includes stochastic models, liquidity risk, information asymmetry, and machine learning application in high-frequency data problems. He is an associate editor of six professional journals for mathematics, statistics and financial engineering. He has numerous papers published in academic journals such as the Journal of Banking and Finance, Journal of Futures Markets, Quantitative Finance and Cutting Edge in Risk, which is a top practitioner’s journal. He has worked as a consultant at Invest.
Fall 2021 Semester
Cheng Peng, Drew Kouri, and Stan Uryasev | 3 September 2021 | Optimal Design of Experiments Focused on Distribution Tails
Speaker: Cheng Peng, Drew Kouri, and Stan Uryasev
Title: Optimal Design of Experiments Focused on Distribution Tails
Date/Time: 3 Sep. 2021, 10:00-11:00 AM
Abstract: This paper studies an optimal design problem with unsymmetric, possibly heavy tailed distributions. Uncertainties at the point of interest are estimated with quantiles. We provide upper bounds for quantiles by using Conditional Value-at-Risk. Quantile regression is used for the estimation of these upper bounds. We do not assume a large number of experiments (i.e., it is not an asymptotic convergence result). Although the design is focused on one tail of the distribution, allocation of tests is similar to the traditional C-optimal design. A case study on acoustic testing demonstrates the suggested methodology.
Bruno Dupire | 17 September 2021 | Functional Itô Calculus and the Characterization of Attainable Claims
Speaker: Bruno Dupire, Head of Quantitative Research at Bloomberg LP
Title: Functional Itô Calculus and the Characterization of Attainable Claims
Date/Time: 17 Sep. 2021, 10:00-11:00 AM
Abstract: We extend some results of the Itô calculus to functionals of the current path of a process to reflect the fact that often the impact of randomness is cumulative and depends on the history of the process, not merely on its current value. We express the differential of the functional in terms of adequately defined partial derivatives to obtain an Itô formula. We develop an extension of the Feynman-Kac formula to the functional case and an explicit expression of the integrand in the Martingale Representation Theorem. We establish that under certain conditions, even path dependent options prices satisfy a partial differential equation in a local sense. We exploit this fact to find an expression of the price difference between two models and compute variational derivatives with respect to the volatility surface. We also characterize which contingent claims can be attained by dynamic trading of the underlying or by a static position in European options or both.
Speaker Bio: Bruno Dupire is head of Quantitative Research at Bloomberg L.P., which he joined in 2004. Prior to this assignment in New York, he has headed the Derivatives Research teams at Société Générale, Paribas Capital Markets and Nikko Financial Products where he was a Managing Director. He is best known for having pioneered the widely used Local Volatility model (simplest extension of the Black-Scholes-Merton model to fit all option prices) in 1993 and the Functional Itô Calculus (framework for path dependency) in 2009. He is a Fellow and Adjunct Professor at NYU and he is in the Risk magazine “Hall of Fame”. He is the recipient of the 2006 “Cutting edge research” award of Wilmott Magazine and of the Risk Magazine “Lifetime Achievement” award for 2008.
Rafael Frongillo | 1 October 2021 | Toward a General Theory of Elicitation Complexity
Speaker: Rafael Frongillo
Title: Toward a General Theory of Elicitation Complexity
Date/Time: 1 Oct. 2021, 10:00-11:00 AM
Abstract: In statistics, finance, machine learning, economics, and several other domains, the question arises of which statistical functionals are elicitable, meaning they can be expressed as the minimizer of expected loss for some loss function. Several important functionals are not elicitable, however, including many risk and uncertainty measures of interest. To circumvent this negative result, several authors have proposed notions of "higher-order" elicitability, wherein a d-dimensional functional is elicited, and then condensed to recover the functional of interest. The smallest possible value of d is called the elicitation complexity. We will discuss several related definitions as part of a unifying framework, and give upper and lower bounds on elicitation complexity for a broad class of risk and uncertainty measures. The talk will conclude with several interesting open problems.
Gary Kazantsev | 15 October 2021 | Machine Learning in Finance
Speaker: Gary Kazantsev, Head of Quant Technology Strategy, Office of the CTO at Bloomberg LP
Title: Machine Learning in Finance
Date/Time: 15 Oct. 2021, 10:00-11:00 AM
Abstract: Machine learning is changing our world at an accelerating pace. In this talk we will discuss the recent developments in how machine learning and artificial intelligence are changing finance, from a perspective of a technology company which is a key participant in the financial markets. We will give an overview and discuss the evolution of selected flagship Bloomberg ML and AI projects, such as sentiment analysis, question answering, social media analysis, information extraction and prediction of market impact of news stories. We will discuss practical issues in delivering production machine learning solutions to problems of finance, highlighting issues such as interpretability, privacy and nonstationarity. We will also discuss current research directions in machine learning for finance. We will conclude with a Q&A session.
Speaker Bio(https://www.techatbloomberg.com/people/gary-kazantsev/) Gary is the Head of Quant Technology Strategy in the Office of the CTO at Bloomberg. Prior to taking on this role, he created and headed the company’s Machine Learning Engineering group, leading projects at the intersection of computational linguistics, machine learning and finance, such as sentiment analysis of financial news, market impact indicators, statistical text classification, social media analytics, question answering, and predictive modeling of financial markets. Prior to joining Bloomberg in 2007, Gary had earned degrees in physics, mathematics, and computer science from Boston University. He is engaged in advisory roles with FinTech and Machine Learning startups and has worked at a variety of technology and academic organizations over the last 20 years. In addition to speaking regularly at industry and academic events around the globe, he is a member of the KDD Data Science + Journalism workshop program committee and the advisory board for the AI & Data Science in Trading conference series. He is also a co-organizer of the annual Machine Learning in Finance conference at Columbia University.
Sebastian Jaimungal | 29 October 2021 | Reinforcement Learning for Dynamic Convex Risk Measures
Speaker: Sebastian Jaimungal
Title: Reinforcement Learning for Dynamic Convex Risk Measures
Date/Time: 29 Oct. 2021, 10:00-11:00 AM
Abstract: We develop an approach for solving time-consistent risk-sensitive stochastic optimization problems using model-free reinforcement learning (RL). Specifically, we assume agents assess the risk of a sequence of random variables using dynamic convex risk measures and develop policy gradient update rules for non-stationary policies. To do this, we use a time-consistent dynamic programming principle and devise an actor-critic style algorithm using neural network structures to obtain optimal strategies. Finally, we demonstrate the performance and flexibility of our approach by applying it to several examples. This is joint work with Anthony Coache.
Speaker Bio: Prof. Jaimungal is a full Professor of Mathematical Finance at the University of Toronto's Department of Statistical Sciences where he is the Director of the Master of Finance and Insurance program. He is a fellow of the Fields Institute for Mathematical Sciences, Associate Member of University of Oxford’s Man Institute, a former chair of the SIAM activity group in Financial Mathematics & Engineering, and is on the editorial board of SIAM J. on Financial Mathematics and Quantitative Finance, among others. His current research interests span stochastic control and games, machine learning, and algorithmic trading.
Giulio Trigila and Dan Stefanica | 12 November 2021 | Normalizing Optimal Transport Flows and Time Series Analysis
Speaker: Giulio Trigila and Dan Stefanica
Title: Normalizing Optimal Transport Flows and Time Series Analysis
Date/Time: 12 Nov. 2021, 10:00-11:00 AM
Abstract: Normalizing Flows (NF) were originally proposed as a tool for density estimation and became extremely popular with the work by D. J. Rezende and S. Mohamed using Neural Network within the context of NF. After a brief introduction to the theory of Optimal Transport (OT), this talk will introduce Optimal Transport Normalizing Flows (OTNF), highlighting the advantages of integrating OT with NF leading to a computational tool for the removal of variability, conditional density estimation and simulation. A new, non-adversarial, formulation of OTNF is presented together with applications and numerical results relative to time series analysis.
Speaker Bio: Giulio Trigila has a background in Physics and Applied Mathematics with a focus in applied probability and numerical optimization. His main research concerns data driven algorithms for the solution of the optimal transport problem and its application to data analysis. He obtained a M.S. in Theoretical Physics at the University of Rome La Sapienza and a Ph.D. in Applied Mathematics from the Courant Institute of Mathematical Sciences. Subsequently he held an appointment as a Post-doctoral fellow at the Technical University of Munich.
Dacheng Xiu | 10 December 2021 | Predicting Returns with Text Data
Speaker: Dacheng Xiu
Title: Predicting Returns with Text Data (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3389884)
Date/Time: 10 Dec. 2021, 10:00-11:00 AM
Abstract: We introduce a new text-mining methodology that extracts information from news articles to predict asset returns. Unlike more common sentiment scores used for stock return prediction (e.g., those sold by commercial vendors or built with dictionary-based methods), our supervised learning framework constructs a score that is specifically adapted to the problem of return prediction. Our method proceeds in three steps: 1) isolating a list of terms via predictive screening, 2) assigning prediction weights to these words via topic modeling, and 3) aggregating terms into an article-level predictive score via penalized likelihood. We derive theoretical guarantees on the accuracy of estimates from our model with minimal assumptions. In our empirical analysis, we study one of the most actively monitored streams of news articles in the financial system--the Dow Jones Newswires--and show that our supervised text model excels at extracting return-predictive signals in this context. Information in newswires is assimilated into prices with an inefficient delay that is broadly consistent with limits-to-arbitrage (i.e., more severe for smaller and more volatile firms) yet can be exploited in a real-time trading strategy with reasonable turnover and net of transaction costs.
Speaker Bio: Dacheng Xiu is Professor of Econometrics and Statistics at the University of Chicago Booth School of Business. His current research focuses on developing machine learning solutions to big-data problems in empirical finance. Xiu’s work has appeared in the Journal of Finance, Review of Financial Studies, Econometrica, Journal of Political Economy, the Journal of the American Statistical Association, and the Annals of Statistics. He is a Co-Editor for the Journal of Financial Econometrics, an Associate Editor for the Review of Financial Studies, Journal of Econometrics, Management Science, Journal of Business & Economic Statistics, etc. He has received several recognitions for his research, including the Fellow of the Society for Financial Econometrics, Fellow of the Journal of Econometrics, the AQR Insight Award, and the Swiss Finance Institute Outstanding Paper Award. Xiu earned his PhD and MA in applied mathematics from Princeton University.
Spring 2021 Semester
Steven Kou | 5 February 2021 | The Economics of Stablecoins
Speaker: Steven Kou
Title: The Economics of Stablecoins
Date/Time: 5 Feb. 2021, Friday, 10:00 - 11:00 AM EST
Abstract: Stablecoins, which are cryptocurrencies pegged to other stable financial assets such as U.S. dollar, are desirable for payments within blockchain networks, whereby being often called the ``Holy Grail of cryptocurrency.'' However, existing cryptocurrencies are too volatile for these purposes. To understand the impact of the proposed coins on the speculative and non-speculative demands of cryptocurrencies, we study equilibrium with and without the stable coins. Based on option pricing theory in presence of jump risk and black-swan type events, we show the robustness of a particular design based on a dual-class structure that offers a fixed income crypto asset (class A coin), a stable coin (class A’ coin) pegged to a traditional currency, and leveraged investment instruments (class B and B’ coins).
Bio: Steven Kou is a Questrom Professor in Management and Professor of Finance at Boston University. Previously, he taught at National University of Singapore (from 2013 to 2018), Columbia University (from 1998 to 2014), University of Michigan (1996-1998), and Rutgers University (1995-1996). He teaches courses on FinTech and quantitative finance. Currently he is a co-area-editor for Operations Research and a co-editor for Digital Finance, and has served on editorial boards of many journals, such as Management Science, Mathematics of Operations Research, and Mathematical Finance. He is a fellow of the Institute of Mathematical Statistics and won the Erlang Prize from INFORMS in 2002. Some of his research results have been incorporated into standard MBA textbooks and have implemented in commercial software packages and terminals, e.g. in Bloomberg Terminals.
Emilio Lorente | 19 February 2021 | Machine Learning Based Asset Allocation Ranking
Speaker: Emilio Lorente
Title: Pricing Slow-Moving Debt Crises
Date/Time: 19 Feb. 2021, Friday, 10:00 - 11:00AM EST
Abstract: We will explore the possibility of a Machine Learning based asset allocation ranking, and its implementation through CVaR portfolio optimization. To achieve this, the covariance matrix is obtained from the Entropy-Pooling (Attilio Meucci, 2014) approach. We will see the end-to-end portfolio construction process in a real-life wealth mandate, where extreme value theory is applied to its risk assessment. The portfolio simulation coherently uses the Entropy-Pooling based covariance matrix to produce the expected cumulative distribution function of the portfolio.
Speaker Bio: Emilio Lorente is the founder and CIO of Recognition Asset Management Software, S.L. Emilio has more than 20 years of experience in the multi-asset management industry. He has developed profound work covering the entire investment cycle for Wealth (Lloyds Bank), Private Banking (Banco Popular) and Institutional Investors (Aberdeen Standard Investments). He is a designated expert on Machine Learning methods applied to markets. More importantly, he applied state of the art techniques to become a pioneer in the application of Machine Learning, Optimization Algorithms, High Performance Computing and Man-Machine collaboration technology in portfolio management. His academic background covers Masters Degrees in Economics, Quantitative Finance) and Artificial Intelligence. He is a certified member of the Global Association of Risk Professionals. He was invited as Senior Lecturer from different Universities in Spain, UK and Latin America to impart MSc and MBA courses in quantitative methods for Asset Allocation, Derivatives, and Risk Management.
Josef Teichmann | 5 March 2021 | Consistent Recalibration Models and Deep Calibration (joint work with Matteo Gambara)
Speaker: Josef Teichmann, Professor in the Department of Mathematics, ETH Zurich
Title: Consistent Recalibration Models and Deep Calibration (joint work with Matteo Gambara)`
Date/Time: 5 Mar. 2021, Friday, 13:00 - 14:00PM EST
Abstract: Consistent Recalibration models (CRC) have been introduced to capture in necessary generality the dynamic features of term structures of derivatives' prices. Several approaches have been suggested to tackle this problem, but all of them, including CRC models, suffered from numerical intractabilities mainly due to the presence of complicated drift terms or consistency conditions. We overcome this problem by machine learning techniques, which allow to store the crucial drift term's information in neural network type functions. This yields first time dynamic term structure models which can be efficiently simulated.
Speaker Bio: Josef Teichmann is a professor for Mathematical Finance at ETH Zurich since 2009. He holds a PhD from Vienna University in the area of infinite-dimensional geometry from 1999 and has worked as an Associate Professor at TU Vienna from 2002 to 2009. His research interests lie in Stochastic Finance, Stochastic Partial Differential Equations, Rough Analysis, and Machine Learning.
Stavros Zenios | 19 March 2021 | A Global Political Risk Factor and Hedging Politics in International Equities
Speaker: Stavros Zenios
Title: A Global Political Risk Factor and Hedging Politics in International Equities
Date/Time: 19 Mar. 2021, Friday, 10:00 - 11:00 AM
Abstract: Using novel measures of politics-policy uncertainty we document predictable variation in stock market returns across countries. We identify a global political risk factor (P-factor) commanding a significant risk premium. Augmenting the global market portfolio with the P-factor significantly reduces pricing errors and improves cross- sectional fit. We identify transmission channels through both cash flow and discount rate channels. We also show how to hedge political risks using portfolio optimization, and find that political risk goes some way towards explaining the home equity bias puzzle, but it does not solve it.
Speaker Bio: Stavros A. Zenios is Professor of Finance and Management Science at University of Cyprus, Senior Fellow at the Wharton School Financial Institutions Center (USA), and Non-resident Fellow of Bruegel (Brussels). He is a Member of the National Academy of Cyprus. He is the author of two books and numerous scholarly articles in leading international journals in risk management, financial engineering, and management science. He is currently working on sovereign debt sustainability issues, including the pricing of political stability and economic policy uncertainty risks. His work on personal financial planning received the 2006 EURO Excellence in Practice Award, and he also received awards for work on the performance of financial institutions. His work on financial modeling and robust optimization is cited extensively, and his book Practical Financial Optimization: Decision making for financial engineers (Blackwell-Wiley) is used in advanced classes in European and North -American Universities. His book with Patrick Harker on the Performance of Financial Institutions (Cambridge University Press) was translated in Chinese, and in 1997 he received the INFORMS prize for his book with Yair Censor Parallel Optimization (Oxford University Press). He served two terms as Rector of University of Cyprus, where he was the first Dean of the School of Economics and Management, and as president of UNICA-Universities of the European Capitals. He was vice-chairman of the Cyprus Council of Economic Advisors and on the Board of the Central Bank of Cyprus. He consulted extensively with international institutions and commercial enterprises, such as the World Bank, the European Stability Mechanism, Union Bank of Switzerland, METLIFE Insurance, BSI, among others.
Ariel Caticha | 2 April 2021 | Entropic Inference
Speaker: Ariel Caticha
Title: Entropic Inference
Date/Time: 2 Apr. 2021, Friday, 10:00 - 11:00 AM EST
Abstract: The concept of entropy has its origins in the 19 th century in the discovery of thermodynamics (Carnot, Clausius, Kelvin) and statistical mechanics (Maxwell, Boltzmann, Gibbs). A series of developments starting around the middle of the 20 th century (mostly due to Shannon and Jaynes) liberated the concept of entropy from its physics origins and elevated it into a general purpose tool for processing information. Thus was born the old Method of Maximum Entropy MaxEnt. In a parallel line of research Bayesian inference enjoyed a remarkable period of expansion in the latter half of the century. The two methods of inference flourished to large extent independently of each other entropic methods do not yet enjoy widespread acceptance within the orthodox Bayesian community. The connection between them has been an endless source of controversy and even their compatibility has been repeatedly brought into question. Further developments extending into the early 21 st century have, however, culminated in the complete unification of entropic and Bayesian inference methods. The goal of this lecture is to summarize the argument for the method of maximum entropy as the universal method of inference and show that it includes the old MaxEnt, all Bayesian methods, and the general subject of large deviations as special cases. The consequences of these methods of entropic inference are potentially enormous both for statistics (entropic priors, model selection, experimental design, etc.) and for science in general. In physics, for example, it is well known that the laws of thermodynamics can be derived from entropic methods. What might at first sight be surprising is that entropic principles can also be used to derive the dynamical laws of mechanics, both classical and quantum. Perhaps the laws of physics themselves are not laws of nature but merely rules to process information about the world.
Speaker Bio: Ariel Caticha is professor at the Department of Physics, University at Albany SUNY. He was born in Uruguay, and educated in Brazil (UNICAMP: BS and MS in physics) and in the USA (Caltech: PhD in physics). In recent years his research has focused on the connection between physics and information. Catichas papers on entropic inference and on its applications to the foundations of statistical mechanics, quantum mechanics, and general relativity can be found at https://www.albany.edu/physics/faculty/ariel-caticha . He also devotes a considerable effort to teaching physics having mentored 13 Ph.D. students (plus 6 ongoing). He has received the University at Albanys Excellence in Teaching and Advising Award and also the SUNY Chancellors Award for Excellence in Teaching.
Brent Lindquist | 16 April 2021 | Advanced Analytics for the Real Estate Investment Market
Speaker: Brent Lindquist
Title: Advanced Analytics for the Real Estate Investment Market
Date/Time: 16 APR. 2021, Friday, 10:00 - 11:00 AM EST
Abstract: Portfolio management encompasses four principal areas: optimization techniques to maximize return relative to risk; risk information and management tools to assess changing market conditions and actively manage optimization; backtesting to estimate exposure to risk; and portfolio insurance to hedge risk exposure. We consider each of these in the context of portfolios comprised of REIT ETFs. We discuss dynamic optimization, which provides stochastic forecasts based upon improved estimates of tail-risk. Backtest results demonstrate improvement resulting from dynamic optimization. We address early warning systems for potential market disruptions. Finally, we consider option pricing where the underlying asset is the managed portfolio.
Speaker Bio: W. Brent Lindquist is a computational applied mathematician. A theoretical physicist by training, his interests have included quantum electrodynamics, hyperbolic conservation laws, flow in porous media, and academic administration. He was active in the development of the Frontier software package and is the principal developer of the 3DMA-Rock package. He is now focusing on computational and mathematical finance.
Viktor Kuzmenko | 30 April 2021 | Expectile Risk Quadrangle
Speaker: Viktor Kuzmenko, Glushkov Institute of Cybernetics, Kiev, Ukraine
Title: Expectile Risk Quadrangle
Date/Time: 30 Apr. 2021, Friday, 10:00 - 11:00 AM EST
Abstract: Expectile is a risk measure used in financial applications, such as portfolio portfolio optimization. The expectile is a risk measure similar Value-at-Risk (VaR) and Conditional Value-at-Risk (CVaR). The Fundamental Risk Quadrangle theory describes general properties and relations for risk, deviation, error and regret functions of a random variable. So-called Regular Risk Quadrangles are known for VaR and CVaR. This presentation describes two new quadranges related to expectile and relevant statistical optimization problems.
Fall 2020 Semester
Yuichi Takano | 4 September 2020 | Dynamic Portfolio Selection with Linear Control Policies for Coherent Risk Minimization
Speaker: Yuichi Takano, Associate Professor, University of Tsukuba, Japan
Title: Dynamic Portfolio Selection with Linear Control Policies for Coherent Risk Minimization
Date/Time: 4 Sep. 2020, Friday, 10:00 - 11:00 AM EST
Abstract: This talk is concerned with a linear control policy for dynamic portfolio selection. We develop this policy by incorporating time-series behaviors of asset returns on the basis of coherent risk minimization. Analyzing the dual form of our optimization model, we demonstrate that the investment performance of linear control policies is directly connected to the intertemporal covariance of asset returns. To mitigate overfitting to training data (i.e., historical asset returns), we apply robust optimization. For this optimization, we prove that the worst-case coherent risk measure can be decomposed into the empirical risk measure and the penalty terms. Numerical results demonstrate that when the number of assets is small, linear control policies deliver good out-of-sample investment performance. When the number of assets is large, the penalty terms improve the out-of-sample investment performance. This is a joint work with Jun-ya Gotoh
Peter Cotton | 18 September 2020 | Collective PRediction, Copulas, and Space Filling Curves
Speaker: Peter Cotton, Chief Data Scientist, Intech Investments
Title: Collective Prediction, Copulas and Space Filling Curves
Date/Time: 18 September 2020, Friday, 10:00 - 11:00 AM EST
Abstract: What if you could write an algorithm once that finds its own way to interesting problems to solve, without the need for human intervention? What if anyone could predict any source of live data by writing three lines of Python code? I will discuss the potential for collective real-time prediction, and demonstrate a prototypical host at Microprediction.Org where algorithms predict everything from electricity prices to cryptocurrencies to helicopter trajectories.
Speaker Bio: Dr. Cotton heads Intech’s data science efforts in collaboration with the investment team and leads Intech’s crowdsourced research efforts. He was responsible for algorithmic trading, privacy preserving analytics and crowdsourcing at JPMorgan. Previously, he was the founder of Benchmark Solutions, a company that pioneered large-scale financial data assimilation and was later sold to Bloomberg. Peter began his career at Morgan Stanley where he was one of several independent inventors of closed-form synthetic CDO pricing. Dr. Cotton earned an undergraduate degree in physics and mathematics from the University of New South Wales and a PhD in mathematics from Stanford University.
Michele Leonardo Bianchi | 2 October 2020 | Forward-looking portfolio selection with multivariate non-Gaussian models
Speaker: Michele Leonardo Bianchi, Regulation and Macroprudential Analysis Directorate, Bank of Italy
Title: Forward-looking portfolio selection with multivariate non-Gaussian models (joint work with Gian Luca Tassinari, University of Bologna)
Date/Time: 2 October 2020, Friday, 10:00 - 11:00 AM EST
Abstract: In this study we suggest a portfolio selection framework based on time series of stock log-returns, option-implied information, and multivariate non-Gaussian processes. We empirically assess a multivariate extension of the normal tempered stable (NTS) model and of the generalized hyperbolic (GH) one by implementing an estimation method that simultaneously calibrates the multivariate time series of log-returns and, for each margin, the univariate observed one-month implied volatility smile. To extract option-implied information, the connection between the historical measure P and the risk-neutral measure Q, needed to price options, is provided by the multivariate Esscher transform. The method is applied to fit a 50-dimensional series of stock returns, to evaluate widely-known portfolio risk measures and to perform a forward-looking portfolio selection analysis. The proposed models are able to produce asymmetries, heavy tails, both linear and non-linear dependence and, to calibrate them, there is no need of liquid multivariate derivative quotes.
Speaker Bio: Michele Leonardo Bianchi is an Advisor in the Macroprudential Analysis Division at the Regulation and Macroprudential Analysis Directorate of the Banca d’Italia. From 2008 to 2014 he worked in the Division of Risk and Financial Innovation Analysis at the Specialized Intermediaries Supervision Department of the Banca d’Italia. He earned an Italian Laurea in Mathematics in 2005 from the University of Pisa and completed his PhD in Computational Methods for Economic and Financial Decisions and Forecasting in 2009 from the University of Bergamo. In 2017 he received the Italian national scientific qualification (associate professorship). He has extensive experience in the analysis and statistical modeling of financial data. He coauthored two book published by Wiley and World Scientific in quantitative finance and authored articles on quantitative finance, financial econometrics, probability theory, and nonlinear optimization published in international scientific journals, including Journal of Banking and Finance, Quantitative Finance, International Journal of Theoretical and Applied Finance, Theory of Probability and Its Applications, Probability and Mathematical Statistics, The Econometrics Journal, Computational Economics, and The Journal of Derivatives.
Chris Bemis | 16 October 2020 | Constraints and Investor Views in Portfolio Optimization
Speaker: Chris Bemis, Head of Quantitative Analysis and Research, Whitebox Advisors, US
Title: Constrainst and Investor Views in Portfolio Optimization
Date/Time: 16 October 2020, Friday, 10:00 - 11:00 AM EST
Abstract: We study the effect linear constraints have on risk in the context of mean variance optimization (MVO). Jagannathan and Ma (2003) establish an equivalence between certain constrained and unconstrained MVO problems via a modification of the covariance matrix. We extend their results to arbitrary linear constraints and provide alternative interpretations for the effect of constraints on both the input parameters to the problems at hand and why ex-post performance is improved in the constrained setting. In addition, we present a signal modification strategy similar in approach to that of Black-Litterman.
Speaker Bio: Dr. Bemis is Head of Quantitative Analysis and Research for Whitebox Advisors, focusing on cross-asset alpha drivers for a variety of asset classes. Most recently this work has yielded several systematic credit strategies for Whitebox Advisors. Dr. Bemis earned his PhD in applied mathematics from the University of Minnesota, where his work involved both modeling and optimization for portfolios of risky assets, and where he remains as an adjunct professor in their Math Finance program.
Abolfazl Safikhani | 30 October 2020 | Non-stationary Spatio-Temporal Modeling of COVID-19 Progression in the U.S.
Speaker: Abolfazl Safikhani, Department of Statistics, University of Florida
Title: Non-stationary Spatio-Temporal Modeling of COVID-19 Progression in the U.S
Date/Time: 30 October 2020, Friday, 10:00 - 11:00 AM EST
Abstract: The fast transmission rate of COVID-19 worldwide has made this virus the most important challenge of year 2020. Many mitigation policies have been imposed by the governments at different regional levels (country, state, county, and city) to stop the spread of this virus. Quantifying the effect of such mitigation strategies on the transmission and recovery rates, and predicting the rate of new daily cases are two crucial tasks. In this talk, we propose a modeling framework which not only accounts for such policies but also utilizes the spatial and temporal information to characterize the pattern of COVID-19 progression. Specifically, a piecewise susceptible-infected-recovered (SIR) model is developed while the dates at which the transmission/recovery rates change significantly are defined as “break points” in this model. A novel and data-driven algorithm is designed to locate the break points using ideas from fused lasso and thresholding. In order to enhance the forecasting power and to describe additional temporal dependence among the daily number of cases, this model is further coupled with spatial smoothing covariates and vector auto-regressive (VAR) model. The proposed model is applied to several U.S. states and counties, and the results confirm the effect of “stay-at-home orders” and some states’ early “re-openings” by detecting break points close to such events. Further, the model performed satisfactorily short-term forecasts of the number of new daily cases at regional levels by utilizing the estimated spatio-temporal covariance structures. This is a joint work with Yue Bai and George Michailidis.
Sajad Obaydi | 13 November 2020 | Constrained Optimization Problems in Co-insurance
Speaker: Sajad Obaydi, Business Actuary, The Hartford, London, UK
Title: Constrained Optimization Problems in Co-insurance
Date/Time: 13 November 2020, Friday, 10:00 - 11:00 AM EST
Abstract: Insurance companies that partake in the Lloyd’s of London market insure complex and exotic risks, which have the potential to produce multibillion-dollar losses. This is made possible by co-insurance, i.e. insurers sharing risks with one another. In this market, each insurer must decide what proportion of each contract it is willing to take on. Greater participation in a contract would mean a larger share of premium income, but at a risk of incurring greater loss. Therefore, by varying the degree of participation of each contract, an insurer can achieve an optimal portfolio that maximizes expected profit, while satisfying multiple commercial constraints. In the case study I will discuss the building blocks of such an optimization problem and the various constraints a practitioner may encounter. Finally, I will demonstrate an approach to solve this problem using modern optimization software.
Spring 2020 Semester
Robert J. Frey| 3 April 2020 | Heavy Tailed Distributions: Theory and Practice
Speaker: Robert J. Frey
Title: Heavy Tailed Distributions: Theory and Practice
Date/Time: 3 April 2020, Friday, 10:00 - 11:00 AM EST
Abstract: Dr. Robert J. Frey is a former Managing Director of Renaissance Technologies Corp (1992–2004) and presently serves as a Research Professor on the faculty of Stony Brook University where he is the Founder and Director of the Program in Quantitative Finance within the Department of Applied Mathematics and Statistics. He is the Founder, and Chief Executive Officer of global fund of hedge funds group FQS Capital Partners.
Wei Zhu | 10 April 2020 | Real-time Prediction of Bitcoin Bubble Crashes
Speaker: Wei Zhu (joined talk with Min Shu)
Title: Real-time Prediction of Bitcoin Bubble Crashes
Date/Time: 10 April 2020, Friday, 10:00 - 11:00 AM EST
Short Bio: Dr. Wei Zhu (http://www.ams.sunysb.edu/~zhu/) is a Professor in Statistics at the Department of Applied Mathematics and Statistics, Stony Brook University. Her research focuses on the areas of errors in variable regression, structural equation modeling, experimental design, and in recent years, time series regression and statistical learning, with applications in biomedicine and finance.
Abstract: In the past decade, Bitcoin as an emerging asset class has gained widespread public attention because of their extraordinary returns in phases of extreme price growth and their unpredictable massive crashes. We apply the log-periodic power law singularity (LPPLS) confidence indicator as a diagnostic tool for identifying bubbles using the daily data on Bitcoin price in the past two years. We find that the LPPLS confidence indicator based on the daily Bitcoin price data fails to provide effective warnings for detecting the bubbles when the Bitcoin price suffers from a large fluctuation in a short time, especially for positive bubbles. In order to diagnose the existence of bubbles and accurately predict the bubble crashes in the cryptocurrency market, this study proposes an adaptive multilevel time series detection methodology based on the LPPLS model and a finer (than daily) timescale for the Bitcoin price data. We adopt two levels of time series, 1 h and 30 min, to demonstrate the adaptive multilevel time series detection methodology. The results show that the LPPLS confidence indicator based on this new method is an outstanding instrument to effectively detect the bubbles and accurately forecast the bubble crashes, even if a bubble exists in a short time. In addition, we discover that the short-term LPPLS confidence indicator being highly sensitive to the extreme fluctuations of Bitcoin price can provide some useful insights into the bubble status on a shorter time scale – on a day to week scale, while the long-term LPPLS confidence indicator has a stable performance in terms of effectively monitoring the bubble status on a longer time scale – on a week to month scale. The adaptive multilevel time series detection methodology can provide real-time detection of bubbles and advanced forecast of crashes to warn of the imminent risk in not only the cryptocurrency market but also other financial markets.
Cristian Homescu | 17 April 2020 | FMachine Learning for Quantitative Investment and Wealth Management
Speaker: Cristian Homescu, Bank of America Merrill Lynch
Title: Machine Learning for Quantitative Investment and Wealth Management
Date/Time: 17 April 2020, Friday, 10:00 - 11:00 AM EST
Abstract: achine learning for quantitative investment and wealth management (QWIM): What is hype and what is reality? What differences are observed for ML applications in this area compared to ML applications in other areas? Within this context this presentation aims to provide an overview of ML applications (classification, network analysis and clustering, forecasting and prediction, etc.) in QWIM, while also discussing the practical challenges.
Presentation Participants of the webinar can get a copy of transparencies from Giorgi Pertaia (firstname.lastname@example.org)
Erick Delage | 24 April 2020 | Preference Robust Utility-based Shortfall Risk Minimization
Speaker: Erick Delage, HEC Montréal, Université de Montréal, Canada
Title: Preference Robust Utility-based Shortfall Risk Minimization
Date/Time: 24 April 2020, Friday, 10:00 - 11:00 AM EST
Abstract: Utility-based shortfall risk measure is proposed by Föllmer and Schied and has received increasing attentions over the past few years for its potential to quantify more effectively the risk of large losses than conditional value at risk. In this talk, we consider the case when the true utility/loss function cannot be specified either because there is missing information about how the decision maker perceives risk, or because he is simply hesitant about it. We propose a preference robust shortfall risk model that exploits empirical data about subjective judgements to construct a set of plausible utility-based loss functions and suggest minimizing shortfall risk as measured using the worst loss function from this set. We develop tractable reformulations when the underlying probability distribution is discrete. In the case when the probability distribution is continuous, we propose a sample average approximation scheme and show that its optimal solution and value converges to the true ones as the sample size increases.
Speaker Bio: Erick Delage is born in Montreal. He graduated in 2009 with a Ph.D. from Stanford University. He joined the Department of Decision Sciences at HEC Montréal and became full professor in 2019. Since 2014, he has been chairholder of the Canada Research Chair in decision making under uncertainty. His research interests span the areas of optimization, decision analysis, and artificial intelligence with applications in portfolio optimization, derivative pricing, resource allocation, inventory management, and energy planning problems.
Weilong Fu | 1 May 2020 | Fast Pricing of American Options Under Variance Gamma
Speaker: Weilong Fu (joined talk with Prof. Ali Hirsa), Columbia University IEOR Department
Title: Fast Pricing of American Options Under Variance Gamma
Date/Time: 1 May 2020, Friday, 10:00 - 11:00 AM EST
Abstract: We investigate methods for pricing American options under the variance gamma model. The variance gamma process is a pure-jump process which is constructed by replacing the calendar time by the gamma time in a Brownian motion with drift, which makes it a time-changed Brownian motion. In the case of the Black-Merton-Scholes model, there are fast approximation methods for pricing American options, but they cannot be utilized for the variance gamma model. We develop a new fast and accurate approximation method inspired by the quadratic approximation to get rid of the time steps required in finite difference methods and simulation methods, while reducing the error by making use of a machine learning technique on pre-calculated quantities. We compare the performance of our method with those of the existing methods and show that this method is efficient and accurate for practical use.
Speaker Bio: Weilong Fu is a Ph.D. student in Columbia IEOR Department. He graduated from Peking University (Beijing, China) in 2017, with a major in Statisitcs. Prof. Ali Hirsa is the Co-Director of Financial Engineering at Columbia University IEOR Department and the Managing Partner at Sauma Capital, LLC. Previously he was Managing Director and Global Head of Quantitative Strategy at DV Trading, LLC, a Partner and Head of Analytical Trading Strategy at Caspian Capital Management, LLC, and he worked in a variety of quantitative positions at Morgan Stanley, Bank of America Securities, and Prudential Securities. Ali’s research interests are algorithmic trading, machine learning, data mining, computational/quantitative finance and optimization; he is the author of “Computational Methods in Finance” and co-author of "An Introduction to Mathematics of Financial Derivatives", 3rd edition. Ali received his PhD in Applied Mathematics from University of Maryland at College Park under the supervision of Professors Howard C. Elman and Dilip B. Madan.
Harvey Stein | 8 May 2020 | A Unified Framework for Default Modeling
Speaker: Harvey Stein, Bloomberg LP
Title: A Unified Framework for Default Modeling
Date/Time: 8 May 2020, Friday, 10:00 - 11:00 AM EST
Abstract: Credit risk models largely bifurcate into two classes -- the structural models and the reduced form models. Attempts have been made to reconcile the two approaches by adjusting filtrations to restrict information, but they are technically complicated and tend to approach filtration modification in an ad-hock fashion. Here we propose a reconciliation inspired by actuarial science's approach to survival analysis. We model the survival curve and hazard rate curve as stochastic processes. This puts default models in a form resembling the HJM framework for interest rates, yielding a unified framework for default modeling. Predictability of default has a simple interpretation in this framework. The framework enables us to disentangle predictability and the distribution of the default time from calibration decisions such as whether to use market prices or balance sheet information. It supplies a formal framework for combining models, yielding a simple way to define new default models.
Speaker Bio: Dr. Harvey J. Stein is Head of the Quantitative Risk Analytics Group at Bloomberg, responsible for Bloomberg's market risk and credit risk models. Dr. Stein is well known in the industry, having published and lectured on mortgage backed security valuation, CVA calculations, interest rate and FX modeling, credit exposure calculations, financial regulation, and other subjects. Dr. Stein is also on the board of directors of the IAQF, an adjunct professor at Columbia University, a board member of the Rutgers University Mathematical Finance program and of the NYU Enterprise Learning program, and organizer of the IAQF/Thalesians financial seminar series. He received his BA in mathematics from WPI in 1982 and his PhD in mathematics from UC Berkeley in 1991.
Agostino Capponi | 15 May 2020 | Large Orders in Small Markets: On Optimal Execution with Endogenous Liquidity Supply
Speaker: Agostino Capponi, Columbia University IEOR Department
Title: Large Orders in Small Markets: On Optimal Execution with Endogenous Liquidity Supply
Date/Time: 15 May 2020, Friday, 10:00 - 11:00 AM EST
Abstract: We solve a continuous time dynamic Stackelberg game, where a large uninformed seller executes optimally, fully cognizant of the response of Cournot-competitive market makers. The game therefore endogenizes both demand and supply of liquidity. We provide closed form solutions for the value functions of market makers and large sellers, the equilibrium bid/ask price, and the execution intensity. The closed-form solution yields several insights. First, stealth trading is both privately and socially costly because market makers incur additional costs not knowing when execution ends. Second, the presence of a large seller does not unambiguously benefit other participants. Market makers benefit only if there is enough risk-absorption capacity or if the execution period is short. Other investors benefit only when the seller sells at high enough intensity. The model explains quantitatively several empirical facts: order duration and participation rate correlate negatively, and price pressure subsides before execution ends. (based on joint work with Hongzhong Zhang and Albert Menkveld).
Speaker Bio: Agostino Capponi is an Associate Professor of Industrial Engineering and Operations Research at Columbia University, a member of the Data Science Institute, the Center for the Management of Systemic risk, the FDT Center for Intelligent Asset Management, and an External Consultant at the U.S. Commodity Futures Trading Commission, Office of the Chief Economist, on topics related to clearinghouse collateral requirements and financial stability. Agostino’s research interests are in systemic risk and financial stability, economics of clearinghouses, market microstructure, and human-machine interaction systems; he is a recipient of the NSF CAREER award, a prize from the MIT Center for Finance and Policy and the Harvard Crowd Innovation Laboratory, and the Bar-Ilan prize for general research in financial mathematics. Agostino serves on the editorial boards of Mathematical Finance, Applied Mathematical Finance, Operations Research Letters, and as the Department Editor of the Institute of Industrial Engineering Transactions; he also serves as the program director of the SIAM activity group in Financial Mathematics and Engineering. Agostino received his Master and Ph.D. Degree in Computer Science and Applied and Computational Mathematics from the California Institute of Technology.