The bridge between
investment banking and
MathFinance Conference 2017
MathFinance hosts the annual Conference in Frankfurt which is tailored to the European finance community. Providing cutting-edge research and brand new practical applications, the conference is intended for practitioners in the areas of trading, quantitative or derivative research, risk and asset management, insurance as well as for academics studying or researching in the field of financial mathematics.
As always, we expect around 100 delegates both from the academia and the industry. This ensures a unique networking opportunity which should not be missed. A blend of world renowned speakers ensure that a variety of topics and issues of immediate importance are covered.
This event is a must for everyone in the quantitative financial industry.
We created a short summary video of the conference:
We want to thank our sponsors:
Dr. Hans Bühler
Global Head of Equities and Investor Services Quantitative Research
JP Morgan & Chase
We present a very efficient, large time step local volatility-type approach which is globally calibrated to possibly arbitragable to market data. The model is very simple to implement, fast, robust vs shocks and provides access to transition probabilities, which means various efficient numerical methods such as likelihood ratio greeks are accessible.
After eight years in Deutsche Bank London, he now works at JP Morgan in London, running the global Equities and Investor Services Quantitative Research team there.
Dr. Christoph Burgard
Head of Risk Analytics for Global Markets
Bank of America Merrill Lynch
The Second Quantization of Banks
• From derivatives pricing to portfolio modelling
• From bilateral to multilateral risks and network effects
• From the risk neutral world to the real world
• From efficient markets to inefficient markets
• Process automation and optimisation
• Quantitative data analysis
Christoph Burgard heads the Risk Analytics team for Global Markets at Bank of America Merrill Lynch, which he joined in November 2015. Prior to this he spent 16 years at Barclays, where he was leading the Equity Derivatives and XVA front office Quantitative Analytics teams for the investment bank as well as the ALM modelling area for the bank’s treasury department.
Christoph was named Risk Magazine’s Quant of the Year 2015 for his pioneering work on FVA. He has a PhD in Particle Physics from Hamburg University and was a research fellow at CERN and DESY.
Mauricio González Evans
BCC Group International GmbH & Co. KG
Cloud Calculation Service for the FX Volatility Smile
The Cloud Calculation Services for MathFinance AG‘s MFVal Library demonstrates how calculations can be deployed to the cloud by using BCCG’s Financial Data Enterprise Platform called “ONE”. The invocation of calculations is hereby, as an example, triggered from Excel, sent as JSON command to the Cloud Micro Services and the calculation results are received back into Excel. The implementation is a generic approach that can be applied to any Math Library that complies to a specific functions structure. Linux, Docker and Windows Platforms are supported. We show examples how market data can be used to generate the FX volatility smile: volatility on the strike space, variance on the log-moneyness space and the probability density. This allows systematic analysis of market data for the purpose of trading or model validation.
Following a range of roles in Software Development, Dealing Room Support and Project management starting in 1988, Mauricio founded BCC Group back in 2003. From 2007 he focused the company on developing a vendor agnostic Market Data Messaging Platform based on Solace Systems Appliances (FMDP). After 4 years of testing and redesign with joint forces from Solace Systems, Royal Bank of Canada and DZ BANK, BCC Group released FMDP which, as a result got certified by Bloomberg for connection to their data.
Prof. Dr. Matthias Fengler
Professor of Econometrics
University of St. Gallen
We suggest a multi-factor stochastic volatility model based on autoregressive gamma processes. Model innovations are Meixner distributed. The model is specified under the historical measure based on a stochastic discount factor, which features compensation of both return and volatility risk. We are able to fully characterize the data generating process under the risk neutral pricing measure. The model captures major stylized facts of equity data such as unconditional asymmetry and fat-tailedness, conditional skewness and conditional kurtosis, stochastic volatility and rich patterns of cross and serial correlation of the return process. The model is exponentially affine; hence standard option pricing approaches can be applied for evaluating plain vanilla options. We present empirical results for S&P 500 data.
Dr. Wolfgang Gerhardt
Speaker of the Management Board
Bank Vontobel Europe AG
Most observers focus on regulatory changes, like MiFID II and PRIIPS, when currently talking on structured products. However, besides regulation another major factor for competitiveness is technology. Pricing, issuance, execution, documentation, and settlement have to be done – at least for standard structures – automatically. This trend has become apparent with the multi issuer platforms, for example mein-zertifikat.de, which enable even private retail investors in Germany to get real-time prices for investment certificates from different issuers, to request the issue of a certain structure without any obligation for a minimum investment and to purchase the new product some minutes later via a bank at a stock Exchange.
After graduating from Wuerzburg University he had various capital markets responsibilities for CSFB-Effectenbank, Schweizerischer Bankverein (Deutschland), Citibank and Sal. Oppenheim jr. & Cie. From 2008 until 2010 he was a member of the Executive Board at Bank Sal. Oppenheim jr. & Cie. (Schweiz) AG in Zurich responsible for investment banking.
Prof. Dr. Kathrin Glau
Technical University of Munich
Complexity Reduction Techniques for Finance
Model calibration requires fast and accurate numerical methods. In the current paradigm, semi-closed pricing formulas for liquid options are seen as a prerequisite for modelling financial asset evolution. Thus attention is restricted to stochastic processes that are simple enough to allow for straightforward expressions of the pricing formulas. This obviously imposes a severe modelling restriction. However, rising demands to include more realistic features, for instance stylized facts on asymptotics of the implied volatility surface, compel us to consider a wider class of processes and hence more complex models. We therefore propose numerical techniques to reduce the computational complexity of the resulting pricing tasks. In this talk we focus on interpolation of option prices in the parameter space. Both the theoretic and experimental results show highly promising gains in efficiency. The efficiency gain is particularly high in cases where Monte Carlo simulation is required and prices need to be computed for a large number of parameter constellations.
Her research is driven by the interdisciplinary nature of computational finance and reaches across the borders of finance, stochastic analysis and numerical analysis. At the core of her current research is the design and implementation of complexity reduction techniques for finance. Key to her approach is the decomposition of algorithms in an offline phase, which is a learning step, and a fast and accurate online phase. The methods range from model order reduction of parametric partial differential equations to learning algorithms and are designed to facilitate such diverse tasks as uncertainty quantification and calibration, real-time pricing, real-time risk monitoring, and intra-day stress testing.
Dr. Frank Koster
A Universal Pairwise Local Correlation Model
In this paper we develop a local correlation model which uses a universal function g(t,m_i,m_j) to describe the local correlation between any asset-asset pair of a basket of underlyings. The arguments m_i,m_j are spot moneynesses.The universal function is calibrated to fit the implied volatilities of an equity index like DAX or EUROSTOXX50. The advantage of this approach is that we do not need to simulate the complete index when pricing options on a usually small subset of this index. The approach does also not suffer from the so-called chewing gum effect of correlation models where the local correlation depends on just the index value. The main part of our work is to show how to calibrate the universal function for each time step such that the resulting correlation matrices are positive definite almost anytime and that the function is sufficiently smooth to allow for stable evaluation.
Frank Koster works as a quant for DekaBank, a German fund manager. He holds a Ph.D. in Scientific Computing/Numerical Analysis from the University of Bonn (Germany).
His first position outside academics was with Mercedes-Benz where he developed and applied simulation methods and shape optimization software for Formula-One race engines (engines with his contributions were awarded ‘Race Engine of the Year’ in 2008 and 2009).
After that he moved into Finance working for Sal.Oppenheim and MacQuarie Capital. His main areas of work are numerical methods/models for equity derivatives.
Prof. Dr. Frank Lehrbass
Professor of Risk Management
Replacing VaR by ES – much ado about nothing?
Basel plans to replace the risk measure Value at Risk 99 by Expected Shortfall 97.5 as of 2018. We briefly compare both measures from a theoretical angle – thereby going a bit beyond the coherence properties – before we carry out an empirical investigation concerning equity and credit portfolios. The results put the regulatory initiative into question. Not only can it happen that the opposite of the intended impact is achieved, but also that the whole exercise turns out to be much ado about nothing. The presentation is based on joint work with Prof. Dr. Daniel Ziggel and M. Sc. Christina Strate, both from FOM University of Applied Sciences.
University of Delft
Symposium on Numerical Methods:
On an efficient one and multiple time-step Monte Carlo simulation of the SABR mode
In this work, we propose an efficient Monte Carlo simulation for the SABR model. The technique is based on an efficient simulation of SABR’s integrated variance process. The integrated variance process appears in the SABR model simulation since it is part of the conditional cumulative distribution of the SABR forward asset dynamics. We base our approach on the derivation of the cumulative Distribution function of the integrated variance and the use of a copulas to approximate the conditional Distribution (integrated variance conditional on the SABR volatility process). For that, a recursive procedure based on Fourier numerical techniques recovers the probability density function given the corresponding characteristic function. Resulting is a fast and accurate simulation algorithm. The one time-step version can be employed to price European options under the SABR dynamics. This converts this approach into an alternative to Hagan analytic formula for short maturities and calibration procedures. On the other hand, the multiple time-step extension of our technique is specially useful for long-term options and for exotic options.
Before his PhD research period, he worked at Department of Mathematics in University of A Coruna, also on the application of high-performance computing to SABR-like models.
Currently, his research interest moves to the use of Machine Learning within the computational finance framework.
University of Oxford
Symposium on Numercial Methods:
Calibration of a Four-Factor Hybrid Local-Stochastic Volatility Model with a New Control Variate Particle Method
We propose a novel and generic calibration technique for four-factor foreign-exchange hybrid local-stochastic volatility models with stochastic short rates. We build upon the particle method introduced by Guyon and Labordère [Nonlinear Option Pricing, Chapter 11, Chapman and Hall, 2013] and combine it with new variance reduction techniques in order to accelerate convergence. We use control variates derived from a calibrated pure local volatility model, a two-factor Heston-type LSV model (both with deterministic rates), and the stochastic (CIR) short rates. Our numerical experiments show that because of the dramatic variance reduction we are able to calibrate the four-factor model at almost no extra computational cost when the corresponding calibrated two-factor model is at our disposal. The method can be applied to a large class of hybrid LSV models and is not restricted to our particular choice of the diffusion. The calibration procedure is performed on real-world market data for the EUR-USD currency pair.
Dr. Roel Oomen
Global Co-Head of electronic FX spot trading
Prof. Dr. Natalie Packham
Professor of Mathematics and Statistics
Berlin School of Economics and Law
Current developments in model risk measurement
Model risk has been a concern ever since models have been used for pricing and risk management of trading positions. Recent updates in regulatory frameworks (FRTB, EBA “prudent valuation”) set out concrete requirements for model risk measurement. After giving an overview of regulatory developments and requirements, we develop value-at-risk and expected shortfall measures for model risk based on realised P&L from model risk. We further demonstrate how the model risk framework can be used to assess the hedge quality in incomplete markets with jumps. This helps explain why in some markets deliberately misspecified, but complete, market models are preferred over more appropriate but incomplete market models.
Prof. Rolf Poulsen
Professor of Mathematical Finance
University of Copenhagen
How Accurately Did Markets Predict the GBP/USD Exchange Rate Around the Brexit Referendum?
We develop a model for the British pound/US dollar exchange rate around the Brexit referendum in June 2016. Applying the model to a combination of betting market odds and financial option prices, we show that while Leave was the least likely outcome (more unlikely, in fact, than betting odds would immediately suggest), predictions of the exchange rate conditional on the outcome of the referendum were accurate and the market was able to separate its views on the likelihood and the impact of Brexit.
Dr. Peter Quell
Head of Portfolio Analytics for Market/Credit Risk
Adaptive Market Risk Measurement in the Trading Book
Current regulatory capital calculation in most financial institutions heavily relies on stationarity assumptions related to the underlying risk drivers. Nevertheless empirical studies show that non stationarity is one of the prevailing stylized facts in financial markets. This presentation describes a straightforward extension of already existing internal market risk models to cope with rapidly changing market conditions. The performance of this so called adaptive market risk method is demonstrated using financial market data as well as Monte Carlo simulation exercises based on different time series models as well as Fractional Brownian Motion.
Dr. Peter Quell is Head of the Portfolio Analytics Team for Market and Credit Risk in the Risk Controlling Unit of DZ BANK AG in Frankfurt. He is responsible for methodological aspects of Internal Risk Models, Economic Capital and Model Risk. Prior to joining DZ BANK AG Peter was Manager at d-fine GmbH where he dealt with various aspects of Risk Management Systems in the Banking Industry. He holds a MSc. in Mathematical Finance from Oxford University and a PhD in Mathematics. Peter is member of the editorial board of the Journal of Risk Model Validation.
Dr. Wolfgang Scherer
Head of Model Validation Credit Trading
Quantum Computing in Finance: Hype or Hyperspeed?
The basic and novel features of quantum computing are presented followed by a brief outline of its theoretical promise as shown in the most prominent algorithms. An overview of current approaches to quantum computing is given. A potential application of quantum computing in finance is reviewed and the emerging enthusiasm of its applications in finance is critically assessed.
Dr. Sebastian Schlenkrich
Manager Financial Engineering
Quasi-Gaussian Model for Model Validation and Pricing Analysis
Validation of pricing models for structured rates products relies on powerful benchmark models that allow capturing the various risk factors like volatilities, skew and correlation. In this presentation we analyze the class of Quasi-Gaussian interest rate models. This includes model parametrization and calibration methodologies.
Moreover, we demonstrate how Quasi-Gaussian models can be applied to disentangle effects arising from the different risk factors involved. This improves transparency on product features and may help to understand price quotes from different internal and external sources.
Sebastian Schlenkrich is Manager in the Financial Engineering unit at d-fine, a leading consultancy company specialised in risk and finance. In this role he manages and delivers client projects on current valuation and risk management topics. Previously, in the Macro Valuation Methodologies Team at UBS Investment Bank, London he held global responsibilities for methodology and tool development of pricing model validation.
A focus of his research and work are valuation methodologies for interest rate, FX and hybrid derivatives. Furthermore he works in the field of Algorithmic Differentiation and its application in finance.
Sebastian holds a Ph.D. in Mathematics from Technische Universität Dresden and a MSc in Mathematical Finance from University of Oxford.
Prof. Dr. Thorsten Schmidt
Unbiased estimation of risk measures
Recently, the estimation of risk measures gained a lot of attention, partly because of the backtesting issues related to elicitability. In this work we shed a new and fundamental light on the biasedness of estimation of risk measures. We show that once the parameters of a model need to be estimated, one has to take additional care when estimating risks. The typical plug-in approach, for example, introduces a significant bias which leads to an underestimation of risk. Moreover, backtesting will consequently fail with models which do fit the data due to the additional bias. We give a number of motivating examples which show the outperformance of unbiased estimators in many circumstances.
(This is joint work with Marcin Pitera.)
Thorsten Schmidt is Professor for Mathematical Stochastics at University Freiburg (successor of Ernst Eberlein).
Prior to this he was professor for Mathematical Finance at Chemnitz University of Technology since 2008, held a replacement Professorship from Technical University Munich in 2008 and was Associate Professor at University of Leipzig from 2004 on. His Ph.D. he obtained from University in Giessen in 2003 on credit risk with infinite dimensional models.
Besides his interests in Mathematical Finance, in particular interest rates, credit risk and energy markets, he has a strong background in statistics and probability theory. His research focusses on topics in mathematical finance and the theory and application of stochastic processes. This includes credit risky markets, interest rate markets, dynamic term structure models, insurance mathematics, energy markets and related fields.
Dr. Peter Schwendner
Senior Lecturer at the Institute for Wealth and Asset Management
ZHAW School of Management and Law
Sovereign Bond Network Dynamics
From 2004 to 2016, the market perception of the sovereign risks of euro area government bonds experienced several different phases, reflected in a clear time structure of the correlation matrix between the yield changes. “Core” and “peripheral” bonds cluster in a bloc-like structure, but the correlations between the blocs are time-dependent and even become negative in periods of stress.
Using noise-filtered partial correlation influences, this time-dependency can be evaluated and visualized using network graphs.
Our results support the view that market-implied spill-over risks have decreased since the European rescue and stability mechanisms came into force.
In 2015, spill-over risks reappeared during the Eurogroup’s negotiations with Greece, although the periphery yields did not show risk spreads that were as large as those in 2012.
We also discuss the 2016 market movements before and after the Brexit referendum, the US election and the Italian constitutional referendum.
Head of Quantitative Analytics Europe / Global Head of QA Macro, Structured Products and Strategies
Chair of the Panel discussion on “recent challenges in derivatives technology”
Tino studied Mathematics and Computer Science at University of Jena and University College Cork, finishing his thesis as intern at Commerz Financial Products in Frankfurt.
Dr. Manuel Wittke
Dr. Mikhail Beketov
Robo advisors are portfolio management platforms that are based on quantitative algorithms and work with minimum human intervention. Currently, the robo advisors are considered as a disruptive trend in wealth management, and according to the different forecasts the robo advisors will globally manage from $2 to $8 trillion by 2020, which is 2.5 – 10% of total global assets under management.
The quantitative methods used in the robo advisors are diverse, yet most of the systems employ classical Markowitz’s mean-variance optimization. This approach is usually complemented by various methods to mitigate such issues as highly-concentrated portfolios, input-sensitivity, and estimation error maximization (constraints set, Black-Litterman, capitalization-specific allocation, etc.).
We provide an overview of both the business and quantitative methods of the robo advisors. Besides, we outline promising approaches and challenges in the implementation and testing of the robo advisory systems.
Manuel is working for Deloitte since 2012. He is a senior manager and member of Deloitte’s quant-team. His main areas of consulting are portfolio risk simulation and model validation. Within this context he is involved in several portfolio risk simulation projects, the conceptual design and implementation of financial markets applications like robo advisors, computation of xVA numbers and benchmarking of portfolios. He is leading the EMEA valuation library and German model validation initiative within Deloitte. Before his career at Deloitte he was a financial engineer in Postbank’s Quantitative Analytics team. Manuel obtained his Diploma and his PhD in Financial Economics from Bonn University.
Mikhail is Manager in the quant-team at Deloitte´s service line Financial Risk Solutions. The main current professional interests of Mikhail are quantitative portfolio and risk management, investment/trading strategies, and financial econometrics. He is an expert in building financial applications and is leading the quantitative asset management initiative. Mikhail has more than 12 years of experience as a quantitative researcher/analyst in the asset management industry and as a research scientist. He has published 37 articles in international scientific journals and since 2016 he is also a lecturer of Econometrics at the Frankfurt School of Finance and Management.
Chief Risk Officer
State Street Bank International GmbH
The future of quant in risk management
More than 20 years ago Kris Wulteputte joined the finance industry with an engineer degree, after a short stint in academia as a teaching assistant for the course in stochastic processes. At JPMorgan in London and New York he was part of the team promoting risk measures like Value at Risk and Average Shortfall, introducing a credit portfolio model as well as a long-term forecasting model, before spinning off RiskMetrics Group. Later he rejoined the Finance industry first as a market risk manager, then head of risk, CRO and member of the board of a significant institution. In his address he will reflect on his early days in the industry, on the changing focus and role for quants, on the catharsis brought by the financial crisis, and on the way the risk function currently engages with quants and where this may be leading in the medium term.
Kris Wulteputte is CRO and member of the Executive Managing Board for State Street Bank International GmbH. Based in Munich, he is responsible for the risk management function across the different risk areas for the International Bank and its branches (Italy, Poland, Switzerland, Austria, Luxembourg, the Netherlands, UK). He chairs the Risk Committee and is a member of the senior governance bodies at SSB GmbH and of the Supervisory Board of SSB Luxembourg.
Mr. Wulteputte has over 20 years of relevant experience in the banking industry and in risk management. He joined State Street from a Dutch custodian bank, where he was CFRO and a member of the board responsible for Finance, Risk, Legal and Compliance. In the late ‘90s he was a member of the Risk Management Services Group at JPMorgan in London and New York and a founding partner of RiskMetrics Group, closely involved with the introduction of quantitative risk measures like Value at Risk and Average Shortfall, and with the introduction of methodologies for market risk management, credit portfolio management and long-term scenario generation. After returning to continental Europe he held various risk management positions in The Netherlands.
Mr. Wulteputte holds a degree in Engineering with great distinction from Ghent University (Belgium) part of which was the completion under the ERASMUS program of an M.Sc. in Digital Systems Engineering at UMIST (UK).
He started his career as a teaching assistant in stochastic processes (Ghent University) and has been a visiting lecturer in risk management at the VU university in Amsterdam (The Netherlands). He was an independent member of the risk committee at LCH.Clearnet, a leading multi-asset clearing house. He has spoken widely on risk management issues and the Basel capital accords.
University of Antwerp
ADI Finite Difference Schemes for the Calibration of Stochastic Local Volatility Models: An Adjoint Method
In contemporary financial mathematics, stochastic local volatility (SLV) models form state-of-the-art models to describe asset price processes. The local component of the SLV model, the so-called leverage function, is defined in a natural way such that the SLV model yields the same fair value for vanilla options as the underlying local volatility (LV) model. Determining this leverage function is, however, a highly non-trivial task. For example, the fair option values defined by the LV model can often not be obtained analytically and have to be approximated, e.g. by numerically solving the corresponding backward PDE.
Consider any given discretization by finite differences of the one-dimensional backward PDE from the LV model and suppose discretization of the two-dimensional backward PDE corresponding with the SLV model is performed by similar finite difference formulas. In this talk we shall propose a calibration technique with the useful property that it determines the leverage function such that both discretizations define exactly the same approximation for the fair value of vanilla options. In this calibration procedure, which involves a two-dimensional PDE problem, alternating direction implicit (ADI) time stepping schemes are used as they are highly efficient in comparison to classical implicit methods. Ample numerical experiments are provided that illustrate the effectiveness of this calibration procedure.
Prof. Dr. Uwe Wystup
se nā lī lī lī nā nā lī
lī nā nā nā nā lī lī lī
na lī nā lī le nā lī nā
lī lī lī nā nā nā nā lī
Uwe earned his PhD in mathematical finance from Carnegie Mellon University, is currently Professor of Financial Option Price Modeling and Foreign Exchange Derivatives at University of Antwerp and Honorary Professor of Quantitative Finance at Frankfurt School of Finance & Management.
Together with his team at MathFinance he provides independent (re-)structuring, valuation, model validation and expert witness services.
His first book Foreign Exchange Risk was published in 2002, quickly became the market standard and has also been translated into Mandarin. His second book FX and Structured Products appeared in 2006. Many of his papers appeared in scientific journals.