Dr. Arzé Karam - Working Papers

Modern Market Microstructure: Liquidity, Volatility and Stability (2025)

This review was undertaken at the request of the Financial Conduct Authority (FCA) and funded through an FCA grant.

Market microstructure studies how the design and evolution of financial markets shape price formation, liquidity, and financial stability. I review the literature on the link between market liquidity and volatility, focusing on how shifts in market participation, the growth of algorithmic and high-frequency trading, the rise of passive investing, and the increasing role of non-bank institutions have influenced modern market dynamics. I also review papers on the impact of retail trading, market fragmentation across exchanges, and the role of dark pools and systemic internalisers in European and UK markets. Additionally, I conclude with a discussion on how the rise of decentralized finance (DeFi) and private markets introduces new challenges and opportunities for liquidity provision and market stability.

Full Paper (PDF) |

When Markets Get Confused: Misperception versus Inventory

with Olga Balakina, Claes Backman and Anastasiia Parakhoniak (2025)

This research project has been supported by the British Academy/Leverhulme Grant SRG1819

We exploit instances where investors mistakenly trade similar-ticker stocks to study short-term, non-fundamental pricing errors. These confusion events trigger significant abnormal returns and wider effective spreads, consistent with temporary mispricing. Our results point to misperception as the primary driver of confusion-induced mispricing, while inventory frictions play a limited role. We show that algorithmic trading accelerates price correction and narrows spreads, mitigating the impact of these belief-driven shocks. Our findings suggest that short-term non-fundamental price distortions stem from confused beliefs rather than liquidity-provider constraints, and that algorithms help restore efficiency in modern equity markets.

Full Paper (PDF) |

Presented at finance seminars at SAFE Frankfurt University and Durham University Business School.

A discrete-time quantum walk model of extreme market events a complexity-based approach

with Stuart Adams, Christopher McCarty and Jack Waller (October 2025)

Traditional financial models, such as Geometric Brownian Motion and Markov switching frameworks, struggle to explain flash crashes and liquidity-driven collapses because they treat regime shifts as exogenous and price movements as continuous diffusions. This study introduces a discrete-time Quantum Walk (QW) modeling framework that captures these dynamics as emergent, endogenous processes. The QW approach represents market evolution through quantum interference and superposition, allowing multiple potential price paths to interact—reflecting how liquidity imbalances and feedback loops amplify shocks. By parameterizing QW transitions with liquidity measures and volatility states, the model endogenously reproduces nonlinear regime shifts and sudden liquidity evaporation that precede crashes. Compared to classical stochastic and reduced-form models, QW simulations more accurately replicate fat-tailed return distributions, state-dependent volatility, and early warning signals tied to thinning market depth. Conceptually, this framework reframes market instability as a complex, self-organizing process rather than an external disturbance, offering a more adaptive and internally consistent view of how microstructural fragility evolves into systemic events.

Presented at: QRFE-2nd Quantitative Finance workshop in May 2025, Commodities Futures Trading Commission March 2025, Central Bank of Ireland February 2025, Financial Conduct Authority FSEG principals meeting October 2024

Full Paper (PDF) | Blog II: Quantum Walk for liquidity crises predictions |

Ethereum Urgency Score

with Dimitar Bogoev and Jiexuihui Chen (2025)

Blockchain technology promises decentralized transparency but remains constrained by behavioral opacity, leaving market participants and regulators with limited visibility into the informational quality of on-chain activity. This paper introduces the Urgency Score, a real-time, transaction-level metric that quantifies execution urgency in Ethereum transactions using six behavioral signals, including gas fees, timing patterns, trade size, and atypical counterparty interactions. Using reduced-form price impact models and machine learning (XGBoost), we show that high-urgency transactions contribute disproportionately to ETH/USD price discovery, especially during market stress. Component-level analysis reveals that urgency dimensions are largely independent under normal conditions but cluster episodically around major events (e.g., the Ethereum Merge, FTX collapse), generating systemic urgency waves that amplify information flow and liquidity shocks across DeFi and centralized exchanges. Urgency-based momentum strategies exploiting these signals deliver consistently positive, risk-adjusted returns, while low-urgency flows have limited or negative predictive content. These findings provide a scalable, blockchain-native framework for identifying informational asymmetries and execution strategies in real time, bridging blockchain behavior with established theories of informed trading. They also highlight the need for new theoretical models of decentralized market microstructure that endogenize urgency propagation under blockchain-specific constraints, informing both market design and supervisory oversight.

Presented at One-Day QRFE workshop on Blockchain-based markets, FinTech and Cryptocurrencies.

Full Paper (PDF)

Quote Volatility Pressure and Intraday Momentum

with Dimitar Bogoev (2025)

We develop a novel machine learning-based measure of illiquidity, centered on a Quote Volatility metric that captures short-term oscillations in bid-ask quotes in high-frequency markets. Trained on episodes of order book thinning and rapid quote reversals, our model incorporates nonlinear interactions between market depth, price impact, and large trading volumes. The resulting illiquidity metric effectively detects real-time liquidity stress, particularly during intraday momentum trading, where trend-following strategies amplify price dislocations during periods of deleveraging. We observe that spikes in quote volatility occur both before and during systemic liquidity crises — including the COVID-19 market selloff (March 2020), the onset of the war (early 2022), the tariffs trade war (April 2025), and the Japanese market crash (August 2024) while remaining stable in normal periods. This real-time metric provides predictive insight into market resilience, execution risk, and trading costs, offering a valuable tool for arbitrageurs, market makers, and regulators in mitigating liquidity-driven disruptions.

Presented at: Central Bank of Ireland 2025, Financial Conduct Authority FSEG principals meeting 2024, QRFE one-day workshop on Market Microstructure, Fintech and AI, IFABS meetings at Oxford University, Commodity Futures Trading Commission, Paderbon University and Leicester University.

Full Paper (PDF)| Blog I: Liquidity Crisis Prediction using ML|