Papers

In this paper we investigate the information content of the lower part of the spectrum of financial correlation matrices, as a source of information on market synchronization. In a financial context, a classical application of Principal Component Analysis and Random Matrix Theory identifies the largest eigenvalues as indicators of dominant market factors and synchronization patterns. We complement this perspective by showing that the smallest eigenvalues also contain relevant information about the effective structure of financial markets. The paper presents the methodological proposal and validates its effectiveness through comprehensive real data experiments in both descriptive and predictive settings.

Cryptocurrency exchange-traded products (ETPs) listed on European exchanges provide a regulated environment for studying intraday market anomalies. We study four Bitcoin and Ethereum ETPs traded on Xetra and Nasdaq Stockholm over the period January 2024 - December 2025 using one-minute bars. As a benchmark, we adopt an extreme value theory approach in which anomalous bars are defined as returns falling below a threshold estimated by fitting a generalised Pareto distribution to left-tail exceedances. We then propose three new binary anomaly indicators. The first, a cross-venue divergence anomaly, identifies venue-specific price divergence between the two exchanges. The second is a no-recovery anomaly that identifies extreme price drops followed by little or no recovery over the next ten active bars. The third is a momentum-reversal anomaly that identifies extreme price drops following positive short-term momentum. Although each anomaly type represents fewer than 1% of one-minute bars, statistical analysis using Mann-Whitney U tests shows that anomaly observations exhibit significantly higher effective spreads, higher values of liquidity-related ratios, and more pronounced order-flow imbalances than non-anomalous bars. Furthermore, employing an out-of-sample prediction methodology with four classifiers - random forest, logistic regression, extreme gradient boosting, and light gradient boosting machine - shows that all four anomaly types are predictable one bar ahead, with AUC-ROC values of up to 0.82. Permutation importance indicates that short-term volatility and drawdown measures are generally more useful for prediction than microstructure variables.

Climate perils are linked through event ordering and state-dependent propagation, features not fully captured by joint loss distributions alone. This paper develops a Cascading Climate Risk Network (CCRN) for multi-peril reinsurance that separates calendar-scale climate conditioning from within-event propagation on a directed acyclic graph (DAG). The model combines complementary-log-log triggering hazards with bounded severity activation, mapping physical states to insured losses via a capacity-bounded demand-surge transformation. For fixed shocks, the event-scale cascade reaches a unique finite-step closure. Monotone comparative statics provide a pathwise upper-corner loss bound over rectangular stress sets, yielding a transparent contract-level stress-testing guarantee under common aleatory inputs. Comprehensive numerical experiments, including copula and Bayesian-network benchmarks, sensitivity analyses, and uncertainty propagation, demonstrate that while central layer prices remain robust across matched-marginal dependence structures, far-tail and high-layer behaviors differ materially. Directional propagation, annual event frequency, and dependence strength emerge as the principal risk drivers. The study provides a controlled synthetic verification of the proposed architecture.

Stablecoins have rapidly emerged as an important class of digital assets and a component of the digital financial ecosystem. Despite their growing importance, the statistical properties of stablecoin transaction activity remain largely unexplored. To the best of our knowledge, this is the first study to investigate scaling behavior in stablecoin transaction data, focusing on USDT and USDC. We analyze approximately 370 million USDT and USDC transactions recorded on the Ethereum blockchain across six periods spanning June 2024 to February 2026. Based on interactions between Externally Owned Accounts (EOAs) and Smart Contracts (SCs), we classify transactions into four categories: EOA-EOA, EOA-SC, SC-EOA, and SC-SC. Using maximum-likelihood estimation of power-law exponents, we find that transaction value distributions exhibit heavy-tailed scaling for both stablecoins across all periods and interaction categories. We identify two distinct scaling regimes: EOA-involved categories cluster around 1.45-1.60, whereas SC-SC transactions exhibit higher exponents of approximately 1.72-1.73. Sensitivity analysis confirms that this separation is robust across periods, stablecoins, and fitting sample sizes. Counterfactual analysis shows that changes in category weights alone cannot explain the observed variation in the overall exponent. Across different sample sizes, the counterfactual path accounts for only about 10%-35% of the total temporal range observed in the actual data. Overall, our results indicate two broadly differentiated scaling regimes in the tail of stablecoin transaction values. Power-law tail behavior is observed throughout stablecoin transaction activity, but the exponent depends on whether transactions are driven by EOAs or SCs. These findings provide a basis for further research on scaling behavior and transaction heterogeneity in blockchain-based financial systems.

Crypto-listed equity perpetuals trade while the primary cash market is closed, yet still need a mark for margin, funding, and liquidation. We model the closed-window mark as the fixed point of an oracle operator with two blocks: external anchoring and self/peer derivative reference. From marks and proxies alone the two are observationally equivalent: every reduced form admits infinitely many topology decompositions, and a path-law argument extends this to the full mark dynamics, so lead-lag and information-share estimators have power equal to size. Disclosure breaks the tie – disclosed diagonal adjustment identifies the normalized topology, and disclosed support with forbidden anchors gives a row-level test that identifies, falsifies, or leaves a positive-dimensional class under a rank condition and finite-sample tolerance. Empirically, a disclosed OKX row survives pre-open falsification while a pure-external baseline shows the test's limited power, and an eight-week deep-closed panel with cash-reopen validation bounds the live-external content of closure variance. Cross-venue agreement is not price discovery unless disclosure or the cash reopen breaks the equivalence class.

We develop a rigorous mathematical framework for the governance of systems of K self-adapting generative AI models under the principles of Model Risk Management (MRM). When multiple models share a meta-learning coupling through an interaction matrix, the per-agent Lyapunov analysis that underpins standard MRM is provably insufficient: individual agents can each satisfy their declared stability bounds while the joint system is in a regime of emergent ensemble-level drift. We formalize this gap through the Joint Lyapunov Proof (JLP)—a cryptographic and stochastic protocol that attests, without revealing proprietary weights, that the aggregate dynamics satisfy MRM Ongoing Monitoring standard at every validation epoch. Our main contributions are the following. We give a complete characterization of the infinitesimal generator of the joint quadratic Lyapunov function. We derive the exact critical coupling threshold above which the system loses mean-square stability. We prove a Noise-Floor Theorem and identify the correct target for zero-knowledge attestation. A per-epoch Succinct Non-Interactive Argument of Knowledge (SNARK) on the live weights is derived. All theoretical claims are validated against five numerical studies using a multi-agent softmax system.

Model risk management (MRM) guidance assumes a static model lifecycle, in which models are developed, independently validated, and implemented without further autonomous modification. Continually self-adapting generative AI systems — models that update their own weights during production deployment — fundamentally violate this assumption and render point-in-time validation inadequate. This paper addresses the resulting governance problem in two parts. Part I develops a rigorous telemetry architecture for such models, operating simultaneously in discrete and continuous time. We establish a Minimal Sufficient Statistic for audit purposes, construct a tamper-evident Merkle chain for discrete weight sequences, derive the appropriate continuous-time generalization via the Ito formula, and propose event-driven logging via KL divergence stopping times that is both computationally tractable and meaningful for validation. Part II asks what happens when the model provider is adversarial. A firm deploying such a model has strong incentives to conceal learning updates that would trigger mandatory validation review. We formalize this as the Model Hiding Problem and provide a systematic taxonomy of six distinct attack strategies against the Part I architecture, spanning discrete and continuous time, with a formal countermeasure for each. Together the two parts establish a dual-regime architecture in which continuous telemetry is necessary but not sufficient, narrowing—but never replacing—periodic invasive audit. The framework is model-architecture-agnostic and is designed to satisfy the three pillars of traditional MRM.

We examine the effect of political power-sharing on local economic activity. This effect depends on the relative importance of the risks associated with unchecked power and the potential efficiency gains or losses arising from checks and balances. Our research design exploits a geographic discontinuity design due to the haphazard overlap of electoral and administrative boundaries that generates quasi-random variation in the number of politicians governing adjacent regions. We supplement this design using an episode of electoral delimitation that allows us to exploit within-region variation in the number of politicians. We find increasing the number of politicians governing an area can lead to new firm creation, lower unemployment, and greater real economic activity. Our results suggest that non-aligned multiple politicians enhance state efficiency by imposing checks and balances on each other, leading to lower regulatory obstacles, less cronyism, and improved provision of public infrastructure, creating an economically favorable environment for firm creation.

Completely monotonic inverse marginal (CMIM) utilities, introduced in [MSZ24], constitute a tractable class of preferences that includes many of the most important utility functions used in mathematical finance, such as power and exponential utilities. In stochastically dominant markets, their Bernstein representation induces a hidden linear structure in the dual optimization problem that serves as the foundation for the present analysis. In this paper, we investigate the sensitivity of optimal investment with respect to perturbations of investor preferences within the CMIM class. Exploiting Bernstein's representation theorem, we show that, under stochastic dominance, affine perturbations of Bernstein measures induce an affine representation of the dual value function. As a result, the dependence of the optimization problem on preferences can be analyzed through a scalar budget equation, allowing us to prove analyticity of the associated Lagrange multiplier with respect to the perturbation parameter and to derive convergent analytic expansions of arbitrary order for the primal value function and the optimal terminal wealth, with explicit recursive formulas expressed through Bell polynomials.

The near-term growth of the commercial space economy and sustainability of the low-Earth orbit (LEO) environment depends on the commercial prospects of large LEO satellite telecommunications constellations (“mega-constellations”). If successful, mega-constellations can spur competition between launch providers, which may lead to lower launch prices. They may also demand orbital sustainability services and/or generate additional collision risk, which may in turn affect space sustainability interests. Consumer demand for mega-constellations' internet services is critical to their commercial success. This analysis presents a method to estimate consumer demand for mega-constellations' internet services, using SpaceX's Starlink over December 2021–November 2023 as an example. This analysis finds that Starlink users are highly concentrated in the wealthiest countries. It also finds that consumer demand for Starlink is growing more slowly than orbital capacity is being added and more slowly than demand for non-Starlink internet overall. Continuation of these trends may indicate that the market for mega-constellation internet is less lucrative than previous forecasts have indicated.

Foundation models for time series forecasting demonstrate impressive zero-shot generalization but often underperform on specialized domains such as high-frequency finance. We present a comprehensive study of hybrid neural-classical correction for adapting frozen TimesFM (200M parameters) to stock return prediction during the volatile opening trading hour. We compare two neural correction architectures - AttnCorrect (multi-head self-attention, approximately 471K parameters) and GatedLinear (low-rank bilinear projection with gating, approximately 49K parameters) - each augmented with Random Forest residual learning. Through systematic ablation across 10 major technology stocks (NVDA, MSFT, AAPL, GOOG, GOOGL, AMZN, META, AVGO, TSLA, NFLX) spanning 2 million data points, we reveal critical insights: (1) The hybrid neural-classical approach achieves 0.597 pooled correlation and 6.4x mean per-day correlation improvement over frozen TimesFM; (2) Classical residual learning (Random Forest) provides the largest single-component contribution, matching or exceeding the neural correction component; (3) Simpler neural architectures surprisingly outperform complex ones when classical residual learning is removed; (4) Self-attention provides the largest neural-only contribution. GatedLinear+RF achieves best overall performance with 9x fewer neural parameters than AttnCorrect+RF. We report three complementary correlation metrics - mean per-day, cross-day cumulative, and pooled - to provide a complete picture of predictive quality. Our results provide practical guidance: effective foundation model adaptation requires careful integration of neural and classical components, with classical methods playing a crucial complementary role.

Section 1502 of the Dodd–Frank Act, enacted in 2010, requires U.S.-listed companies using tin, tantalum, tungsten, and gold (3TG) from the Democratic Republic of the Congo and adjoining countries to disclose information on the minerals' origins. Concerns have been raised that the regulation may have induced a de facto embargo through avoidance of sourcing from the covered region. However, how the price responsiveness of mineral exports evolved under changing institutional and market conditions remains insufficiently understood. Since tungsten production in the covered region is concentrated almost entirely in Rwanda, this study examines the price responsiveness of Rwandan tungsten exports from January 2009 to December 2023. Because missing export quantity data prevent continuous observation of export unit values, we apply the identification approach of Nakano and Nishimura (2025), combining monthly mirror trade data from UN Comtrade with exchange rates and a world average price. An importer fixed-effects model is estimated using export value as the dependent variable, with the sample divided into four periods according to changes in the institutional and market environment. The results reveal substantial temporal variation in price responsiveness. A statistically significant negative price response is observed in Period 1 ($\eta=-20.814$, $p<0.01$), disappears in Period 2 ($\eta=1.814$, $p>0.10$), reappears in Period 3 ($\eta=-5.277$, $p<0.01$), and disappears again in Period 4 ($\eta=0.440$, $p>0.10$). Coefficient-difference tests confirm significant changes between Periods 1 and 2 ($p=0.0021$) and between Periods 3 and 4 ($p=0.0006$). These findings suggest that the price responsiveness of Rwandan tungsten exports varied substantially over time rather than following a uniform trajectory after the regulation.

Open-weight language models from Chinese AI labs caught up on benchmarks relative to proprietary frontier models in recent months. Yet their reliability on real-world financial tasks remains largely untested. We updated the Financial Touchstone benchmark, which now has 2,967 question context-answer triplets across 495 international annual reports. We also apply a new set of models on the benchmark, expanding coverage from eleven to twenty models across ten providers, including recent open-weight models such as GLM 4.7, GLM 5, Kimi K2.6, and DeepSeek V3.2, as well as Alibaba's proprietary flagship Qwen3-Max. Anthropic's Claude Opus 4.6 achieves the highest accuracy (88.4%), while Google's Gemini 2.5 Pro maintains the lowest hallucination rate (0.08%). Notably, the open-weight Kimi K2.6 ranks third in accuracy, and the non-reasoning models GLM 5 and Mistral 3 rank fourth and fifth, challenging the assumption that reasoning architectures or proprietary weights are a prerequisite for strong financial comprehension. Information retrieval remains the primary bottleneck, accounting for 48.9% of all failures. We also document a new finding: geopolitical content filters in Chinese models refuse legitimate financial questions (0.08% of attempts), sometimes without clear reason, and the refusal behavior depends on the access route as much as on the model. The complete dataset and evaluation framework are publicly available.

The daily return of a stock is often restricted to an exchange-imposed band to curb extreme fluctuations. Any attempted price movement beyond this band is clipped, leaving an unobserved excess. We introduce a minimal stochastic latent-state model in which a fraction of this hidden excess is retained for the next day. This retention generates memory, even though the daily stochastic driving shocks are independent. For symmetric driving shocks with regularly varying tails, the stationary latent return preserves the tail index of the noise, but has an enhanced tail amplitude. In the wide-band limit, a close of the daily return at either limit of the band admits a single-dominant-shock description. We show that after such an event, the mean return on the following day has the same sign and grows proportionally to the band width, while the probability of reaching the same limit again approaches a finite value. Reaching the opposite band limit on the following day requires a second extreme shock of opposite sign and is power-law suppressed. Simulations support these analytical predictions. Empirical data from stocks subject to daily price limits are qualitatively consistent with the predicted same-sign response and its increase across wider price bands.

Artificial intelligence is associated with larger research teams, yet in mathematics, among the most codifiable fields, individual researchers working with AI now produce research-grade results. A span-of-control model reconciles these observations. AI lowers execution cost, which expands laboratory scale, and automates codifiable tasks, which lowers the member share of each unit. Team size is therefore quasi-concave in AI capability, with at most one peak. The model predicts that a fully codifiable team peaks when effective automation coverage reaches a closed-form threshold, typically near complete coverage, and, among fields with shared primitives that possess an interior peak, those with less irreducibly human task content peak first. Under explicit priors, the 90 percent forecast intervals for the fully codifiable peak span 2026 to 2030.

How much capital a trading strategy can absorb before its edge disappears is a causal question about how much is deployed, but it is answered with observational proxies that rest on incompatible assumptions. We ask what experiment would answer it instead, and show that two features of the problem interact to constrain any answer. Deployed capital erodes the edge gradually, so a trial of fixed length measures less than the eventual effect; and parallel implementations of one strategy trade the same securities, so they are not independent units. Comparing implementations on the same date removes market-wide shocks, which is what makes the comparison credible. But the crowding created by the strategy's own accumulated position is common to those implementations too, and an arbitrary date effect absorbs it exactly: the comparison that makes the experiment robust is the one that prevents it from measuring the crowding capacity is about. A same-date design recovers one implementation's private response at the prevailing level of aggregate positioning, and reaching the aggregate effect requires either implementations with deliberately different exposure to that position or variation in it over time. We characterise what each route identifies and what it costs, establish how far a fixed holding period understates the eventual effect and how to correct for it, and show what a finite set of deployment levels can and cannot reveal. A calibration on a purpose-built panel illustrates the resulting design rules and prices a study that would follow them.

In this paper, we consider an optimal stopping problem with infinite horizon, non-negative pay-offs and non-linear evaluations $\rho_{S,\tau}$ indexed by two indices: $S$ and $\tau$, where $S$ is the time of evaluation and $\tau$ is the time when the pay-off is revealed. The agent's stopping strategies are constrained to be in the set of so-called Bermudan stopping times $\Theta$. Under suitable assumptions on the non-linear evaluations $\rho$ and on the pay-off, we show that a dynamic programming principle holds in this framework. We investigate the existence of $\varepsilon$-optimal stopping times, as well as the existence of optimal stopping times. We show that an $\varepsilon$-optimal stopping time exists. We also prove that the first time when the value family hits the pay-off is optimal if and only if it is finite. We also provide Doob's type convergence for non-negative \emph{$(\Theta, \rho)$}-supermartingales in the case where $\rho_{S,\tau}=\rho_S$ depends on the first index only. We provide an example from BSDEs with infinite horizon.

Business users confronted with a moving metric need to know which part of their data moved and why. Existing data-explanation methods typically return predicates: conjunctions of attribute-value conditions that isolate responsible records. Predicates are exact and directly executable as filters, but they describe axis-aligned regions and may not compactly capture segments defined by combinations of continuous tendencies. This paper presents Self-Explaining Segment Trees (SEST), an architecture in which an explanation is a multivariate cluster in a feature subspace selected for relevance to a designated key performance indicator (KPI). SEST selects the subspace once per KPI using Shapley attributions over a decision-tree surrogate, recursively partitions the population while choosing the branching factor independently at each node through mixture-model silhouette search, and attaches to every node a dual explanation payload: standardized effect sizes over numeric features and type-dependent contribution profiles over user-designated dimensions. These explanations are computed from untransformed data so surfaced values retain source units and category labels. A stance layer reduces any depth of the tree to its extremal KPI-suppressing and KPI-amplifying segments. We establish termination and a node-count bound determined by the depth limit and minimum segment size, and characterize per-tree construction cost as quadratic in population size in the degenerate case and geometrically decaying across depth in the balanced case. This is an architecture and methodology paper; we report no predictive-accuracy or validation results and leave outcome validation to future work.

The paper places the Samuelson–Merton “util-prob” construction and preference-based state-price interpretation inside a continuous-time equivalent-martingale-measure framework. The contribution is not the existence of a marginal-utility pricing kernel, but a compensated stochastic-exponential representation of the “util-prob” density. Behavioral weighting by marginal utility becomes an equivalent martingale measure only after the drift generated by risk aversion and prudence is compensated. Under positivity, integrability, normalization, and martingale conditions, the behavioral density is generated by the Brownian kernel $-A_t\sigma_t$, where $A_t$ is Arrow–Pratt risk attitude and $\sigma_t$ is state volatility. The required compensator identifies the finite-variation term that must be removed before marginal-utility weighting defines a true martingale density. The framework preserves put–call parity under a common admissible kernel and yields a Cox–Ross risk-premium identity linking option and underlying premia through volatility exposures. Empirically, the paper maps option-implied and physical densities into a signed Arrow–Pratt decomposition, with positive curvature interpreted as risk aversion and negative curvature as risk-seeking or speculative tilt. In a panel of liquid index ETF and single-name options, arbitrage-penalized EBMM has lower pooled holdout squared-error loss than arbitrage-penalized SVI when both models receive the same total-variance positivity and butterfly-density penalties, although slice-level results remain mixed. The exercise is best read as a density-ratio admissibility test rather than a replacement for production volatility-surface estimation.

We introduce StableEval Arena, a cost-aware benchmark framework for evaluating agentic AI systems on stablecoin peg-risk prediction. StableEval Arena evaluates LLM-backed agentic systems on diagnosing peg stress and forecasting deviations from the one-dollar peg over a hidden seven-day horizon, using leakage-safe historical replay with exchange price-volume data and market-context features. We report two complementary experiment blocks: a 120-case stress-enriched validation block and a 507-case natural-distribution full-arena evaluation block. Across six LLM-backed agent configurations and baselines, StableEval Arena measures prediction quality, calibrated-label behavior, structured-output reliability, latency, token consumption, and estimated inference cost. Rather than ranking agents by accuracy alone, the framework treats trustworthiness as a joint property of forecast quality, operational reliability, and computational cost. The results show a gap between protocol-following reliability and financial-risk reliability: agents reliably produce valid structured outputs at modest measured cost, but still miss most rare severe-stress and sustained-depeg cases. To support auditing and replication, we release the benchmark dataset on Hugging Face and the source code on GitHub.

We provide a first causal analysis of the behavioral consequences of the friendship paradox-the fact that people's friends in a network have more connections than average. We find that people's behavior is biased by their network position: they do not best respond to what they should infer the average behavior of the population to be, but instead simply to the average behavior of their friends. Moreover, we find that they fail to learn to overcome such a bias when relocated within the network, varying their observational environment. In these games of complements, the friendship paradox generates a systematic upward distortion in actions, increases behavioral dispersion, and persists despite learning opportunities.

Hawkes-based microstructural foundations for rough volatility, leverage, and rough Heston-type limits were developed by El Euch et al. (2018, Finance Stoch., 22(2), 241–280) and connected to the affine rough Heston framework of El Euch and Rosenbaum (2019, Math. Finance, 29(1), 3–38). The rough Hawkes–Heston model with common price–volatility jumps of Bondi et al. (2024, Math. Finance, 34(4), 1197–1241) extends this framework by adding state-dependent common jumps to rough affine volatility. We provide a microstructural foundation for its variance and common-jump mechanism by constructing a Poisson-embedded marked Hawkes order-flow model. Ordinary arrivals generate rough continuous volatility and leverage through a nearly unstable heavy-tailed Hawkes mechanism, while rare marked arrivals represent common shock events that produce simultaneous price jumps and volatility excitation. Under the nearly unstable scaling and the reduced-form admissibility conditions, the complete rescaled price/variance/jump system converges along the full sequence to the unique complete canonical rough Hawkes–Heston weak solution. The Hawkes renewal structure yields a Mittag–Leffler Volterra representation, which is then rewritten in Riemann–Liouville fractional form. The limiting coefficients are expressed explicitly in terms of the microscopic parameters. The construction provides a microstructural foundation for the variance and common-jump mechanism of the rough Hawkes–Heston model. Numerical experiments illustrate the convergence of our microstructural foundation to the rough Hawkes-Heston model.

We consider a market maker who can only obtain and dispose of inventory by responding to a sequence of sealed-bid enquiries, and whose customers arrive with imbalanced intent: sellers more often than buyers, or the reverse. Under the assumption that the best competing response is exponentially distributed around a commonly discerned fair price, we observe a symmetry in the steady state solution that compresses the imbalanced problem onto the perfectly balanced one. Order imbalance is absorbed, exactly, by a translation of the market maker's skew, a widening of her quotes, and a multiplication of her effective cost of carry. The adjustment is simple even though the solution it adjusts is not, and it involves no free parameter beyond the observable market width. The exponential assumption is needed only locally, at the quotes actually made, and the width that enters is the locally observed one. Among the consequences: a market maker with zero inventory should still skew; skew responds to imbalance at first order whereas width responds only at second order; and the popular "constant width, linear skew" heuristic is recovered as the small-skew solution in the special case of balanced flow and quadratic holding cost.

Financial question answering is typically evaluated by answer correctness, yet in SEC filings a plausible and even numerically correct answer can be grounded in the wrong evidence. Similar facts and disclosures recur across sections of a filing, across reporting periods of the same firm, and across comparable firms. FinRank targets this provenance-sensitive retrieval problem by requiring systems to identify evidence for the intended entity, reporting period, and disclosure context. The benchmark contains 1185 manually authored question-answer records over the 10-K and 10-Q filings of 22 companies. Each record includes a reference answer, gold supporting passages, and hand-curated hard negatives drawn from confusable passages within filings, across reporting periods, and across comparable firms. FinRank evaluates passage retrieval, reranking, and hard-negative discrimination as separately measured tasks. Baseline results demonstrate the difficulty of this setting: among the evaluated systems, even a 7B instruction-tuned embedder reaches only 44.8% Recall@10 on the pooled evidence corpus; sub-billion-parameter encoders gain at most 3.5 points over BM25, a finance-adapted embedder trails BM25 by 9.7 points, and pairwise accuracy falls by 13.0-20.5 percentage points when random negatives are replaced with the curated hard negatives. FinRank provides an evidence-first benchmark for developing financial question answering systems that are not only accurate but also grounded in the correct disclosure.

This paper documents an applied natural-language-processing framework for measuring the tone of Brazilian Monetary Policy Committee (Copom) statements. The project is explicitly inspired by iSent, Itaú's Central Bank sentiment classifier, particularly its sentence-level division of official communication into hawkish, dovish, neutral, and out-of-context classes. The implementation extends that idea in three directions. First, an LLM identifies short hawkish and dovish expressions and assigns each a 0-to-1 intensity weight. Second, the document index combines sentence counts with document-specific average signal intensities, producing a bounded score from -1 to 1. Third, a separate full-document layer measures forward-guidance direction, guidance explicitness, uncertainty level, and change in uncertainty. The empirical sample is restricted to communications dated August 2016 or later and contains 80 statements and 1,498 classified sentences from August 31, 2016 through August 5, 2026. Across this sample, 33.3% of sentences are hawkish, 18.0% dovish, 42.1% neutral, and 6.5% out of context. The average document score is +0.107, while the most hawkish reading is +0.570 in August 2021. The latest statement, dated August 5, 2026, scores +0.232, with eight hawkish, two dovish, and nine neutral sentences. Its structural overlay is more nuanced: guidance is directionally ambiguous but partly explicit, while uncertainty is classified as central and higher than at the prior meeting. Tone and the guidance-direction score have a contemporaneous Pearson correlation of 0.719. These are descriptive outputs, not a validated forecast of Selic decisions or DI returns. The main contribution is therefore methodological: a transparent, incremental, auditable system that separates rhetorical tone from policy guidance and uncertainty.

Recent work shows that Large Language Models (LLMs) encode more information in their activations than they show in their responses. We ask whether the information contained in activations could be used as a source of measurement for the presence of concepts. We compare two algorithms for extraction of concepts, i.e., Recursive Feature Machine (RFM) algorithm and linear probing against an embedding baseline, surface baselines, and the model's own response using a human-annotated Environmental, Social and Governance (ESG) dataset of financial texts. The best linear probe ranges from 0.6 to 3.3 percentage points below fine-tuned domain classifiers. The probes also outscore the model's own response in all comparisons. Furthermore, the probes outperform the RFM concept vectors in all accuracy comparisons. In a synthetic mixture test, their mean-pooled score tracks the number of concept sentences in a paragraph better than the concept-vector score. Put together, a linear probe on a frozen LLM is a cost-effective and generalizable alternative to fine-tuning a domain classifier.

We study Lambda-quantiles, a generalisation of classical quantiles in which the constant probability level $\lambda \in [0,1]$ is replaced by a functional parameter $\Lambda \colon \mathbb{R} \to [0,1]$. We consider the general case of non-monotone $\Lambda$, which arises naturally if closure properties of the class of corresponding Lambda-quantiles with respect to inf-aggregation or with respect to mixtures are required. As preliminary results, we characterise finiteness, constancy, and what we call the attainment property known from classical quantiles. We then consider the problem of reconstructing $\Lambda$ from the values of $\Lambda$-quantiles on a suitable family of simple distributions, showing its identifiability under mild assumptions. Next, we substantially refine several results obtained in the literature on weak upper and lower semicontinuity and on the property of convexity of the level sets with respect to mixtures, obtaining in both cases almost complete characterisations without any monotonicity assumption. We then move to the case in which $\Lambda$ has bounded variation, which enables us to prove a mixture representation result: any such $\Lambda$-quantile can be rewritten as a Lambda-quantile with an increasing functional parameter, evaluated at a mixture of the original distribution with a fixed reference distribution at a fixed weight, thus reducing the complexity of the parameter from bounded variation to monotone. Finally, we introduce and study the notion of the ordinal covariance group of a risk measure, showing that in the case of a $\Lambda$-quantile it coincides with the compositional invariance group of $\Lambda$ and with a certain group of measure-preserving transformations of the signed measure associated with $\Lambda$.

We develop a certified, scalable approximation for high-dimensional Wasserstein distributionally robust portfolio optimization. For expected-utility maximization under order-one Wasserstein ambiguity, standard duality yields a semi-infinite convex program. For long-only portfolios with box support under the one-norm ground metric, an exact sample-specific vertex reformulation provides an exponential-size computational benchmark. We then majorize the utility by supporting hyperplanes and dualize the support subproblems, obtaining a finite hyperplane–dual formulation over compact polyhedral supports. Under the one-norm ground metric and polyhedral portfolio constraints, this formulation is a polynomial-size linear program. The uniform utility-approximation error bounds both the robust-value error and the near-optimality gap for the original robust problem. Experiments validate the certified approximation and demonstrate monthly 476-asset rebalancing and computational scalability to 1,000 assets.

This paper extends prior work on linguistic pattern based facility layout optimization by enhancing the LP Alinks framework with an explicit spatial uniformity criterion. While earlier studies demonstrated that linguistic patterns can effectively encode expert knowledge and guide agent based layout emergence, their optimization scope remained limited to cost oriented objectives. To address this gap, we introduce the Normalized Coverage Score (NCS), a scale adjusted measure of spatial evenness that complements the classical flow distance economic objective and enables systematic exploration of cost uniformity trade offs. A full factorial experiment comprising 324 conditions evaluates the combined effects of problem size, link density, virtual force scaling, and two families of membership functions on both objectives. Five way and nested three way ANOVAs reveal that structural factors (number of objects and link density) exert a dominant baseline influence on both economic performance and spatial uniformity. However, while economic cost is overwhelmingly governed by these structural characteristics, uniformity is substantially more sensitive to the linguistic pattern parameters, which control the dispersion compaction dynamics of the emerging layouts within the structural constraints. Based on these interactions, we derive a parameter-selection matrix that prescribes settings for cost minimization, uniformity maximization, or balanced compromise. Comparative analyses with Drezner's method, MDS, and non metric MDS demonstrate that the extended LP Alinks framework consistently attains superior uniformity while maintaining competitive economic performance, making it a robust and interpretable decision support tool for early stage facility layout design.

Economics uses different behavioural models for risk, time, losses, valuation, and social choice. I study a unified choice experiment in which the same decision makers face all these domains. I hide a decision maker's choices in one domain and ask a frozen tabular foundation model to recover them from that decision maker's choices elsewhere and labelled choices by other participants. The foundation model improves on the training-sample median, and the gain disappears when visible choices are shuffled across decision makers. I then estimate one random-utility model over the foundation model's learned representation. This structural model applies the same utility function in every domain, retains most of the foundation model's reduction in prediction error, predicts domains excluded from utility estimation, and reproduces how behavioural measures co-move across people. The resulting model separates three objects: a learned common choice domain, one systematic utility function on that domain, and one random component that generates stochastic choice on observed menus.