This paper develops fixed-count bootstrap methods for likelihood-based inference in autoregressive conditional duration (ACD) models with random calendar-time spans, addressing finite-mean, boundary, and infinite-mean regimes and the resulting non-standard limit behavior. It shows fixed-count bootstrap is valid for κ ≥ 1 and yields a Gaussian bootstrap t statistic for all κ > 0, while percentile bootstrap remains first-order valid under 0 < κ < 1. The results are supported by Monte Carlo simulations and an empirical application to cryptocurrency ETFs. [page::1]
FinSMART introduces a market-aligned reinforcement learning framework for financial sentiment analysis that directly optimizes sentiment signals using realized market outcomes, enabling continual adaptation through market-aware retraining [page::0][page::1]. The approach uses a dual-filter reward with asymmetric incentives and a GRPO optimization to align sentiment with profitable market activity, achieving substantial improvements in profitability, risk-adjusted performance, and signal quality compared with state-of-the-art methods (e.g., FinDPO), including a 220% increase in cumulative trading returns over the strongest baseline and improved RankIC [page::0][page::1][page::2]. The framework further demonstrates the practicality of market-aware retraining—replacing costly human labels with observed articles and outcomes to sustain performance gains under evolving market dynamics [page::6][page::7].
This paper studies an electoral competition model where parties maximize legislative power via the Myerson value rather than vote shares. Voters are uniformly distributed on [0,1], and each voter supports the closest policy. A coalition-communication graph emerges: two parties link if their policy distance is at most d, and a party’s payoff equals its Myerson value in the graph-restricted voting game. The analysis derives symmetric pure-strategy equilibria for n = 2, 3, 4. In the two-party case, the unique equilibrium converges to the median (unaffected by d). In the three-party case, a continuum of equilibria exists as long as the two extremes communicate; as d → 0, the median voter theorem holds. In the four-party case, the unique symmetric equilibrium places two parties at (1−d)/2 and two at (1+d)/2; as d → 0, convergence to the median occurs. Across 3- and 4-party setups, tighter coalition communication (smaller d) exerts a centripetal force toward the center. An application with two extreme parties shows endogenous cordon sanitaire behavior and again median convergence as d → 0. These results highlight the role of coalition-power incentives in multiparty policy competition and the tractability of the Myerson value as a bargaining power measure [page::0][page::8][page::9][page::10]
We derive a mesoscopic model of optimal execution with passive limit orders, grounding it in microstructural observations of order-flow imbalance and limit-order fill probabilities [page::1]. The model yields a reduced-form passive impact rate that decays exponentially with quote distance and links directly to fill intensity [page::2]. An optimal liquidation problem is solved where the trader selects the distance of passive sell quotes, trading off higher fill probability against larger cumulative impact and potential non-execution risk [page::3]. Empirical calibration using NASDAQ equities and public FX data supports the core microstructural assumptions and highlights extensions to heterogeneous decay rates, transient impact, and target execution schedules [page::4].
This paper reports a large natural field experiment (70,884 applications) showing that AI voice agents interviewing applicants can reduce variance in information collection and improve hiring outcomes: offers rise by 12% (9.73% vs 8.70%), job starts rise by 18%, and one-month retention rises by 18%, with no observed decline in the productivity of hired workers [page::2][page::15][page::18]. The mechanism—controlled variance—stems from AI enforcing structured interviews while remaining responsive to applicants, yielding richer, more comparable hiring-relevant information [page::3][page::4]. Transcript analyses show that AI-led interviews standardize topic coverage and guideline adherence, and correlate with linguistic features linked to higher offer probabilities [page::3][page::20]. Applicants generally respond well to AI interviewing, with most in the Choice condition preferring AI, though some backlashes (e.g., AI aversion, technical failures) exist; cost analyses indicate AI becomes cost-effective in higher-wage or high-volume settings under favorable pricing, though break-even depends on price tiers and wage environments [page::28][page::38][page::103]. Overall, AI can improve screening efficiency in high-volume, routine tasks, while human evaluators continue to play a role in signal weighting and complex assessments [page::40].
FinanceHarness introduces a layered, expert-guided harness and a point-in-time financial deep research benchmark (FinanceGym) to address the gap between generic deep-research systems and analyst-style financial reports. The PIT sandbox enables per-question cutoffs and forward-looking evaluation, while FinanceGym provides 400 expert-annotated questions with 2,464 rubric items across topics, sectors, and reasoning types, split into pre-cutoff evidence retrieval and post-cutoff outcome anticipation [page::2][page::4][page::5]. The framework demonstrates that even strong open-weight models struggle to bridge retrieval with forward-looking synthesis, with FinanceHarness achieving a 32.4% overall rubric score on a 27B backbone—close to larger backbones and well above baseline open-weight systems—and GRPO trainer adds a further ~0.4 points, underscoring the value of a shared, temporally controlled research environment [page::12][page::9][page::11]. The paper details the architecture (orchestration, tools, runtime, and skills), dataset pipeline, evaluation contract, and illustrative case studies indicating the harness's capacity to produce citation-grounded analyst-style reports within a reproducible PIT framework [page::6][page::7][page::8].
AWARE-FX is an auditable AI/NLP decision-support system that converts corporate annual-report text into traceable firm-year FX hedging-disclosure measures. It combines a professional-source lexicon, negation and accounting-status logic, channel-specific financial encoders, exact evidence gates, conservative aggregation, and an audit ledger. Across 24,909 Hong Kong firm-years (2008–2025) it retrieves and scores 543,527 snippets, with a rigorous evaluation including ablations, a stratified human audit, multi-seed encoder comparisons, and temporal testing. FinBERT generally achieves higher mean F1 than Contemporary encoders across seven of eight encoder-task combinations; temporal F1 ranges from 0.702 to 0.872. Abstaining on the least-confident 20% of temporal observations raises retained-sample F1 by 0.050–0.077. Deterministic Qwen3-8B shows strength on commodity and negation evidence but weaker performance on foreign-debt and accounting-context labels, highlighting limits of general-purpose LLMs for specialized financial tasks. The strict FX score is negatively associated with linked FX exposure measures, while the generic broad score is not, providing external construct validation rather than causal hedging effects. The study contributes a tested, auditable decision-support pipeline where retrieval, status logic, classification, uncertainty handling, aggregation, and external validation remain independently auditable [page::1]
This study tests whether GPT-4 can outperform humans in generating high-quality new-product ideas aimed at college students at a $50 price cap. Across three pools (student ideas, GPT-4 zero-shot, GPT-4 few-shot), AI-generated ideas show higher average purchase intent, with few-shot prompting performing best [page::12]. However, AI ideas are perceived as less novel and more similar to each other, indicating a narrower solution landscape [page::16][page::17][page::15]. Importantly, AI-generated ideas dominate the top decile, being about seven times more likely to appear in the top 10% than human ideas, underscoring substantial value in AI-driven ideation for high-quality outputs [page::3][page::19]. The results imply a productivity premium for AI in ideation and a potential shift toward idea evaluation and AI-human collaboration to maintain diversity and novelty [page::3][page::21].
<paper_abstract>This study employs a time-varying parameter vector autoregression (TVP-VAR) to quantify dynamic connectedness and directional volatility spillovers between seven critical mineral ETF portfolios and macro-financial variables (energy market, carbon emissions, market sentiment, and global infrastructure). The analysis uses daily data from 2013-05-01 to 2023-05-02 and distinguishes pre- and post-COVID-19 periods to capture structural breaks during the energy transition. Findings reveal ESG scores significantly influence spillovers, with higher-ESG portfolios more likely to act as net transmitters, while energy and carbon-emission futures tend to be net receivers. The COVID-19 episode induced substantial shifts in transmission roles, and ESG-based risk links emerge as a key systemic factor for investors seeking hedging and diversification in a dynamic macro-financial network. This work contributes by leveraging ETF data to reflect investable risk and by linking ESG performance to a portfolio’s directional role within a broader connectedness framework. [page::0]
This paper proposes an ex-ante spatial-temporal carbon response framework for day-ahead nodal carbon intensity (NCI) forecasting and proactive dispatch, leveraging a hierarchical multi-agent deep learning model (ACGD) with DSAM and an LLM-based cooperatio n system. It couples RES-forecasting (Tier-1 ANN) with a CNN-GRU-DSAM (Tier-2) for NCI prediction, and coordinates Geographically Dispatchable Loads (MESSs and DDCs) for spatial-temporal low-carbon scheduling. Case studies on a modified IEEE 33-bus network show that reducing scheduling latency by one hour enables more than 30% system-emission reduction. The work shifts NCI from a retrospective metric to a predictive operational signal, enabling proactive carbon management in cleaner power systems [page::1]
