Main Conference Day OneTuesday, 18 November 2025 - GMT (Greenwich Mean Time, GMTZ)

From 14.00 - 15.30

Beyond the well-charted territory of equity portfolio construction lies the complex challenge of fixed income – a domain where traditional quantitative approaches have struggled. While equity markets benefit from simple asset properties and good data access, fixed income presents formidable obstacles. The few quantitative methods in fixed income – notably stratified sampling for passive strategies – merely scratch the surface. These techniques are incomplete workarounds rather than solutions to the fundamental challenge: the absence of high-quality fixed income risk models.

Our research tackles this through sophisticated interest rate and granular issuer spread curves, coupled with systematic factor models similar to equity approaches. We'll explore theoretical aspects of model construction, including "risk entities" that capture fixed income's multidimensional nature, along with the unique challenges of implementing these models in portfolio optimization frameworks.

Join us to discover research breakthroughs enabling truly quantitative fixed income portfolio construction and optimization.

Traditional diversification metrics work well in calm markets but become misleading during market stress. During extreme events, correlations that appear stable at 0.3 can suddenly spike to 0.9, leaving portfolios more concentrated than expected. This session explores why diversification breaks down during stress periods and provides practical techniques for identifying hidden factor concentrations, building more robust diversification measures, and stress-testing portfolio resilience.

In this talk, I discuss the existing ML models for Limit Order Book (LOB) data and explore the prospects of emerging generative LOB models in high-frequency trading, optimal execution, and surveillance. Over the past eight years, several machine learning models have been developed for forecasting from LOB data. One of the most studied tasks is mid-price prediction with Level II LOB data. Examples of the key contributions are the DeepLOB model from Oxford University and the TABL model and its variants from Tampere University. These approaches are based on supervised learning, framing the task as a classification problem: whether the average-filtered mid-price moves up, stays the same, or goes down. Despite the apparent simplicity of this formulation, the methods have proven surprisingly effective in modeling very short-term price dynamics. Moreover, the models are computationally efficient and well suited for low-latency applications in high-frequency trading and market making. More recently, GAN-based models (e.g., one from Oxford) have emerged, showing promise in predicting full LOB state snapshots.

With the rise of large language models, transformer-based generative methods for sequential data have also gained attention. Level III LOB message flow data forms a natural application area for such models. Development in this direction has only started in the past couple of years and is still very much in progress, yet the latest findings are highly promising. It is plausible that within the next one to two years, we will see generative foundation models for LOB data capable of reliably forecasting message flows, LOB snapshots, and derived features such as the mid-price. These models may also enable counterfactual “what-if” analyses, opening new opportunities for data-driven market impact modeling and market surveillance. The main challenges ahead include ensuring sufficient inference speed for low-latency trading and achieving robustness in real-world market conditions.

Using tick-level data from the leading crypto asset markets (Binance and Coinbase), we quantify how market orders, limit-order submissions, and cancellations each drive bitcoin’s price dynamics. While market orders move prices more per event, limit orders dominate price discovery overall. Binance consistently leads Coinbase in incorporating new information, despite having a coarser tick size. Cross-exchange price impacts are asymmetric and more pronounced during volatile periods. Our findings suggest that exchange-specific microstructure affects price formation in crypto markets, which remain fragmented and frictional despite fungibility. These insights inform execution strategy and regulatory design in decentralized trading environments.

The talk will explore the interpretability of the extreme copulas with examples from finance and from the air combat scenarios

• Multi-risk exposure and hedging cost tackling
• Multi-strategy investment and budget allocation
• Unified modelling and solving procedure

Constructing an efficient hedge portfolio for derivatives within risk limits, which are typically based on Greeks, is crucial from both trading and risk management perspectives. Therefore, we propose simple algorithms to optimize the hedge portfolio with complex Greeks, inspired by the proximal point method with the Sinkhorn algorithm. One algorithm focuses on optimizing cost, while the other optimizes both cost and CVaR (conditional value at risk) with constraints to limit the residual Greeks. Our algorithms consist of iterations involving only simple vector and matrix calculations, similar to the Sinkhorn algorithm, which is now widely used in machine learning; thus, linear programming solvers are not needed. In numerical examples, we explore two ways to improve the hedge portfolio for Bermudan swaptions, moving away from conventional anti-diagonal vega hedging with coterminal swaptions. The results of the optimized cost, CVaR, and hedge positions produced by our algorithms are consistent with insights on hedging under various constraints.

Climate change presents unprecedented challenges to global financial systems, necessitating quantitative methodologies for effective risk assessment and management. This talk explores climate transition risk modeling frameworks and their application in financial risk assessments. We examine how transition risks propagate through economic and financial systems via transmission channels, highlighting the roles of integrated assessment models, macroeconomic models, and the Bank of Canada’s Merton Model framework in translating climate scenarios into financial metrics. The core focus of this presentation is the OSFI Standardized Climate Scenario Exercise (SCSE), which provides a standardized approach for financial institutions to evaluate climate-related risks. We specifically focus on the methodology to assess transition risk impacts on market risk for corporate bonds. Through a comparative analysis of the oil extraction and renewable energy sectors, we demonstrate how sector-specific exposures respond differently to transition pathways and climate policies. We conclude with a discussion of key modeling challenges and emerging opportunities.

• Universal regimes for nominal, real and inflation rates
• Universal regimes across economic periods and across markets
• Fundamentals behind the universal regimes. Critical role of Central Banks. Tenor dimension
• Universal regimes across P and Q universes
• Expanding Fisher equation from rate levels to its volatilities: is knowing two out of three enough?