to top
Select language:    en ENGLISH  |  cs CZECH
Moderné nástroje
pre finančnú analýzu
a modelovanie 2019
17. ročník konferencie
Datum:30.5.2019 (štvrtok)
Miesto:Národná banka Slovenska, Bratislava, Slovenská republika
Jazyk:Slovenský, český, anglický
Solving Large Optimization Problems in Finance: How MATLAB Can Help You
Jorge Paloschi, MathWorks, United Kingdom
Solving optimization problems is an important task in Finance. Examples of this are fitting models to data and optimizing portfolios composition based on different criteria. MATLAB provides a very good set of tools for these tasks, with different solver types and heavy customization of how to use them.
We will focus mainly on performance. That is, improving the execution time when optimizing large problems or smaller ones that take many iterations to converge. We will show how to use the different tools available in MATLAB (as, for instance, using a parallel environment). We will also address the provision of gradients and Hessians, and how this will affect both performance and robustness.
BIO: Dr. Jorge Paloschi has been a Senior Consultant at MathWorks for more than 11 years, focusing mainly on Finance related projects. Prior to MathWorks, he worked for several years in the front office of one of the biggest investment banks in London (Credit Derivatives desk). Jorge holds a first degree in Maths and a PhD from Imperial College in London and has published many scientific papers on solving sets of algebraic nonlinear equations.
Public Investment and EU Funds in a Small Open Economy Integrated in the Euro Area
Martin Železník, Národná banka Slovenska
Small open economies within the European Union can be extensively influenced by the utilization of the structural and investment EU funds. Even more so if they are eligible to draw from the Cohesion Fund targeted at the less developed EU countries. In many of these countries, we observe an EU-funds cycle that causes spikes in total investment as the programming period draws near to its end, and a decline after the new programming period begins. As the share of EU-funded public investment and the public investment financed from domestic sources varies highly over time, we decided to explore the differences in the transmission of these two types of public investment shocks into the real economy. We use a version of the EAGLE model calibrated for the Slovak economy integrated in the euro area and extended with EU funds mechanisms. We find that if the part of the total investment that is funded from domestic sources comes from an increase in taxes, the EU-funded investment delivers larger improvement in real GDP. The difference is especially striking for investment funded by an increase of social security contributions paid by firms.
BIO: Martin Železník absolvoval odbor Hospodárska politika na Ekonomicko-správní fakulte Masarykovej univerzity v Brne. V súčasnosti pracuje ako expert analytik ekonometrického modelovania v Národnej Banke Slovenska. Pôsobí tiež ako zástupca NBS v pracovnej skupine pre ekonometrické modelovanie v rámci Európskej Centrálnej Banky.
A New Foreign Block for the CNB Core Model
Stanislav Tvrz, Česká národní banka
In this work we present the structure and empirical results of the new foreign block of the Czech National Bank's core prediction model. The new foreign block is more structural than the previous version, which makes it possible to interpret the foreign outlook in a storytelling way. At the same time the model also incorporates the oil price as a production factor.
BIO: Stanislav Tvrz works at the Monetary Department of the Czech National Bank. Previously he worked at the Monetary Policy Department of the National Bank of Slovakia. He specializes in macroeconomic modelling and forecasting, especially with the use of DSGE models.
Measuring Market Risk with Entropy
Martin Veselý, Česká národní banka
Since Great Financial Crisis many analysts have been searching for a risk measure with as little as possible assumptions about underlying data. A metric called entropy seems to fulfil this requirement. This metric is used in information theory to describe a "surprise" contained in a message. Since a risk can be considered as the "surprise" as well, the entropy seems to be an appropriate measure. Firstly, a mathematical model of entropy is discussed, then a method how to calculate it is presented and finally it is shown that the entropy can be actually used as a market risk measure in a practical way.
BIO: Martin Vesely works at Czech National Bank since 2012 as a risk manager. He focusses mainly on credit risk. Moreover, Mr. Vesely is responsible for a reporting tools development and he is an IT business partner at his department. He graduated from Faculty of Nuclear Sciences and Physical Engineering summa cum laude in 2013. He also holds BBA in Financial Management.
Gone with the Wind: Safe Debt Levels from the Perspective of Pension Reform
Zuzana Múčka, Rada pre rozpočtovú zodpovednosť
We assess the adverse fiscal and macroeconomic implications of population ageing in Slovakia, how they can be reduced through pension reforms, and what would be the impact in case reforms were reversed. The latter aspect is highly relevant due to recent withdrawal of earlier reform measures. Compared to other studies we exploit the country-specific information contained in the 2018 Ageing Report projections and integrate this information into a general equilibrium model with overlapping generations. We find that population ageing - if left unaddressed - has major adverse macroeconomic effects, with the debt-to-GDP ratio in the baseline scenario increasing by more than 100pp. By contrast, implementing comprehensive reform packages helps to alleviate the adverse impact and to spread the adjustment burden more equally across generations. Furthermore, reversal of earlier pension reforms leads to substantial fiscal and macroeconomic costs - going well above the costs of the baseline scenario.
BIO: Zuzana Múčka works as a senior analyst and modeler for the Council for Budget Responsibility (Slovakia), prior to that she worked in the IT sector (software engineer and analyst for Microsoft and local company Anasoft). She finished her master and postgraduate studies at the Faculty of mathematics, physics and informatics, Comenius University and the University of Sheffield, United Kingdom. During her studies she participated in several internships at the University of Jyvaskyla, Finland.
Optimalizácia výroby plynovej elektrárne Malženice
Kristián Barát, ZSE Energia a.s.
Témou príspevku je riadenie optimalizácie výroby plynovej elektrárne Malženice. Príspevok prináša praktický pohľad na riadenie výroby plynovej elektrárne reflektujúc na technické vlastnosti elektrárne ako aj trhové podmienky - najmä vývoj cien elektrickej energie, plynu a emisných povoleniek. Každodenná prevádza prináša zaujímavé situácie ako sa vysporiadať s príležitosťami na trhu s energiami a maximalizovať tento potenciál v prospech elektrárne.
BIO: Kristián Barát is the Head of analytical team of energy sourcing division at ZSE Energia a.s. Previously he worked as Head of risk management team in Západoslovenská energetika a.s. After 10 years of experience in risk management his new focus is on portfolio management, market analysis, renewable energy and prediction of energy production and consumption.
Risk Modelling Using Monte Carlo Methods – Design and Validation of an Insurance Model
Philip Gruber, arithmetica Consulting GmbH, Austria
Using a Monte Carlo model for risk management purposes requires careful thought about the design and validation of the model. This is especially important when approval of supervisors is involved, as is the case for internal models in Solvency II. Automated validation tests improve the validation process and enable the validators to focus on the topics which matter most. In this presentation, the initial design decisions involved in creating a simulation model are presented together with common validation tests and ways to automate them.
BIO: Philip Gruber is a senior consultant in the insurance & banking department of arithmetica, an Austrian consultant company specialised in actuarial topics. He is project head of the development of a partial internal model for Non-Life Solvency II risk, which is in active use since the inception of Solvency II 2016. His responsibilities include the design of the underlying mathematical model and the development of the actual software.
He holds a master’s degree in actuarial and financial mathematics from TU Vienna and is a regular member of the Actuarial Association of Austria.
Optimization of Financial Models Using Evolutionary Algorithms and GPU Computing
Michal Hojčka, Riccardo Gismondi, R7 CORP k.s.
Traditional gradient-based optimization techniques are not very useful in case of multimodal or non-differentiable functions which is also the case with many financial models. We use Evolutionary Algorithms (Genetic Algorithm (GA), Covariance Matrix Adaptation Evolution Strategy (CMA-ES), etc.) to provide an alternative and more robust way of optimization. In order to deal with huge computational requirements we utilize GPU parallel computing with CUDA.
BIO: Michal Hojčka has PhD in applied mathematics and for the last four years has been developing software in R7 Corp.
BIO: Riccardo Gismondi, CEO of R7 Corp, has Master in Mathematics. He has huge experience in the fields of Quantitative Finance and Algorithmic Trading with many international clients (Banks, Investment Funds).
Master Class
Application Development Workflow - From Algorithm to Production Systems
Jorge Paloschi, Michal Blaho, MathWorks, Humusoft s.r.o.
Paloschi Blaho
Sharing analysis results with your colleagues can be challenging, especially when they are not MATLAB users. MathWorks provide several tools and workflows to solve this challenge. During this masterclass we will focus on application building, packaging and deployment to showcase application development workflows successfully used in many industries.
Deploying MATLAB implemented application for production implies undertaking several steps to ensure a safe process. We will be discussing these topics as well as what tools are provided by MathWorks to make this not only a robust process but also make software developers and managers life easier.
BIO: Dr. Jorge Paloschi has been a Senior Consultant at MathWorks for more than 11 years, focusing mainly on Finance related projects. Prior to MathWorks, he worked for several years in the front office of one of the biggest investment banks in London (Credit Derivatives desk). Jorge holds a first degree in Maths and a PhD from Imperial College in London and has published many scientific papers on solving sets of algebraic nonlinear equations.
BIO: Michal Blaho pracuje v spoločnosti Humusoft ako aplikačný inžinier pre Slovenskú republiku. Je absolventom inžinierskeho a doktorandského štúdia v odbore Automatizácia na Fakulte elektrotechniky a informatiky STU v Bratislave. Medzi jeho hlavné odborné záujmy patrí modelovanie systémov, riadenie systémov a generovanie kódu.

© HUMUSOFT 1991 - 2019