Machine Learning for Market Data Anomaly Detection & Gap Filling
Numerix Webinar Featuring Intesa Sanpaolo & Deepfin Labs
Historical market data is a critical input for many market risk calculations, and an institution’s daily calculations require enormous volumes of data. However, the quality of the calculations is directly linked to the quality of the underlying data, so if data is anomalous or missing, the risk measures will be incorrect and misleading, and risk management decisions will be impacted.
Can Machine Learning (ML) techniques be utilized to identify anomalous data, and to fill in gaps in the data where is it missing, to improve the quality of these risk calculations? How proficient are these techniques at performing this task, and how much human intervention is still required?
Join presenters from Intesa Sanpaolo, Deepfin Labs and Numerix , as they shared their approaches to this task, along with results from preliminary testing.
During the webinar, the presenters covered:
- Why anomaly detection and gap filling are crucial for risk management
- Two ML techniques for anomaly detection with application to interest rate curve data
- One ML technique for gap filling with FX forward data across multiple currency pairs
- Results from preliminary testing
- An interview discussing the similarities, differences, and novelty of the different techniques
Featured Speakers
Manola Santilli, PhD
Manola joined the Market and Financial Risk Management area of Intesa Sanpaolo in 2015 as a market risk analyst in the Internal Model Market Risk Office. She works in market data management, with a focus on market data collection and statistical analysis of historical time series across all asset classes, market risk metrics monitoring, analysis and implementation of FRTB.
Previously she worked as a quantitative analyst in Société Générale.
She holds a PhD in Statistical Economics, with a thesis on advanced stochastic volatility models for derivative pricing, and an MSc in Economics and Finance.
Marco Scaringi
Marco Scaringi joined the Financial and Market Risk Management area of Intesa Sanpaolo in 2017 as quantitative analyst. His work covers pricing and risk management of financial instruments across all asset classes, with a focus on model validation, model risk, fair value adjustments, interest rate modelling, funding and counterparty risk and prudent valuation. His research focuses on interest rate models and XVAs, financial bubble analysis, portfolio optimization and Financial Benchmarks transition.
He holds an MSc in Theoretical Physics from University of Milan, with a thesis on advanced statistical mechanics techniques applied to the description and detection of financial bubbles through optimization heuristics. He also holds a post lauream degree Executive Course of Quantitative Finance from MIP, Graduate School of Business, Polytechnic of Milan, with a thesis concerning interest rate and XVA modelling.
Marco Bianchetti, PhD
Marco Bianchetti joined the Market Risk Management area of Intesa Sanpaolo in 2008. His recent work covers derivative pricing and risk management across all asset classes, with a focus on new product development, model validation, model risk management, funding and counterparty risk, and Quasi Monte Carlo. Previously Dr. Bianchetti worked for 8 years in the front office Financial Engineering area of Banca Caboto (now Banca IMI), developing pricing models and applications for interest rate and inflation trading desks. He is a frequent speaker at international conferences and training sessions in quantitative finance. He holds an MSc in theoretical nuclear physics and a PhD in theoretical condensed matter physics.
Pascal Lemieux
Pascal Lemieux is Director of Financial Engineering in EMEA at Numerix, where he works on the development of fintech software solutions. His primary focus is leveraging Numerix’s proprietary SDKs to engineer software for performing risk and valuation calculations for trading firms and financial institutions across Europe. Mr. Lemieux is also a principal contributor to the creation of the Machine Learning framework at Numerix. He has more than 15 years of professional experience in quantitative finance, working for leading firms such as J. P. Morgan and Royal Bank of Canada (RBC) in London, UK, and Société Générale in Paris, France. Mr. Lemieux holds a BEng degree in Engineering Physics and an MSc in Financial Mathematics. He is also the author of scientific publications on semiconductor modelling.
As Chief Marketing Officer and Executive Vice President of Global Marketing & Corporate Communications, James leads the company’s global marketing and corporate communications efforts, spanning a diverse set of solutions and audiences. He oversees integrated marketing communications to clients in the largest global financial markets and to the Numerix partner network through the company's branding, electronic marketing, research, events, public relations, advertising and relationship marketing.
Since joining Numerix in 2008, James has launched the organization’s award-winning thought leadership program, bringing to light challenges and insights from Numerix market experts. He also hosts the Numerix Video Blog, tackling the challenges pressing the derivatives markets—from regulatory issues to trading strategies.
Prior to joining Numerix, James served as Managing Director of Global Marketing and Communications for Fitch Ratings. During his tenure at Fitch, he built the firm’s public relations program, oversaw investor relations and led marketing and communications plans for several acquisitions. Prior to Fitch, James was a member of the communications team at Moody's Investors Service.
