We introduce a semiparametric approach for forecasting Value-at-Risk (VaR) and Expected Shortfall (ES) by modeling the conditional scale of financial returns, defined as the difference between two specified quantiles, via restricted quantile regression. Focusing on downside risk, VaR is derived from the left-tail quantile of rescaled returns, and ES is approximated by averaging quantiles below the VaR level. The method delivers robust, distribution-free estimates of extreme losses and captures skewness, heavy tails, and leverage effects. Simulation experiments and empirical analysis show that it often outperforms established models, including GARCH and joint VaR-ES conditional-quantile approaches. An application to daily returns on major international stock indices, spanning the COVID-19 period, highlights its effectiveness in capturing risk dynamics.

Large Language Models (LLMs) excel at general code generation, yet translating natural-language trading intents into correct option strategies remains challenging. Real-world option design requires reasoning over massive, multi-dimensional option chain data with strict constraints, which often overwhelms direct generation methods. We introduce the Option Query Language (OQL), a domain-specific intermediate representation that abstracts option markets into high-level primitives under grammatical rules, enabling LLMs to function as reliable semantic parsers rather than free-form programmers. OQL queries are then validated and executed deterministically by an engine to instantiate executable strategies. We also present a new dataset for this task and demonstrate that our neuro-symbolic pipeline significantly improves execution accuracy and logical consistency over direct baselines.

We study optimal auction design for Maximum Extractable Value (MEV) auction markets on Ethereum. Using a dataset of 2.2 million transactions across three major orderflow providers, we establish three empirical regularities: extracted values follow a log-normal distribution with extreme right-tail concentration, competition intensity varies substantially across MEV types, and the standard Revenue Equivalence Theorem breaks down due to affiliation among searchers' valuations. We model this affiliation through a Gaussian common factor, deriving equilibrium bidding strategies and expected revenues for five auction formats, first-price sealed-bid, second-price sealed-bid, English, Dutch, and all-pay, across a fine grid of bidder counts $n$ and affiliation parameters $ρ$. Our simulations confirm the Milgrom-Weber linkage principle: English and second-price sealed-bid auctions strictly dominate Dutch and first-price sealed-bid formats for any $ρ> 0$, with a linkage gap of 14-28\% at moderate affiliation ($ρ=0.5$) and up to 30\% for small bidder counts. Applied to observed bribe totals, this gap corresponds to \$10-18 million in foregone revenue over the sample period. We also document a novel non-monotonicity: at large $n$ and high $ρ$, revenue peaks in the interior of the affiliation parameter space and declines thereafter, as near-perfect correlation collapses the order-statistic spread that drives competitive payments.

Economic growth is conventionally analyzed at the national level, yet cities generate the bulk of global output. Here we construct GDP trajectories for 8,808 functional urban areas (FUAs) across 165 countries over 1993-2019 using satellite-derived nighttime light data and identify 17 distinct, persistent growth regimes through clustering of full temporal trajectories. Rather than converging toward a common frontier, FUAs inhabit distinct economic niches-analogous to ecological niches-defined by shared volatility profiles, shock responses, and long-run dynamics that transcend national boundaries. Cities within the same country frequently belong to different regimes, while structurally similar cities on different continents share the same one; regime membership explains 16% of within-country growth variance beyond country fixed effects. National-level convergence emerges as an aggregation artifact: conditional convergence operates within regimes, not globally. A directed propagation network reveals that shocks transmit along lines of structural similarity rather than geographic proximity, with advanced economies exporting disturbances and emerging economies absorbing or amplifying them. Within-country spatial inequality declines with industrialization maturity, consistent with growth initially concentrating in leading cities before diffusing across the urban system. The global economy is better understood as an ecology of heterogeneous urban growth regimes than as a collection of nations on a shared development path.

Auto-deleveraging (ADL) mechanisms are a critical yet understudied component of risk management on cryptocurrency futures exchanges. When available margin and other loss-absorbing resources are insufficient to cover losses following large price moves, exchanges reduce positions and socialize losses among solvent participants via rule-based ADL protocols. We formulate ADL as an optimization problem that minimizes the exchange's risk of loss arising from future equity shortfalls. In a single-asset, isolated-margin setting, we show that under a risk-neutral expected loss objective the unique optimal policy minimizes the maximum leverage among participants. The resulting design has a transparent structure: positions are reduced first for the most highly levered accounts, and leverage is progressively equalized via a water-filling (or ``leverage-draining'') rule. This policy is distribution-free, wash-trade resistant, Sybil resistant, and path-independent. It provides a canonical and implementable benchmark for ADL design and clarifies the economic logic underlying queue-based mechanisms used in practice. We further study the multi-asset, cross-margin setting, where the ADL problem becomes genuinely multi-dimensional: the exchange must allocate a vector of required reductions across accounts with portfolios exposed to correlated price moves. We show that under an expected-loss objective the problem remains separable across accounts after introducing asset-level shadow prices, yielding a scalable numerical method. We observe that naive gross leverage can be misleading in this context as it ignores hedging within portfolios. When asset prices are driven by a single dominant risk factor, the optimal policy again takes a water-filling form, but now in a factor-adjusted notion of leverage, so that more effectively hedged portfolios are deleveraged less aggressively.

We study cubic cardinality-constrained portfolio optimization, a higher-order extension of the standard Markowitz formulation where three-way sector co-movement terms augment the quadratic risk-return objective. Classical heuristics like simulated annealing (SA) and tabu search require Rosenberg quadratization of these cubic interactions. This inflates the variable count from n to 5n and introduces penalty terms that substantially distort the augmented search landscape. In contrast, Hyper-Adaptive Momentum Dynamics (HAMD) operates directly on the native higher-order objective using a hybrid pipeline combining continuous Hamiltonian search, exact cardinality-preserving projection, and iterated local search (ILS). On a cubic portfolio benchmark under matched 60-second CPU budgets, HAMD achieves substantially lower decoded native cubic objective values than SA and tabu search, yielding single-seed relative improvements of 87.9%, 71.2%, 59.5%, and 46.9% at n = 200, 300, 500, and 1000. In a detailed three-seed study at n = 200, HAMD attains a median native objective of 195.65 (zero variance), while SA and tabu yield 1208.07. Decoded-feasibility analysis shows SA satisfies all exact cardinality and Rosenberg auxiliary constraints, yet decodes to a native objective 80-88% worse than HAMD, demonstrating a surrogate-distortion effect rather than simple infeasibility. Exact calibration on small instances (n = 20, 25, 30) confirms HAMD finds the provably global optimum in 9/9 trials. These results demonstrate that native higher-order search offers a substantial advantage over quadratized surrogate optimization for constrained cubic portfolio problems.

We study the problem of asset liquidation in financial systems. During financial crises, asset liquidation is often inevitable but can lead to substantial losses if a significant amount of illiquid assets are sold simultaneously at depressed prices -- a phenomenon known as price impact. To tackle this challenge, we consider a two-period liquidation model that allows for early liquidation prior to clearing, thereby mitigating price impact at clearing, and we develop a worst-case approach to solve the decision-making problem on the optimal size of early liquidation. Specifically, we propose a robust optimal strategy -- a tractable liquidation approach that maximizes the worst-case value of liquid assets at clearing, taking into account the uncertainty of other banks' early liquidation decisions. We derive a (semi-)closed-form representation of this strategy in a practical scenario involving permanent price impact and analyze its sensitivity to that impact's magnitude. We further identify its closed-form expression in another practical scenario featuring interbank exposures. Our findings, although built upon a stylized model, offer valuable guidelines for developing robust liquidation strategies that mitigate losses resulting from asset liquidation.

Forecasting crude oil prices remains challenging because market-relevant information is embedded in large volumes of unstructured news and is not fully captured by traditional polarity-based sentiment measures. This paper examines whether multi-dimensional sentiment signals extracted by large language models improve the prediction of weekly WTI crude oil futures returns. Using energy-sector news articles from 2020 to 2025, we construct five sentiment dimensions covering relevance, polarity, intensity, uncertainty, and forwardness based on GPT-4o, Llama 3.2-3b, and two benchmark models, FinBERT and AlphaVantage. We aggregate article-level signals to the weekly level and evaluate their predictive performance in a classification framework. The best results are achieved by combining GPT-4o and FinBERT, suggesting that LLM-based and conventional financial sentiment models provide complementary predictive information. SHAP analysis further shows that intensity- and uncertainty-related features are among the most important predictors, indicating that the predictive value of news sentiment extends beyond simple polarity. Overall, the results suggest that multi-dimensional LLM-based sentiment measures can improve commodity return forecasting and support energy-market risk monitoring.

Using transaction-level trade data from Polymarket's 2024 U.S. presidential election market, we study how prediction markets process shocks. We analyze three events: the Biden-Trump debate, the assassination attempt on Trump, and Biden's dropout. Trading rises after each shock, especially among incumbent traders with pre-event exposure against a Trump victory, who are also more likely to flip positions. Price adjustment differs across shocks. The debate-induced price jump largely reverses, the assassination-attempt repricing persists, and Biden's dropout triggers two-sided trading with little net price change. These patterns link post-news price dynamics to liquidity and disagreement about how shocks map into election odds.

A tight alignment between the degree vector and the leading eigenvector arises naturally in networks with neutral degree mixing and the absence of local structures. Many real-world networks, however, violate both conditions. We derive bounds on the divergence between the degree vector and the eigenvector in networks with degree assortativity and local mesoscopic structures such as communities, core-peripheries, and cycles. Our approach is constructive. We design sufficiently general degree-preserving rewiring algorithms that start from a neutral benchmark and monotonically increase assortativity and the strength of local structures, with each step inducing a perturbation of the adjacency matrix. Using the Stewart--Sun Perturbation Bound, together with explicit spectral-norm control of the rewiring steps, we derive upper bounds on the angle between the eigenvector and the degree vector for modest levels of assortativity and local structures. Our analytical bounds delineate regions of `spectral safety' in which a node's degree can be used as a reliable measure of its systemic importance in real-world networks. We also substantiate our analytical bounds with numerical simulations that compute the exact angles of deviation.

In this paper, a new approach for solving the problems of pricing and hedging derivatives is introduced in a general frictionless market setting. The method is applicable even in cases where an equivalent local martingale measure fails to exist. Our main results include a new superhedging duality for American options when wealth processes can be negative and trading strategies are subject to a cone constraint. This answers one of the questions raised by Fernholz, Karatzas and Kardaras.

We develop a unified framework for modeling multiple term structures arising in financial, insurance, and energy markets, adopting an extended Heath-Jarrow-Morton (HJM) approach under the real-world probability. We study market viability and characterize the set of local martingale deflators. We conduct an analysis of the associated stochastic partial differential equation (SPDE), addressing existence and uniqueness of solutions, invariance properties and existence of affine realizations.

Elicitable functionals and (strictly) consistent scoring functions are of interest due to their utility of determining (uniquely) optimal forecasts, and thus the ability to effectively backtest predictions. However, in practice, assuming that a distribution is correctly specified is too strong a belief to reliably hold. To remediate this, we incorporate a notion of statistical robustness into the framework of elicitable functionals, meaning that our robust functional accounts for "small" misspecifications of a baseline distribution. Specifically, we propose a robustified version of elicitable functionals by using the Kullback-Leibler divergence to quantify potential misspecifications from a baseline distribution. We show that the robust elicitable functionals admit unique solutions lying at the boundary of the uncertainty region, and provide conditions for existence and uniqueness. Since every elicitable functional possesses infinitely many scoring functions, we propose the class of b-homogeneous strictly consistent scoring functions, for which the robust functionals maintain desirable statistical properties. We show the applicability of the robust elicitable functional in several examples: in a reinsurance setting and in robust regression problems.

This paper investigates how natural language communication with an AI agent affects human cooperative behaviour in indefinitely repeated Prisoner's Dilemma games. We conduct a laboratory experiment (n = 126) with two between-subjects treatments varying whether human participants chat with an AI chatbot (GPT-5.2) before every round or only before the first round of each supergame, and benchmark against human-human data from Dvorak and Fehrler (2024) (n = 108). We find four main results. First, cooperation against the AI is high and initially comparable to human-human levels, but unlike in the human-human setting, where cooperation converges to near-complete levels, cooperation against the AI plateaus and never reaches full cooperation. Second, repeated communication, which substantially increases cooperation in human-human interactions, has no detectable effect in the human-AI setting. Third, strategy estimation reveals that human-AI subjects favour Grim Trigger under pre-play communication and remain dispersed under repeated communication, whereas human-human subjects converge to Tit-for-Tat and unconditional cooperation respectively. Fourth, human-AI conversations contain more explicit strategy commitments but fewer emotional and social messages. These results suggest that humans cooperate with AI at high rates but do not develop the trust observed in human-human interactions. Cooperation in the human-AI setting is sustained through conditional rules rather than through the social bonds and mutual understanding that characterise human-human cooperation.

In this paper, we propose a novel frequency-severity joint trip-level risk index that combines the frequency of abnormal driving patterns with a severity component reflecting how extreme such behavior is relative to a portfolio-level baseline. Severity is quantified through an inverse-probability penalty that increases with the rarity of observed tail extremes, rather than being interpreted as a claim size. Based on high-frequency telematics data, we construct a multi-scale representation of longitudinal acceleration using the maximal overlap discrete wavelet transform (MODWT), which preserves localized driving patterns across multiple time scales. To capture severity as tail rarity, we model the portfolio distribution using a Gaussian-Uniform mixture with a layered tail structure, where Gaussian components describe typical driving behavior and the tail is partitioned into ordered severity layers that reflect increasing extremeness. We develop a likelihood-based estimation procedure that makes inference feasible for this mixture model. The resulting severity layers are then used to construct multi-layer tail counts (MLTC) at the trip level, which are modeled within a Poisson-Gamma framework to yield a closed-form posterior risk index that jointly reflects frequency and severity. This conjugate structure naturally supports sequential updating, enabling the construction of dynamically evolving driver-level risk profiles. Using the UAH-DriveSet controlled dataset, we demonstrate that the proposed index enables reliable discrimination across behavioral driving states, identification of high-risk trips, and coherent ranking of drivers, yielding a purely behavior-driven risk measure suitable for actuarial ratemaking and potentially mitigating fairness concerns associated with traditional covariates.

This paper revisits the classic instrument choice problem in a setting with consumption externalities, through the lens of robust mechanism design. A regulator can implement any incentive-compatible policy but is uncertain about how individual demand is correlated with marginal externalities, and evaluates policies by worst-case welfare. The optimal policy is a quantity control: a floor for positive externalities and a ceiling for negative externalities. If the sign of the correlation is known, a uniform tax or subsidy can be optimal. The framework also applies to regulatory uncertainty and costly screening, providing a welfare-based explanation for the prevalence of non-price policies.

Social relationships are known to shape human behavior, yet when and how social ties influence strategic cognition remains unclear. We adopt a dual-measure approach that combines observed gameplay behavior with elicitation of partner-specific beliefs at each decision point, allowing us to examine how social ties shape both decisions and predictions across interaction structures. Dyads classified as having no ties, weak ties, or strong ties played three canonical economic games: the Dictator Game, Ultimatum Game, and Centipede Game, while also making predictions about their partner's actions. Using a mixed design that held partners constant across games while varying social distance between dyads, we examined how relational proximity affected the alignment between behavior and partner-specific beliefs. Across two norm-saturated games (Dictator and Ultimatum), neither offers nor belief calibration differed reliably by social distance. In contrast, in the sequential Centipede Game, where outcomes depend on anticipating a specific partner's future actions, strong-tie dyads both cooperated longer and expected later termination than no-tie dyads, with beliefs and behavior shifting in parallel. These results indicate that social ties become strategically relevant when the interaction structure makes partner-specific accountability cognitively necessary, but not when behavior is governed primarily by shared norms or institutional constraints. The findings provide a structural account of when relational knowledge enters strategic cognition and help reconcile mixed results in prior work on social distance in economic games.

This paper makes the Millennium Prize problem P vs NP operational in quantitative finance by studying cardinality-constrained portfolio selection. Starting from the convex Markowitz mean-variance program with CAPM-based expected returns (Rf plus beta times ERP), we impose a hard sparsity rule that limits the portfolio to K assets out of approximately 94 industry portfolios (Damodaran). The constraint couples discrete subset selection with continuous weight optimization, yielding a mixed-integer quadratic program and an NP-hard search space that grows combinatorially with n and K. We therefore evaluate scalable approximation schemes (greedy screening, Monte Carlo sampling, and genetic algorithms) under a replication-oriented protocol with random-seed control, distributional performance summaries (median and quantiles), runtime profiling, and convergence diagnostics. Dependence structure is documented via correlation and covariance diagnostics and positive-semidefinite checks to link algorithm behavior to the geometry implied by the risk matrix. To support the title's derivatives component, we add a European call option priced by the Black-Scholes model and map it into CAPM-consistent moments using delta-based linearization, validated with a bump test and moneyness/maturity sensitivity. Results highlight how the cardinality constraint reshapes the attainable efficient frontier, why stability and computational-cost trade-offs matter more than single-best runs, and how common-factor dependence can limit diversification in K-sparse solutions. The study provides a reproducible template for NP-hard portfolio optimization with transparent inputs and extensible derivative overlays.