Tag: Conference

Video Description

Can we open the black box of large language models and make AI in finance truly transparent?

As financial institutions adopt LLMs for tasks like trading, compliance, and advisory, the need for transparency is more critical than ever. Understanding how these models internally reason about financial topics is key to ensuring trust and regulatory alignment.

This session covers how mechanistic interpretability which is also termed as “MRI of AI” can reveal internal patterns in LLMs related and use it for applications such as trading, sentiment analysis and hallucination reduction.

Speaker Bio

Hariom Tatsat

Hariom has years of experience bridging AI, machine learning, quantitative techniques, and finance. He is an O’Reilly author and published researcher, with multiple research contributions in AI, machine learning, and mechanistic interpretability, particularly focused on making large language models more transparent and reliable in financial and agentic AI settings. He has been a featured speaker at several conferences and industry forums and received the Indian Achiever Award in Machine Learning. He completed his MS at UC Berkeley and his BE at IIT (India).

Video Description

• AI model risk is adversarial and should be managed like fraud, rogue-trader, or cyber risk, not like market or credit risk
• Frontier models can actively deceive — and they reason about it
• In-depth evaluations of frontier models reveal deliberate scheming: models fake alignment to get deployed, strategically underperform on evaluations, and attempt to undermine oversight
• A model that can actively deceive breaks the core assumption of current regulatory frameworks for AI: that what you tested is what you deployed
• The right framework for AI model risk is the cyber risk playbook — red-teaming, adversarial evaluation, assume-breach, and continuous monitoring, not periodic spot checks or statistical revalidation
• Governing AI with non-adversarial control frameworks is a major regulatory and governance gap that demands urgent industry action

Speaker Bio

Dr . Alexander Sokol

Alexander Sokol is the founder, Executive Chairman, and Head of Quant Research at CompatibL, a trading and risk technology company. He is also a co-founder of Numerix, where he served as CTO from 1996 to 2003.

Alexander won the 2018 Quant of the Year Award together with Leif Andersen and Michael Pykhtin for their joint work revealing the true scale of the settlement gap risk that remains in the presence of initial margin. Alexander’s other notable research contributions include systemic wrong-way risk (with Michael Pykhtin), joint measure models and the local price of risk (with John Hull and Alan White), the use of autoencoder manifolds for interest rate modelling (with Andrei Lyashenko and Fabio Mercurio), and the mean reversion skew.

Alexander graduated from high school at the age of 14 and earned a PhD from the L.D. Landau Institute for Theoretical Physics at the age of 22. He was the winner of the USSR Academy of Sciences Medal for Best Student Research of the Year in 1988.

Video Description

Inter-company relationships change over time, but they are difficult to incorporate into traditional return-based covariance estimates. This paper proposes a practical three-step framework for transforming financial news into a covariance input for portfolio optimization. First, a large language model extracts dynamic company relationship networks from daily financial news. Second, a graph neural network refines these relationships by combining the news-derived network with market features, producing company representations that reflect both narrative information and observed market conditions. Third, the refined company similarities are converted into a covariance proxy and used in a global minimum variance portfolio. Compared with traditional return-based covariance estimators and text-embedding baselines, the proposed framework improves out-of-sample risk-adjusted performance, volatility control, and downside-risk management. The results indicate that text-embedding baselines capture meaningful financial similarity but that document-level averaging can dilute discriminative firm-level signals, limiting their effectiveness as standalone covariance estimators. Our findings suggest that news-derived information networks can complement historical returns by providing a dynamic covariance input that reflects evolving economic relationships among firms.

Speaker Bio

Yuyu Fan

Yuyu Fan is a Senior Vice President and Director of AI Research on the Data Science team at AB, where she leads the development and implementation of advanced artificial intelligence (AI) and machine learning (ML) models to meet critical business needs. Fan collaborates with various business units within the firm to transform business requirements into data-driven solutions. For instance, she leverages AI to generate investment signals for portfolio managers and partners with the Client team to create models that optimize sales processes. Prior to joining the firm in 2018, Fan worked at the College Board as a psychometrician intern for two years, using ML models to monitor test validity, reliability and security. She holds a BA in sociology from Zhejiang University (Hangzhou, China), MAs in sociology and psychology from Fordham University, and a PhD in psychometrics and quantitative psychology from Fordham University. Location: New York.

Video Description

In recent years, deep learning (DL) models have experienced notable success in predicting high-frequency returns in equities by leveraging extensive order book data and directly extracting features from it. This marks a notable departure from current industry practice, where features are often manually crafted. In this talk, I discuss two of our articles on this topic, addressing several practical open questions in this area, such as determining the most suitable network architecture and input selection for forecasting returns at multiple horizons (e.g. alpha term structures), optimizing the width and depth of neural network components to enhance performance, defining the appropriate historical data window size, and evaluating the benefits of incorporating time as a feature in the models. We evaluate the effectiveness of four DL models in forecasting high-frequency alpha term structures across various settings: a simple LSTM, a multi-head LSTM, an LSTM Seq2Seq without attention, and an LSTM Seq2Seq with attention. We find that surpassing the performance of a simple LSTM in the return forecasting task is surprisingly challenging.

Speaker Bio

Professor Petter Kolm

Petter Kolm is a Professor at Miami Herbert Business School, University of Miami. He was previously a Professor at NYU Courant where he directed the Mathematics in Finance Master’s program. In 2021, he was honored as “Quant of the Year” by Portfolio Management Research and the Journal of Portfolio Management for his significant contributions to quantitative portfolio theory. In 2026, he was named “Buy-Side Quant of the Year” by Risk.net.

He has co-authored numerous influential articles and books on quantitative finance, portfolio management, and financial data science. Petter serves on several company advisory boards; editorial boards for leading academic journals; and boards of directors for professional associations. Previously, he worked in Quantitative Strategies at Goldman Sachs Asset Management.

As a consultant and expert witness, Petter provides expertise in machine learning, portfolio management, risk management, and systematic trading. He earned his Ph.D. in Mathematics from Yale University, an M.Phil. in Applied Mathematics from the Royal Institute of Technology (KTH), and an M.S. in Mathematics from ETH Zurich.

Video Description

Large language models are now ubiquitous, and with them the ability to read, summarise and structure text that was out of reach a decade ago. One deliverable in particular has become almost trivial: the extraction of semantic triples – node, edge, node – from unstructured corpora at scale. This has put an old question back on the table. If building a firm network no longer requires hand-curated data, are the signals that a semantic network produces worth anything in the cross-section, and if so, what kind of thing are they?

We build a knowledge graph of global developed-market equities from twenty-five million LLM-extracted triples and use it to settle a single question: is a semantic centrality signal compensation for a risk exposure, or a genuine pricing error?

We first set out what centrality means on a graph of this kind, introducing the familiar measures – degree, Katz, PageRank, eigenvector – and explaining which we take forward and why. We then sort firms on point-in-time centrality within a large global universe, form a central-minus-peripheral long-short, and take it through the tests the network-signal literature usually stops short of. We ask whether the spread survives a standard external six-factor model. We characterise what the long and short legs actually hold – their size, their volatility, how heavily the corpus writes about them – so that any premium can be named rather than merely labelled. Finally, we ask whether the wiring of the network matters at all, by rewiring the graph while preserving each firm’s number of connections.

Speaker Bio

Tony Guida

Tony Guida is a Quantitative Portfolio Manager, researcher, and author with expertise in systematic investing, machine learning, and financial data science. He currently works on the integration of large language models (LLMs) and knowledge graphs into investment processes for a Swiss asset manager.

Tony began his career at Unigestion in 2006, joining the quantitative equity team as a research analyst. He later became a member of the Research and Investment Committee for Minimum Variance Strategies, where he led the factor investing research group for institutional clients. In 2015, he joined EDHEC-Risk Scientific Beta as a Senior Consultant, focusing on risk allocation and multi-factor strategies. In 2016, Tony moved to a major UK pension fund, where he built an in-house systematic equity strategy and co-managed a £8 billion portfolio as Senior Quantitative Portfolio Manager.

In 2019, he joined RAM Active Investments as a Senior Quantitative Researcher in equities, later co-heading the systematic macro hedge fund offering. In 2023, Tony co-founded a hedge fund, where he served as Co-Head of Research and Portfolio Manager, leading the development of cross-asset, machine learning-driven investment strategies.

Tony holds Bachelor’s and Master’s degrees in Econometrics, and a Master’s in Economics and Finance. He has authored and edited several industry-leading books, including:

• Big Data and Machine Learning in Quantitative Investment (Wiley, 2018)
• Machine Learning for Factor Investing: R Version (CRC, 2020)
• Machine Learning for Factor Investing: Python Version (CRC, 2023)

He is also an Advisory Board Member for the Financial Data Professional Institute and serves as a peer reviewer for academic journals in machine learning and finance.

Video Description

We consider the problem of backcasting market data. This can be relevant if long histories of time series are not available, e.g. lack of data due to newly introduced indices like the risk free rates, quotes of newly listed companies on exchanges, or simply non-available in-house data. Financial institutions need to extend their time series using backcasting. This is often a somewhat ad-hoc method. We propose a sound statistical method based on Mixture Models. Since the models work well with a reasonable amount of available data, they can ideally be used for this application. Furthermore, using the method provides an interpretable methodology, gives variance estimates and does not rely on expert judgement, e.g. input parameters. All comes from available data. Furthermore, the method allows to use conditional backcasting and proxying with regard to available, liquid market data. The latter is very relevant since it takes the joint distribution of several market variables into account.

Speaker Bio

Dr. Jörg Kienitz

Jörg is a Quantitative Finance professional for more than 25 years. After holding positions as Head of Quant with Postbannk/Deutsche Bank, Deloitte or LSEG, he is now Director of Quantiative Methods at mrig, a Frankfurt based consultancy. He also works as an adjunct associate professor at UCT, Cape Town, and as an assistant professor at BUW, Wuppertal. Jörg is frequently speaking on conferences, (co-) authored four books with Wiley and Springer and published many research and practiitioners papers in Quantitätive Finance, Journal of Computational Finance, RISK or Wilmott.

Dr. Lorenc Kapllani

Lorenc Kapllani is a Senior Quantitative Finance & Machine Learning Consultant at mrig, a Frankfurt-based consultancy specializing in quantitative finance. Prior to joining mrig, he worked as a Credit Risk Analyst at GEFA Bank in Germany.

He holds a PhD in Applied Mathematics from the University of Wuppertal, where his research focused on deep-learning methods for pricing high-dimensional nonlinear derivatives, risk management, and uncertainty quantification — published across peer-reviewed papers in journals including the IMA Journal of Numerical Analysis.

Alongside his industry role, he continues to research at the intersection of quantitative finance and machine learning.

Video Description

The talk first introduces the basics of quantum mechanics, including qubits and their unique properties of superposition and entanglement. It then examines the principal challenges facing quantum computing, including the sensitivity of qubits to thermal fluctuations, vibration, electromagnetic interference, and other sources of noise. The talk outlines how artificial intelligence is helping overcome these limitations through noise reduction, error mitigation, qubit calibration, AI-assisted quantum circuit design, and hybrid classical–quantum computing.

Speaker Bio

Dr. Gunter Meissner

After a lectureship in mathematics and statistics at the Economic Academy Kiel, Gunter Meissner PhD, joined Deutsche Bank in 1990, trading interest rate futures, swaps, and options in Frankfurt and New York. He became Head of Product Development in 1994, responsible for originating algorithms for new derivatives products, which at the time were Lookback Options, Multi-asset Options, Quanto Options, Average Options, Index Amortizing Swaps, and Bermuda Swaptions. In 1995/1996 Gunter Meissner was Head of Options at Deutsche Bank Tokyo. From 1997 to 2007, Gunter was Professor of Finance at Hawaii Pacific University and from 2008 to 2013 Director of the Master in Financial Engineering Program at the University of Hawaii. Currently, he is President of Derivatives Software (www.dersoft.com), Adjunct Professor of Mathematical Finance at Columbia University and NYU, and Executive Research Director at Bhodi research group, www.Bodhiresearchgroup.com.

Gunter Meissner has published numerous papers and seven books on financial derivatives, risk management, and statistics. His new book “AI Applications in Finance – Opportunities and Limitations” will be published in fall 2026. Gunter can be reached at meissner@hawaii.edu. His CV is at www.dersoft.com/cv.pdf.

Watch the recordings from the AI and Machine Learning in Quant Finance Conference 2026.

Video Description

In this talk, Dr. Adam Rej discusses two mechanisms that can load to a decay of a systematic strategy in production: in-sample overfitting and arbitrage. Thanks to the reconstruction of 72 equity strategies / factors published in academic literature, he can test various proxy variables of overfitting and arbitrage in the cross-section and determine their statistical significance. Dr. Rej concludes with practical implications for systematic investors.

Speaker Bio

Dr. Adam Rej

Adam Rej is Head of Macro Alpha, based in New York. He has previously worked for the Portfolio Construction and Alternative Beta teams. Prior to joining CFM, Adam held post-doctoral positions at École Normal Supérieure (Paris), Institute for Advanced Study (Princeton) and at Imperial College London. His PhD research was at the Max Planck Institute for Gravitational Physics – Potsdam in the field of theoretical physics. Adam joined CFM in 2014.

Video Description

For long term Strategic Asset Allocation, model portfolios are defined at the level of indexes. The possible outcomes at a scale of a few decades is obtained by Monte Carlo simulations, resulting in a probability density for the portfolio values. Such studies are critical for long term wealth plannings, for example in the financial component of social insurances. The base model is a constant drift, a constant covariance and normal innovations, as pioneered by Bachelier. Beyond this model, this presentation summarizes  a multivariate process that incorporate the most recent advances in the models for financial time series. This includes a dynamic drift estimate, the heteroskedasticity (i.e. the volatility’ dynamics), and the fat tails and asymmetry for the distributions of returns. The quantitative outcomes depend critically on the drift, because this is a non random contribution acting at each time step. The changes introduced by the drift dynamics is the partial decoupling between the volatility along the time direction from the standard deviation of the terminal values. Finally, the main statistics for the wealth at increasing time are presented, showing the key features added by the components beyond the basic normal random walk.

Speaker Bio

Gilles Zumbach

Gilles Zumbach has been doing research on many topics in finance, ranging from tick-by-tick time series to very long term market simulations, from risk evaluations to realistic option pricing, from large scale portfolio optimisation to pricing. He has worked for several institutions, including banks, hedge funds and service providers. Gilles has published over 35 research papers in finance, most of them using careful data analyses combined with mathematical models. He wrote a book published by Springer Verlag, linking fundamental research on time series to applications in finance. Recurring themes in his work are processes and volatility. In a former life, he was as a physicist.