Dear students, future innovators, and budding finance enthusiasts, we are excited to welcome you to the Quantitative Finance Seminar Event, hosted by us at LSEG! This unique gathering aims to unite students from diverse academic backgrounds, providing you with the opportunity to learn about the fascinating world of quantitative finance from industry experts and leading academics. In today’s rapidly evolving financial landscape, the significance of quantitative methods cannot be overstated. We have carefully curated an impressive line-up of guest speakers, each a renowned authority in their respective fields, to share their insights and expertise on the most relevant topics in the realm of quantitative finance. Throughout this seminar, you’ll be exposed to a wide array of topics, such as algorithmic trading, risk management, machine learning applications, portfolio optimization, and more.
What’s in it for you?
- Exposure to cutting-edge topics: You will have the opportunity to learn about the latest developments, trends, and technologies in the field of quantitative finance. This exposure can help you make informed decisions about your academic and career paths, as well as stay ahead in a competitive job market.
- Networking opportunities: Attending this event allows you to connect with industry experts, academics, and other students who share similar interests. These connections can lead to potential internships, job offers, or collaborative projects, and can also provide invaluable advice and mentorship.
- Enhanced understanding: By listening to the experiences and insights of professionals and academics in the field, you can gain a deeper understanding of the practical applications of quantitative finance concepts.
The Quantitative Finance Seminar series is a continuation of our original series which debuted in 2021 . This series aims to bring LSEG analytics closer to the modelling community. Talks will cover traditional model types in LSEG portfolio—for example, pricing, commodities, and margining—as well as new model types such as Sustainable Investments and Applied NLP. Most speakers will be model developers focusing on model methodology and applications. They will share the state-of-the-art expertise in quantitative finance. We hope the seminars will result in engaging discussions and help us in tackling the challenges we face as model stakeholders in the industry and academia.
In addition to a keynote presentation, each event will feature a Q&A section enabling attendees to connect with peers and experts, exchange ideas, and form professional relationships that may lead to future collaborations and partnerships. The dates for confirmed speakers are:
|Area and title
|Post Trade – A Model for WWR Collateral Add-ons and the Technology Behind it
The link for our first presentation: https://teams.microsoft.com/l/meetup-join/19%3ameeting_OWM0NGQ2ZDAtOGQ0NC00MzUxLTliNTItNWU1ZDYxYjU3NTQz%40thread.v2/0?context=%7B%22Tid%22%3A%22287e9f0e-91ec-4cf0-b7a4-c63898072181%22%2C%22Oid%22%3A%222dcad1f9-394c-464a-af5b-0c176001f581%22%2C%22IsBroadcastMeeting%22%3Atrue%2C%22role%22%3A%22a%22%7D&btype=a&role=a
Title: A Model for WWR Collateral Add-ons and the Technology Behind it
Claudio Albanese will present his work on wrong-way-risk (WWR) collateral add-ons for clearing and review the underlying Mathematics and Software Engineering.
Abstract: Our WWR model is based on CVA analytics. WWR add-ons are calculated so that the CVA of a member including credit-market correlations equals the CVA calculated neglecting correlations. We illustrate the model with a case study based on five realistic portfolios with about 500k equity options for a CCP with 100 members pursuing a variety of trading strategies. We compare several margin policies, including one based on Expected Shortfall with a 97.5% confidence level plus WWR add-ons and one with far higher confidence level struck at 99.7% but without WWR add-ons. We find that the two policies give rise to nearly the same total collateral requirement across members. However, WWR add-ons trigger a reallocation of collateral across members, raising requirements for members with many short put positions and reducing requirements with respect to members with a well diversified portfolio. Interestingly, a reverse stress testing analysis demonstrates that WWR add-ons substantially reduce default risk for the CCP.
The industrialization of the WWR model will require a security model to run permissioned analytics client-side directly on the ledger. Permissioned analytics are a new concept, an alternative to secure computing frameworks, and has numerous other potential applications to clearing, including cross-margining across exchanges.
The technology used for the WWR case-study in the first part of the talk is based on an innovative mathematical framework for Computational Finance developed by the author. The Mathematics uses the formalism of operator algebras from Quantum Mechanics and – unlike the more traditional risk and pricing models – applies the same solution strategy to all pricing models, all asset classes and all analytics. The resulting universal solver (called Esther) has excellent performance on servers optimized for AI use cases because of two reasons: (i) codes can be vectorized for parallel execution on GPUs and (ii) memory management is streamlined and the hardware footprint is very modest as all models are treated equally. The computational building blocks of the Esther solver are the same as those for Machine Learning (ML), matrix and tensor algebras and Hidden Markov Models. The difference is that while ML is backward looking, Esther is forward looking.
The Esther solver is implemented as a compiler for a modeling language (EDSL) which factors out all the mathematics and low-level software complexities. Just about any risk and pricing problem can be expressed in EDSL and handed over to the solver. Work is in progress to integrate Esther into Digital Assets DAML for the execution of permissioned analytics endogenously in the ledger. Also integration with emerging Large Language Models such as ChatGPT seems well within reach and would allow users to program any risk and pricing application in plain English. Application domains include calibration and reuse of market data and large scale portfolio analytics such as XVA, WWR, reverse stress testing), model risk. The Esther compiler is a multi-platform .NET application.
Bio: Claudio Albanese graduated in Physics at ETH-Zurich. His academic career include faculty positions in Physics at Princeton and the University of Toronto and a Chair in Mathematical Finance at Imperial College London. Having left academia in 2007, he founded a firm, Global Valuation, that develops risk and pricing software based on a novel mathematical framework. The first release of the software is currently powering XVA and margin calculators offered by OSTTRA, a CME-IHS venture, and received several industry awards for the innovative design and performance. The second release of the software has the form of a compiler capable to solve all problems in risk and pricing and is based on a simple modeling language consistent with ISDA CDM.
Linkedin: LSEG in Romania