Liquidation of collateral are the primary safeguard for solvency of lending protocols in decentralized finance. However, the mechanics of liquidations expose these protocols to predatory price manipulations and other forms of Maximal Extractable Value (MEV). In this paper, we characterize the optimal liquidation strategy, via a dynamic program, from the perspective of a profit-maximizing liquidator when the spot oracle is given by a Constant Product Market Maker (CPMM). We explicitly model Oracle Extractable Value (OEV) where liquidators manipulate the CPMM with sandwich attacks to trigger profitable liquidation events. We derive closed-form liquidation bounds and prove that CPMM transaction fees act as a critical security parameter. Crucially, we demonstrate that fees do not merely reduce attacker profits, but can make such manipulations unprofitable for an attacker. Our findings suggest that CPMM transaction fees serve a dual purpose: compensating liquidity providers and endogenously hardening CPMM oracles against manipulation without the latency of time-weighted averages or medianization.

In finance, sequential decision problems are often faced, for which reinforcement learning (RL) emerges as a promising tool for optimisation without the need of analytical tractability. However, the objective of classical RL is the expected cumulated reward, while financial applications typically require a trade-off between return and risk. In this work, we focus on settings where one cares about the time split of the total return, ruling out most risk-aware generalisations of RL which optimise a risk measure defined on the latter. We notice that a preference for homogeneous splits, which we found satisfactory for hedging, can be unfit for other problems, and therefore propose a new risk metric which still penalises uncertainty of the single rewards, but allows for an arbitrary planning of their target levels. We study the properties of the resulting objective and the generalisation of learning algorithms to optimise it. Finally, we show numerical results on toy examples.

Combining forecasts from multiple experts often yields more accurate results than relying on a single expert. In this paper, we introduce a novel regularized ensemble method that extends the traditional linear opinion pool by leveraging both current forecasts and historical performances to set the weights. Unlike existing approaches that rely only on either the current forecasts or past accuracy, our method accounts for both sources simultaneously. It learns weights by minimizing the variance of the combined forecast (or its transformed version) while incorporating a regularization term informed by historical performances. We also show that this approach has a Bayesian interpretation. Different distributional assumptions within this Bayesian framework yield different functional forms for the variance component and the regularization term, adapting the method to various scenarios. In empirical studies on Walmart sales and macroeconomic forecasting, our ensemble outperforms leading benchmark models both when experts' full forecasting histories are available and when experts enter and exit over time, resulting in incomplete historical records. Throughout, we provide illustrative examples that show how the optimal weights are determined and, based on the empirical results, we discuss where the framework's strengths lie and when experts' past versus current forecasts are more informative.

It is well established that the US prewar output was more volatile and less shock persistent than the postwar output. This is often attributed to the data interpolation employed to construct the prewar series. Our analytical results, however, indicate that commonly used linear interpolation has the opposite effect on shock persistence and volatility of a series - it increases shock persistence and reduces volatility. The surprising implication of this finding is that the actual differences between the volatility and shock persistence of the prewar and postwar output series are likely greater than the existing literature recognizes, and interpolation has dampened rather than magnified this difference. Consequently, the view that postwar output was more stable than prewar output because of the effectiveness of the postwar stabilization policies and institutional changes has considerable merit. Our results hold for parsimonious stationary and nonstationary time series commonly used to model macroeconomic time series

Japan's population is shrinking, the share of working-age people is falling, and the number of elderly is growing fast. These trends squeeze public finances from both sides--fewer people paying taxes and more people drawing on pensions and healthcare. Policy discussions often focus on one fix at a time, such as raising taxes, reforming pensions, or boosting productivity. However, these levers interact with each other through feedback loops and time delays that are not yet well understood. This study builds and calibrates an integrated system dynamics model that connects demographics, labor supply, economic output, and public finance to explore two questions: (RQ1) What feedback structure links demographic change to fiscal outcomes, and how do different policy levers work through that structure? (RQ2) Which combinations of policies can stabilize key fiscal indicators within a meaningful timeframe? The model, grounded in official statistics, tracks historical trends reasonably well. Policy experiments show that productivity improvements and controlling per-person costs offer the most effective near-term relief, because they act quickly through revenue and spending channels. In contrast, raising fertility actually worsens the fiscal picture in the medium term, since it takes decades for newborns to grow up and join the workforce. A combined scenario pairing moderate productivity gains with moderate cost control nearly eliminates the deficit by 2050. These findings underscore the importance of timing when evaluating demographic policy. Stabilizing finances within a practical timeframe requires levers that improve the budget directly, rather than those that work through slow demographic channels. The model serves as a transparent testing ground for designing time-aware fiscal policy packages in aging, high-debt economies.

We address the problem of estimating the expected shortfall risk of a financial loss using a finite number of i.i.d. data. It is well known that the classical plug-in estimator suffers from poor statistical performance when faced with (heavy-tailed) distributions that are commonly used in financial contexts. Further, it lacks robustness, as the modification of even a single data point can cause a significant distortion. We propose a novel procedure for the estimation of the expected shortfall and prove that it recovers the best possible statistical properties (dictated by the central limit theorem) under minimal assumptions and for all finite numbers of data. Further, this estimator is adversarially robust: even if a (small) proportion of the data is maliciously modified, the procedure continuous to optimally estimate the true expected shortfall risk. We demonstrate that our estimator outperforms the classical plug-in estimator through a variety of numerical experiments across a range of standard loss distributions.