Koło Naukowe Finansów Obliczeniowych zaprasza na konferencję Quant Days, która odbędzie się w dniach 25-27 listopada 2021 w formie zdalnej na platformie Zoom.
harmonogram wykładów z 25 oraz 26 listopada wraz z abstraktami i linkami Zoom, a także bio prelegentów:
czwartek 25.11
17:45 – 18:55
„Excursions in Mathematics of Quantitative Finance: on robustness and data-driven approaches” (lecture in English)
prof. Jan Obłój, University of Oxford
https://us02web.zoom.us/j/86855324047
abstract:
In the talk I will showcase how fascinating mathematics arise from trying to understand and quantify model uncertainty in modern quantitative finance. I will present various recent developments on the crossroads of statistics, optimal transport, stochastic analysis and optimization. I will also mention how machine learning tools are changing both the modelling approaches, and computational solutions, to some classical problems in mathematical finance. While part of the talk will have a broad outlook, I will also specifically focus on explaining how thinking of data series via its empirical measure – an element of an infinite dimensional space – rather than a collection of points, can help develop robust methods. I will work with Wasserstein distributionally robust optimization, develop its sensitivity analysis and show applications in finance, statistics and machine learning.
piątek 26.11
16:15 – 17:25
„How banks manage default risk for derivative portfolios” (lecture in English)
Karol Partyka, Adam Foster, Jan Malinowski (Goldman Sachs)
https://us02web.zoom.us/j/82608135868
abstract:
During the presentation Goldman Sachs experts in modelling will explain default risk management of financial derivative portfolios. The presentation will provide insight into derivative pricing, key risk measures, exposure simulation and quantification of capital. The content will be supplemented with practical examples of programmatic application of each of the topics.
17:30 – 19:10
„Jak zaawansowane narzędzia analityczne mogą wspierać proces wyceny nieruchomości?” + sesja Q&A
Tomasz Falkowski (EY)
https://us02web.zoom.us/j/83624537351
abstrakt:
Podczas wykładu zostanie omówione podejście do budowy algorytmu wyceny nieruchomości z wykorzystaniem technik nauczania maszynowego (XGBoost, elastic net, ensemble learning) z uwzględnieniem danych o geolokalizacji (w tym feature engineering). Porozmawiamy też o metodach do oceny modeli ML, oraz szerzej o interpretowalności modeli historycznie kategoryzowanych jako black-box.
Więcej informacji na wydarzeniu: https://fb.me/e/3V7VyALxB