Automated market makers (AMMs) are a cornerstone of decentralised finance (DeFi). Constant product markets with concentrated liquidity, such as UniswapV3, are now a well-established design. In these markets, liquidity providers (LPs) face a sequential decision problem: they must decide when to rebalance their positions and which price ranges to allocate capital to as market conditions evolve. We formulate dynamic liquidity provision as a stochastic impulse control problem and use reinforcement learning (RL) to solve it, focusing on providing interpretable solutions. We show that learned policies exhibit rich state-dependent behaviour, allocating liquidity according to mispricing, rebalancing costs, uncertainty, inventory exposure, and heterogeneous risk preferences. These behaviours help compress the left tail of the Profit and Loss (PnL) distribution and avoid catastrophic outcomes under high uncertainty. Finally, we benchmark the RL agents against baseline and sophisticated agents from the AMM microstructure literature and analyse their performance.
The results suggest that adaptive interaction among AI trading systems may create new forms of algorithmic vulnerability relevant for exchanges, trading venues, execution desks, and market-surveillance teams.
Sudip Gupta· The Journal of Financial Dat...· 0 citations
This work proposes Scenario-Context Rollout (SCR), a macroeconomics-guided feedback mechanism to produce a distribution of next-day joint returns under potential economic shocks, and theoretically analyze this problem and shows that combining scenario-scored rewards with tape-realized transitions induces a hybrid fixed point.
Vanya Priscillia Bendatu, Yao Lu· Proceedings of the 32nd ACM...· 0 citations
SHAP analysis reveals that trading activity, lagged liquidity, and market uncertainty are the main determinants of liquidity forecasts, and the findings highlight the complementary role of explainable machine learning in empirical finance.
Veni Arakelia, G. Caporale, Mirto M Gasparinatou et al.· CESifo working papers· 0 citations
Robust hedging valuation adjustment (HVA) is applied as a post-training valuation-adjustment layer that evaluates tracking-loss CVaR together with explicit funding and margin add-ons together with explicit funding and margin add-ons.
A closed-loop multi-agent decision framework that introduces prompt-level learning as a scalable alternative to full model retraining and highlights the potential of prompt-level adaptation for building robust and autonomous financial decision systems.
Kandarp Mukeshkumar Sharda, Aliyu Sani Sambo· NLP & Big Data· 0 citations
Classical put-overlays have long been treated as a reliable hedge against tail risk but the market conditions of 2025 expose their limits in ways that theory didn’t fully anticipate. This paper examines where these strategies break down: in markets defined by elevated volatility, wide bid-ask spreads, and structural frictions that quietly erode the protection investors thought they’d bought. Traditional hedging methods, which rely on the systematic purchase of out-of-the-money (OTM) options, are increasingly hampered by high premiums and reliance on static volatility assumptions. During the significant geopolitical disruptions of 2025, most notably the “Tariff Shock” of April, these traditional models proved inadequate, suffering from “premium bleed” and an inability to account for discontinuous market gaps. To address these systemic vulnerabilities, we formulate the hedging process as a stochastic control problem and implement a Deep Hedging framework utilising Deep Reinforcement Learning (DRL). Optimised specifically for Expected Shortfall (ES), the model utilises Long Short-Term Memory (LSTM) layers to process multi-dimensional state vectors, including implied volatility skew and realised turbulence. Our empirical results demonstrate that this AI-optimised policy achieves a 35% reduction in hedging costs while simultaneously improving tail risk protection and draw-down resilience. Notably, the DRL agent exhibits anticipatory behaviour, transitioning from a reactive to a predictive paradigm by adjusting hedge positions prior to observable volatility spikes. These findings suggest that in structurally incomplete markets, optimal risk management has evolved from simple insurance into a process of continuous, regime-aware policy optimisation. This study provides a robust framework for institutional solvency in an era defined by non-linear correlations and rapid liquidity decay.
Y. Chakrabarti· American Journal of Financia...· 0 citations