financial planning and analysis

This paper addresses longevity risk, market volatility, high operational costs, and default probability in pension plan management. We propose a simulation-based modular optimization framework that innovatively integrates adaptive asset allocation with dynamic benefit adjustment, enabling real-time, joint liability–asset management. A customizable, multi-dimensional performance metric system supports systematic calibration of strategic parameters. Numerical experiments demonstrate that introducing limited flexibility—such as annual portfolio rebalancing and modest benefit adjustments—reduces operational costs by 18%–32% and decreases default probability by up to 76%. The framework thus provides a scalable, empirically validated methodology for designing robust, cost-efficient, and low-default pension systems.

FinOps faces a core challenge in deriving timely insights and enabling efficient decision-making due to heterogeneous, multi-source cloud billing data—exhibiting inconsistencies in format, categorization, and metric definitions. To address this, we propose the first autonomous AI agent framework specifically designed for FinOps, enabling end-to-end automation spanning multi-source bill ingestion, semantic alignment and fusion analysis, and cost-optimization recommendation generation. Grounded in a goal-driven paradigm, the framework integrates both open- and closed-weight large language models, and tightly couples task planning, reasoning, and execution modules to emulate domain-expert understanding, decision-making, and action capabilities. Experimental evaluation demonstrates that the agent matches senior FinOps practitioners in recommendation accuracy, response latency, and optimization benefit estimation—thereby significantly enhancing real-time cloud cost governance.

This study investigates core challenges confronting corporate financial management under remote and hybrid work models—specifically, weakened budgetary control, diminished financial transparency, and inefficient cross-departmental collaboration. Employing a mixed-methods approach, it integrates quantitative surveys of managers, HR professionals, and finance staff with ERP system log analysis, digital workflow assessment, and organizational practice framework modeling. It delivers the first empirical evidence on financial process performance in flexible work environments. Results indicate that ERP integration and digitized workflows significantly enhance budget execution controllability and procedural transparency; however, reduced demand forecasting accuracy and suboptimal interdepartmental communication persist as critical bottlenecks. Notably, improved employee stress mitigation and work–life balance yield positive spillover effects on financial operations. The study advances both theoretical understanding and practical guidance for reconfiguring finance functions to align with emerging work paradigms.

This paper addresses two key limitations in multi-asset active allocation: insufficient integration of momentum and trend signals, and weak tail-risk control. To this end, we propose a dual-layer协同 framework that jointly generates and fuses trend signals at both the asset-class level (equities, bonds, commodities) and the risk-factor level (e.g., value, momentum, volatility), embedding them directly into portfolio optimization. Methodologically, the approach combines rolling-window momentum ranking, multi-horizon trend filtering, risk-parity weighting, and volatility-targeting constraints. Its primary contribution lies in the first systematic, cross-dimensional co-modeling of trend signals across assets and factors—simultaneously enhancing returns and mitigating downside risk. A 22-year backtest demonstrates that the strategy delivers an annualized excess return of 3.2% relative to benchmarks including the Bloomberg Barclays US Aggregate Bond Index and the MSCI ACWI Index, while reducing maximum drawdown by 37%.

Conventional financial ratio analysis at the industry level suffers from statistical deficiencies—including skewness, non-normality, direction dependence (sensitivity to numerator/denominator selection), and outlier susceptibility. Method: This paper pioneers the systematic application of Compositional Data Analysis (CoDA) to finance, introducing clr/alr/ilr transformations for geometric mean aggregation, compositional principal component biplots, compositional k-means clustering, and compositional linear regression, alongside a CoDA-based DuPont decomposition framework. Validation employs the CoDaPack toolkit on Spanish winery financial statements. Contribution/Results: The approach enables unbiased industry-level mean estimation, visualizes structural heterogeneity in financial composition, identifies robust performance clusters, and supports direct modeling of ratios with interpretable, statistically coherent regression. This work establishes a theoretically consistent, robust, and reliable analytical framework for financial ratio modeling.

This work addresses the growing demand for personalized financial advisory services by proposing a novel role-playing framework grounded in large language models. Existing approaches struggle to encode consistent, scalable investment reasoning capabilities, often yielding overly generic advice through standard prompting. In contrast, the proposed method uniquely integrates fund disclosure documents, portfolio holdings changes, market context, and fund manager commentary within a multi-agent loop comprising actor, evaluator, and refiner components that iteratively refine the advisor’s performance. This enables the system to develop transferable, manager-specific investment reasoning—not merely stylistic adaptation. Experimental results demonstrate superior performance over baselines in portfolio reconstruction and commentary alignment tasks, while generating more concrete and actionable investment advice in dialogues involving market scenario generation and investor profile matching.

研究通过FWBench工具评估了语言模型在成本限制下选择和使用时间序列预测进行决策的能力，测试了包括小型语言模型在内的十种配置。

This study addresses the limitations of traditional investment approaches that utilize data solely for asset pricing while neglecting its systematic impact on portfolio construction and performance attribution under real-world constraints, often leading to suboptimal decisions. To overcome this, the paper proposes a novel three-stage knowledge-optimization framework encompassing decision architecture design, portfolio selection, and performance evaluation. For the first time, it integrates data, models, and business insights into actionable knowledge units embedded throughout the investment process. The framework innovatively introduces a knowledge-augmented proxy for the ex-ante Sharpe ratio to enhance knowledge-driven performance attribution. Empirical results across multiple scenarios demonstrate that the proposed method significantly elevates the explicit knowledge value in investment decisions, yielding improved portfolio performance and interpretability under practical constraints.

This study addresses the limitations of traditional target-date funds (TDFs), which lack an explicit link to a defined retirement income goal and fail to model risk control at the portfolio level. The authors propose a novel framework centered on an exogenously specified income target, incorporating a time-declining conditional value-at-risk (CVaR) constraint path that enables dynamic asset allocation within a feasible risk set. Two new evaluation metrics—probability of goal attainment and cumulative risk—are introduced. The approach eschews period-by-period optimality assumptions, instead combining stochastic asset allocation sampling with multi-asset backtesting, validated in the context of Chile’s pension reform. Findings reveal that the age at which risk reduction begins is critically important, and that contribution density has a hard lower bound below which investment returns alone cannot compensate for structural underfunding.

This work addresses the limitations of existing financial agent evaluation frameworks, which often rely on static benchmarks or focus solely on final returns, thereby lacking traceability of decision-making processes and hindering fine-grained, fair performance assessment in dynamic markets. To overcome these challenges, the authors propose a unified performance tracking platform for financial agents that, for the first time, enables persistent logging of the complete decision trajectory—from market observation to trade execution. By integrating a time-consistent market data interface, a multi-agent collaborative architecture, and an end-to-end logging system, the platform supports interactive, cross-market and cross-model attribution analysis. Deployed across Hong Kong, U.S., and A-share markets, the system—augmented with a visual Trading Arena interface—significantly enhances the transparency, interpretability, and diagnostic capability of agent evaluation.