LQG 2025/09/03 to 2025/09/06 In-Person & On-Line Only Autumn Seminar at Robinson College, Cambridge

LQG Journal 2025

The LQG autumn seminar was, as always, graced by a range of fine presentations.  With LQG’s thanks to our speakers and my apologies for the inevitable misunderstandings and omissions.  Robert Macrae

David Buckle, Inquire
E[g(X_1, …, X_n)] = ∫[0,1] … ∫[0,1] g(F_1^(-1)(u_1), …, F_n^(-1)(u_n)) . c(u_1,…,u_n) du_1 … du_n  for dummies

David explained how to sample from a kernel-smoothed 1D empirical distribution.  For the 1% tails he assumed exponentials chosen to match empirical skew and kurtosis, resulting in blindingly fast run-time typically requiring 1 linear interpolation occasionally one log.

Armed with this he approached multivariate sampling via Gaussian copulas, using a root-finding algorithm to ensure that observed correlations are preserved as in general these differ from the copula correlations.  So conditional on tail co-dependence and other requisite copula assumptions you can sample from the prospective returns of any portfolio, and hence directly optimise any utility function based on returns.  You can also scale and shift those returns to investigate how different assumptions should affect choices, for example to investigate the impact of house strategic views.

Naima Hammoud, Bloomberg

Extracting transient factors from cross-sectional residuals

Big market moves (in particular) induce correlations in idiosyncratic returns to a pre-specified factor model.  Naima pointed out that the RMS T-statistic of a factor K can be interpreted as bias to risk if K omitted and shows that performance is significantly improved by incorporating the principal component of the residuals as an additional factor.  105 days is a sensible estimation half-life and shows interesting dynamics around large events such as Covid, possibly picking up an AI factor.  Simon questioned whether weighting by inverse specific variance, despite its elegance, necessarily improved performance for near-cap-weighted portfolios.

David Aikman, NIESR
In the Fed we trust?

Picking a fight with the Chancellor in order to maximise the newsworthiness of an LQG talk sets a high bar for future inaugural speakers.

Money is a social convention underwritten by trust, and trust in central banks should increase resilience to political interference.  To measure trust David uses tweets.  While substantial problems are involved — tribal and bot activity, and according to Ely “People with nothing better to do” — they can sensibly consider covariates and use scandal-related resignations as a plausibly-exogenous instrument.

Fine tuning proved unsatisfactory so they just asked ChatGPT-4o mini to categorise 4 million central-bank-related tweets and retweets as positive, neutral or negative on trust and accumulated the smoothed result as their trust index. This suggests a plausible pattern with falling trust under Trump 1; partial recovery under Covid; and a further fall under inflation.  Given the smooth nature of the index the regression and VAR analyses appeared interesting rather than compelling.

Jose Menchero, Bloomberg
Factor misalignment in portfolio construction

To investigate the quality of a risk model you should compare realised to forecast volatility on hedged portfolios because a long only constraint regularises and can hide problems.  Jose removed each factor in turn from his prespecified-factor covariance model and instead treated the factor exposures as alphas to create portfolios highly exposed to the missing factor and so give a sensitive test of the value of including that factor.  He concluded that gains from inclusion were on the scale of 6% reduction in realised volatility, which is comparable to the estimated gain from using inverse specfic variance weights over sqrt(market cap).   By contrast there seems no value in including alphas to the extent that they are orthogonal to all factors.

Annika Rao-Monari, CleeAI
From syntactic to semantic AI

Chatbots know what an answer looks like and slot material into this structure as appears most appropriate, playing Tetris with sentences.  Syntax is king and while results can be impressive, LLMs hallucinate when they cannot find a reality-based sentence that is a good fit to the emerging pattern.  Annika argued that to progress further we may need a more evaluative approach centred on concepts rather than sentences.

James Eagle
Data visualisation

With Trump firing Fed governors and a labour statistics commissioners, can we assume US institutional reports will remain accurate?  While outright fabrication may remain rare there is endless scope for the manipulation of methodology and presentation.  The less we can trust institutions the more effort we will have to invest in verification and interpretation, and the greater the need for trustworthy, effective and digestible data-based narratives.  James illustrated with a number of examples.

Desi Ivanova, Oxford

Complexity, uncertainty and learning

Desi gave a thorough analysis of the Virtues of Complexity and double descent, showing that the modern ML incarnation is closely related to the familiar pseudoinverse effect in which underspecified regressions are fitted with minimum norm.   She also showed that the influential Kelly paper implied that radial basis functions would perform even better on their problem than their preferred random fourier factor approach, and with vastly less computation.

More generally, Desi argued for a change in approach away from explicitly parsimonious models towards large, rich, complex, non-convex models estimated via gradient descent and with overfitting controlled by careful use of holdouts, dropouts and other modern approaches.  Overfitting should only be a problems when those models are tested outside the domain in which they have been trained.   Old-school quants may well agree… but, bruised by long experience, also suspect that the future holds many such places…

Dori Levanoni, Capstone
Optimal currency hedging

International investments involve exposure to both an asset and a currency and Dori asked how investors might hedge the latter.  Since 1974 from the perspective of a US or UK investor the currency exposure appears to have a small positive association with equity assets but also a small positive return.  Of course for Japanese investors the situation appears very different.  While the USD has been widely discussed as a safe haven since GFC this was far less prevalent in the decades following the collapse of Bretton Woods.  Dori considered a trading strategy that implemented whatever hedge would have performed best during the previous 3 years and found some marginally LQG-compliant evidence that this had performed well.

Pedro Castro, Man Group
FX hedging

Currency volatility is on the scale of 11%, too large to be ignored, and institutions typically hedge about half their FX exposure.  This roughly corresponds to the historic mean-variance optimising equity hedge over 50 years, but outcomes across markets and decades show huge dispersion.  Pedro presented some evidence that hedging so as to maximise carry may perform better than a fixed strategy, though in developed markets the gains are small and mainly accrue before GFC. There may be a small additional improvement from penalising recent volatility, while momentum and PPP seem to add little.

Simon Polbennikov, Barclays
Bond performance attribution

RAIL — return attribution at issue level — starts with a bond’s excess return over duration-matched treasuries and decomposes as spread carry; systematic; and specific components. Systematic contains market; spread duration; quality; and sector components.  All are scaled by DTS (Duration * Spread) ratios.  These systematic exposures account for a substantial part of bond volatility and removing them permits more accurate attribution of performance.  For example, Simon estimated that analyst recommendations generate around 0.15% over the following 6 months.

Ed Fishwick, Blackrock
Themes repeat

Some LQG luminaries  — the Vivaldis of finance — recycle remarkably similar material for decades, whether reverse optimisation or meditations on Chapter 12 uncertainties.

Ed noted that smoothing in PE and RE markets rendered them unrealistically attractive compared to public equities in terms of volatility-based metrics.  However, he questioned which valuation was more appropriate.  In the 1700s when packet boats could not sail between Amsterdam and London the volatility of dual-listed stocks was little changed despite the absence of news.  Should we conclude that excess volatility is intrinsic to public markets; that short-term volatility is a poor risk measure; that public markets should converge towards private market valuations; and that  “The rest is noise”?  This last is apparently a tribute to Alex Ross’ magisterial review of 20th century classical music.

In this talk I will demonstrate the method, discussed in Buckle and de Jong (2024), of simulating non-normal random returns, for many correlated assets, then using such a simulation to construct portfolios.
First, I will show how to simulate returns of a single asset from a return distribution who’s non-normal shape is derived from historical data (but with specified expected future return and volatility).
Second, I will show how to use copulas to simulate returns of many assets such that the returns have some specified correlation, but each asset’s marginal distribution is as in step 1.
Third, I will show how to construct portfolios from these non-normally distributed assets.
We will see that the non-normality does impact portfolios selection relative to the traditional Markowitz approach.
A python script of the entire process will be made available to any audience member scoring me highly on the speaker ratings form!

Chairman
Inquire UK

David is a long term friend of and regular speaker at the LQG.  He has been in the asset management industry for over 25 years, primarily as a portfolio manager or analyst. He has been made redundant many more times than he would have liked and therefore has worked for many blue-chip asset managers including JP Morgan Asset Management, Putnam Investments, Merrill Lynch Investment Managers, Blackrock, UBS Global Asset Management and Fidelity. He also ran an investment boutique for several years deploying overlay strategies. Throughout his career he has contributed to the investment literature, especially in the area of the theory of active management.
Having become a bit disenchanted with the industry of late, David now has a plurality of roles, including being the CEO and CIO of a tiny real asset investment company, running a handful of multi asset portfolios, and is the chairman of the Institute for Quantitative Investment Research (INQUIRE). He spends much of his time authoring articles and is in the process of writing a couple of books on investment matters.
David’s background is as a Phd. mathematician and that discipline remains his hobby. His most recent article is the interpretation of some cuneiform mathematics scribed on a Babylonian stone from 2000BCE.

During periods of market turmoil, cross-sectional factor models may not fully capture all sources of systematic risk. In this presentation, we study how “transient” factors can be extracted from cross-sectional residuals using Principal Component Analysis. We study the strength of the factors using t-statistics and find that, in practice, transient factors are comparable to a strong style factor. We also determine the theoretical conditions under which the factor may be successfully extracted and confirm this result via numerical simulations. We then conduct an empirical study in which we omit a strong factor from a factor risk model and test how well the extracted factor replicates the omitted factor. Finally, we offer some insight into interpreting the transient factors.

Naima Hammoud
Bloomberg

Naima Hammoud joined Bloomberg in August 2022, where she has conducted research spanning both fixed income and equity markets. Within fixed income, Naima worked on portfolio risk models focused on developed markets. She also built a specialized methodology for managing the risk associated with defaulted debt. Currently, she is developing statistical techniques aimed at identifying transient factors in equity markets.
Prior to joining Bloomberg, Naima served as an Assistant Professor of Mathematics at New York University’s Courant Institute of Mathematical Sciences. She holds a Ph.D. in Applied Mathematics from Princeton University, obtained in 2016.

This paper proposes a novel framework to analyse how nonlinear financial thresholds affect downside risk to the macroeconomy and to asset returns. We identify three financial instability (FI) zones — capturing housing market overheating, corporate credit overheating, and depleted financial intermediary capital — and show, using panel quantile regressions and a Bayesian VAR, how crossing these thresholds reshapes the forward distribution of GDP growth. We also examine the implications for risk-return trade-offs in asset markets. Our findings support the use of this framework for macro-financial surveillance, policy design, and risk management.

David Aikman
The National Institute of Economic and Social Research

Prior to joining NIESR, David was Professor (in Practice) of Finance at King’s Business School and the inaugural Director of the Qatar Centre for Global Banking and Finance. He is the co-organiser (with Richard Barwell) of the Bank of England Watchers’ conference, and he co-edits with Richard a blog dedicated to financial stability policy (www.macroprudentialmatters.com). He is currently co-editing (with Prasanna Gai) a new Research Handbook of Macroprudential Policy, which will be published by Edward Elgar.

David is a member of the Bank of England’s Macroprudential Roundtable. Until July 2025, he was a member of the Prudential Regulation Authority’s Cost Benefit Analysis Panel. He has recently been an external expert at the Central Bank of Ireland, the Bank of Canada, and the Central Bank of the UAE.

David spent 17 years working as an economist at the Bank of England – most recently in the role of Technical Head of Division in Financial Stability where he led the Bank’s work on macroprudential policy strategy. Between 2013 and 2015, he was seconded to the Board of Governors of the Federal Reserve System in Washington DC, where he worked as an advisor in the Division of Financial Stability. In 2023, David was a Visiting Professor at Keio University, Tokyo. In 2008, David was a Visiting Scholar at the Bank of Japan’s Institute for Monetary and Economic Studies. David represented the Bank of England in various international fora, including meetings of the Financial Stability Board, the Basel Committee, and the European Systemic Risk Board.

He is the author of various research papers on financial stability and macroprudential policy, and has a Ph.D. in Economics from the University of Warwick.

Factor misalignment in portfolio construction is said to occur when mean-variance optimization is performed on an alpha factor that is not contained within the set of risk model factors. It is well established that portfolio optimizers tend to tilt the portfolio in the direction of the residual alpha, which is the component of alpha that is orthogonal to the risk model factors. Some practitioners are concerned that this may lead to a reduction in portfolio efficiency and argue that building a custom risk model that includes the alpha factor can lead to sizeable gains in portfolio efficiency. In this presentation, we study this question both through realistic numerical simulations and actual empirical backtests. We find that for most factors, including the alpha factor in the risk model leads to only modest gains in portfolio efficiency and in some cases may even lead to a decline in portfolio efficiency. Finally, we provide some guidance and insight on which alpha signals are likely to benefit from inclusion in the risk model and which are not.

Jose Menchero

Jose Menchero
Bloomberg

Jose Menchero serves as Head of Portfolio Analytics Research at Bloomberg. Jose and his team are responsible for developing the analytics and algorithms used for factor risk models, portfolio risk and return attribution, scenario analysis, tail risk, portfolio construction, and portfolio optimization.
Prior to joining Bloomberg, Jose was the founder and CEO of Menchero Portfolio Analytics Consulting. Before founding his consulting firm, Jose worked for eight years at MSCI, where he was Managing Director responsible for developing the Barra equity risk models, and for portfolio construction research. Jose also served for seven years as Director of Research at Thomson Financial, where he developed several risk and return attribution methodologies, as well as equity factor risk models.
Jose has over 30 finance publications in leading practitioner journals. Before entering finance, Jose was Professor of Physics at the University of Rio de Janeiro in Brazil. Jose holds a PhD in theoretical physics from the University of California at Berkeley, and a BS degree in aerospace engineering from the University of Colorado at Boulder. Jose is a CFA Charterholder.

Large Language Models (LLMs) rely on surface-level language pattern recognition — generating fluent answers without understanding meaning, logic, or consequences. In high-stakes environments like finance, this isn’t just insufficient — it’s dangerous. This talk explains how to bring cognitive intelligence to LLMs, replacing generative guesswork with contextual, concept-driven reasoning that mimics the way humans analyze risk, weigh alternatives, and adapt over time. We’ll explore how contextual intelligence architecture enables precision, transparency, and dynamic learning by moving beyond tokens and into true semantic intelligence — making AI safe, auditable, and actually usable in the real-world workflows of modern risk officers.

Annika Rao-Monari
CleeAI

Annika Rao-Monari is the Chairman, Co-Founder and CEO of CleeAI, a pioneer in infrastructure-level AI that builds AI—developing custom agents across industries for enterprise. She holds a Master of Science degree from Imperial College London, where she conducted her thesis research at CERN, and was recognised as a Forbes 30 Under 30 honouree. She previously co-founded several deep tech ventures in AI and blockchain—including Aventus Protocol Network, Coral, and Artos.
CleeAI’s core innovation is the Language-Knowledge Model (LKM): a new AI stack purpose-built for enterprise agents. Unlike traditional LLMs, LKMs are small, fast, and precise—capable of training and deploying custom agents for any role using real-time data, without requiring prompts, GPUs, or cloud integration.
Annika’s work is redefining how institutions use AI—not as static copilots—but as autonomous, explainable systems that understand, reason, and act. Raised across cultures and fluent in four languages, she brings both global perspective and deep technical rigor to the future of intelligent automation.

Visual Storytelling in an Age of Information Overload

Founder
Eeagli.com

James is a Zürich-based data-visualisation specialist and former buy-side economist. His first book, “Visual Storytelling with Data” (Wiley, 10 July 2025), distils a decade of helping asset-management and fintech teams turn complex models into memorable charts. He’s currently co-founder of Plotset, a low-code platform that brings studio-quality visuals to non-coders. Outside work he mentors early-career quants and occasionally attempts to tame Swiss mountain trails on a gravel bike.

Overfitting is widely regarded as a central challenge in machine learning, with complexity controls, bias-variance trade-offs, and careful test set usage presented as fundamental safeguards. But how well do we understand these concepts in practice? What exactly constitutes overfitting across different contexts, and how should we think about model complexity in modern statistical learning? In this talk, we’ll examine these questions through examples from computer vision, natural language processing, and quantitative finance, and consider how different kinds of overfitting—from statistical noise fitting to domain shift—shape our understanding of model performance. We’ll specifically focus on recent work exploring the “virtue of complexity” in finance (Kelly et al., 2024), connect it to classical ML methods, and discuss practical approaches for incorporating uncertainty as a tool for model selection.

Desi Ivanova
University of Oxford

Desi R Ivanova is a Senior Research Fellow (equivalent to Assistant Professor) in computational statistics and machine learning at the University of Oxford’s Department of Statistics. She earned her PhD at Oxford, receiving the G-Research “Best Thesis” prize in 2024. Her research focuses on data-efficient machine learning, covering areas such as optimal experimental design, amortized inference, and causality, and uncertainty quantification, particularly in the context of LLMs (and AI foundation models more broadly). Before academia, she worked as a quantitative researcher at UBS and Goldman Sachs in London.

The complete guide to currency hedging

Investing internationally, regardless of asset class, involves buying two assets: the underling investment and the currency it’s denominated in. That currency asset is not a small asset and it can have a large impact on the risk and return of the overall portfolio. The decision of what to do with that unintended currency exposure must be made. You can choose to get rid of it, keep it, or do something in between. But how should you choose what to do? We’ll go through all the dimensions of the problem and provide a roadmap to finding the answer.

Managing Director
Capstone

Dori Levanoni joined Capstone in November 2023 as a Managing Director, Senior Portfolio Manager, responsible for managing the Capstone Currency Strategy on Capstone’s Solutions Platform.

Prior to joining Capstone, Dori served as the Chief Investment Strategist and Head of Global Macro at First Quadrant. While at First Quadrant, Dori acted as a portfolio manager to many of the firm’s global macro and global equity strategies, in addition to contributing to various research projects and publications. Having spent over 30 years at First Quadrant, Dori was heavily involved in the firm’s investment process, risk assessment, and portfolio optimization and management. Dori started his career at First Quadrant in 1991, spending a brief period researching in the Department of Anatomy and Neurobiology at Washington University in St. Louis in the mid-1990s.

The question of whether, when, and how to hedge foreign exchange risk has been a vexing one for investors since the end of the Bretton Woods system in 1973. Our study provides a comprehensive empirical analysis of dynamic FX hedging strategies over 50 years, examining dynamic FX hedging strategies for 20 currency pairs. While traditional approaches often focus on static long term risk mitigation, we explore the broader implications for expected returns and dynamic management.

Pedro Castro
MAN Group

Pedro is a Managing Director and Head of Portfolio Engineering within Solutions, responsible for all quantitative analysis, portfolio construction modelling and research.
Pedro joined Man Group in 2024. Prior to this, he was Director of Risk and Quantitative Analysis at BlackRock, covering Multi-Asset Strategies and Solutions for 13 years. He was also responsible for driving the adoption and evolution of BlackRock’s strategic portfolio construction research across their investment and consulting platforms. Before that, he was a finance assistant lecturer at the University of Warwick.
Pedro holds a MSc in Finance and Economics, with distinction, from Warwick Business School and a BSc in Economics from Nova SBE. He is a CFA charterholder and a certified Financial Risk Manager (FRM).

Simon will introduce Return Attribution at Issue Level (RAIL) for decomposing returns of corporate bonds into interpretable systematic and security-specific components. RAIL can be applied to track performance of individual recommendations or active portfolios or to validate systematic credit strategies. Unlike traditional attribution models, RAIL operates at the security level and is independent of portfolio construction. It enables performance analysis on a bond-by-bond basis, with the flexibility to aggregate results for portfolio-level insights.

Key applications include:

– Event studies – assessing the impact of earnings announcements, litigation, corporate actions, or adverse news
– Portfolio performance attribution – particularly for active portfolios that diverge from benchmark indices
– Analyst performance tracking

Simon Polbennikov

Simon Polbennikov is a Managing Director in the Quantitative Portfolio Strategy group at Barclays, where he leads empirical research on systematic credit, investment styles, portfolio construction, and risk. His work supports institutional investors with data-driven insights into the investment process. The Quantitative Portfolio Strategy (QPS) team has been ranked #1 in Quantitative Analysis in both the US and Europe by the Extel/Institutional Investor survey for four consecutive years (2022–2025).

Simon is the co-author of two influential books: Systematic Investing in Credit (Frank J. Fabozzi Series, Wiley, 2021) and Measuring ESG Effects in Systematic Investing (The Wiley Finance Series, 2024), which explore the application of quantitative techniques in credit and ESG investing. He joined Barclays in 2008 from Lehman Brothers, where he held a similar role. Simon holds a degree in physics from Moscow State University and a PhD in empirical finance from Tilburg University in the Netherlands.

It is risky to forecast – especially about the future

Ed Fishwick
Senior Managing Director, Chief Risk Officer, BlackRock

Ed Fishwick is the Chief Risk Officer, Global Head of Risk and Quantitative Analysis, and a member of the Global Executive Committee at BlackRock. He is responsible for the investment and enterprise risk of BlackRock. Prior to taking on his current responsibilities in 2022, Mr. Fishwick served as the Global Co-Head of the Risk and Quantitative Analysis Group in London for 15 years. His service with the firm dates back to 2003.

Mr. Fishwick’s previous positions in the industry include Head of Risk Management, and Investment Process Research at AXA Investment Managers, and Head of Research at Franklin Portfolio Managers in Boston.

Ed is the Chairman of the London Quant Group.