The Schedule

24th July 2026

Alexander Barzykin

Title: Strategic OTC market making with informational risks and reputation feedback

Abstract: Foreign exchange dealers provide bid and ask prices to their OTC clients at which they are happy to buy and sell, respectively. They can skew quotes or hedge in the interbank market to manage the inventory risk. In addition, dealers have to mitigate risks directly or indirectly related to the information they reveal to their clients and to the whole market through their actions and the risks associated with the possible informational advantage of their clients. In this talk, we are going to focus on the optimal control approach to OTC market making, specifically targeting response to risks via pricing, execution, and possibly rejection of client requests. Electronic OTC liquidity provision is increasingly shaped not only by the price of the next quote, but also by the dealer's accumulated standing with clients and platforms. We further develop a stochastic-control model in which RFQ win ratios and streaming fill ratios feed back into future flow through reputation gates, creating an explicit trade-off between immediate spread capture and long-term franchise value. The resulting policy naturally alternates between reputation-building campaigns and harvesting phases, and can generate multiple stable client-flow regimes even in a one-dealer control problem.  

James Dalby

Title: From Prompt to Portfolio

Abstract: Modern AI can collapse the distance between a research question and a working answer. Used well, a large language model becomes the connective tissue of a quantitative workflow: it orchestrates your existing tools and writes the glue code and interfaces you would otherwise build and maintain yourself. The leverage, though, comes from the context and harness you put around the model, not from casual one-line prompting.

This talk makes the case by building a real research tool live, from an empty repository, in conversation with Claude. The running example is continuous-time optimal investment, solved with a deep BSDE method: a neural policy, simulated forward and scored by expected utility. We validate it on three problems of rising difficulty: from the classical Merton solution, through a mean-reverting risk premium that makes the optimal strategy non-myopic and adds an intertemporal hedging demand, to an incomplete-market stochastic-volatility (Heston) model with no simple closed form, grading each against known results. The solver is then exposed as a tool server, so a single plain-language request drives the whole pipeline: configure, solve, validate, and plot.

Brian Healy 

Title: Decomposing Implied Volatility Surfaces

Abstract: We present a random matrix model for implied volatility surfaces, followed by a parameter estimation method based on free probability. Surface data from multiple underlying securities are arranged in a matrix $R$. This matrix is then decomposed as into a low-rank contribution from market-wide factors, and a residual. After removing the factors, we further decompose the residual into a parametric form suitable for explaining autocorrelation and cross-correlations.

Nicholas Martin

Title: Agentic Systems for Markets Operations

Abstract: Agentic systems are increasingly being adopted across financial institutions not just as a work tool for employees but also a method to automate traditionally labour-intensive processes. Markets operations is an area with many exploratory and manual workflows that have increasingly come into focus as something that could be automated using agentic systems due to the nature of their ability to look across scattered data, reason, plan and then act. This talk introduces a perspective on agentic adoption within market operations and proffers some hypothesises on future states. The conclusion is that whilst pragmatic agentic adoption is important across financial institutions, market operations are an area that agentic adoption can have transformative change within an institution in the short-to-medium term.

Lifan Xuan

Title: Data challenges in quantitative finance from a practitioner's perspective

Abstract: Quantitative finance has developed a rich set of models and methodologies for derivative pricing and risk analysis. In practice, however, the effective application of these models often depends as much on the underlying data as on the modelling framework itself, particularly as the volume and variety of available information continue to grow. 

In this talk, I will discuss several data challenges that practitioners encounter in day-to-day quantitative modelling, including data quality issues, differences in data structures across asset classes, and the difficulties of model evaluation and validation. Using examples from equity derivatives, I will illustrate how real-world data constraints can affect the practical application of otherwise well-understood models, including volatility calibration and fitting techniques. The aim is to provide academics with insight into the data realities faced by practitioners and to help bridge the gap between theoretical modelling frameworks and the conditions under which they are applied in practice.