|
on Computational Economics |
| By: | Benjamin Coriat; Eric Benhamou |
| Abstract: | This paper presents a novel hierarchical framework for portfolio optimization, integrating lightweight Large Language Models (LLMs) with Deep Reinforcement Learning (DRL) to combine sentiment signals from financial news with traditional market indicators. Our three-tier architecture employs base RL agents to process hybrid data, meta-agents to aggregate their decisions, and a super-agent to merge decisions based on market data and sentiment analysis. Evaluated on data from 2018 to 2024, after training on 2000-2017, the framework achieves a 26% annualized return and a Sharpe ratio of 1.2, outperforming equal-weighted and S&P 500 benchmarks. Key contributions include scalable cross-modal integration, a hierarchical RL structure for enhanced stability, and open-source reproducibility. |
| Date: | 2025–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2507.18560 |
| By: | Giorgos Iacovides; Wuyang Zhou; Danilo Mandic |
| Abstract: | Opinions expressed in online finance-related textual data are having an increasingly profound impact on trading decisions and market movements. This trend highlights the vital role of sentiment analysis as a tool for quantifying the nature and strength of such opinions. With the rapid development of Generative AI (GenAI), supervised fine-tuned (SFT) large language models (LLMs) have become the de facto standard for financial sentiment analysis. However, the SFT paradigm can lead to memorization of the training data and often fails to generalize to unseen samples. This is a critical limitation in financial domains, where models must adapt to previously unobserved events and the nuanced, domain-specific language of finance. To this end, we introduce FinDPO, the first finance-specific LLM framework based on post-training human preference alignment via Direct Preference Optimization (DPO). The proposed FinDPO achieves state-of-the-art performance on standard sentiment classification benchmarks, outperforming existing supervised fine-tuned models by 11% on the average. Uniquely, the FinDPO framework enables the integration of a fine-tuned causal LLM into realistic portfolio strategies through a novel 'logit-to-score' conversion, which transforms discrete sentiment predictions into continuous, rankable sentiment scores (probabilities). In this way, simulations demonstrate that FinDPO is the first sentiment-based approach to maintain substantial positive returns of 67% annually and strong risk-adjusted performance, as indicated by a Sharpe ratio of 2.0, even under realistic transaction costs of 5 basis points (bps). |
| Date: | 2025–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2507.18417 |
| By: | Mohammad Rubyet Islam |
| Abstract: | The pursuit of alpha returns that exceed market benchmarks has undergone a profound transformation, evolving from intuition-driven investing to autonomous, AI powered systems. This paper introduces a comprehensive five stage taxonomy that traces this progression across manual strategies, statistical models, classical machine learning, deep learning, and agentic architectures powered by large language models (LLMs). Unlike prior surveys focused narrowly on modeling techniques, this review adopts a system level lens, integrating advances in representation learning, multimodal data fusion, and tool augmented LLM agents. The strategic shift from static predictors to contextaware financial agents capable of real time reasoning, scenario simulation, and cross modal decision making is emphasized. Key challenges in interpretability, data fragility, governance, and regulatory compliance areas critical to production deployment are examined. The proposed taxonomy offers a unified framework for evaluating maturity, aligning infrastructure, and guiding the responsible development of next generation alpha systems. |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2505.14727 |
| By: | Connor Lennon; Edward Rubin; Glen Waddell |
| Abstract: | Machine learning (ML) primarily evolved to solve "prediction problems." The first stage of two-stage least squares (2SLS) is a prediction problem, suggesting potential gains from ML first-stage assistance. However, little guidance exists on when ML helps 2SLS$\unicode{x2014}$or when it hurts. We investigate the implications of inserting ML into 2SLS, decomposing the bias into three informative components. Mechanically, ML-in-2SLS procedures face issues common to prediction and causal-inference settings$\unicode{x2014}$and their interaction. Through simulation, we show linear ML methods (e.g., post-Lasso) work well, while nonlinear methods (e.g., random forests, neural nets) generate substantial bias in second-stage estimates$\unicode{x2014}$potentially exceeding the bias of endogenous OLS. |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2505.13422 |
| By: | Dulgeridis, Marcel; Schubart, Constantin; Dulgeridis, Sabrina |
| Abstract: | Accounting fraud poses significant financial and reputational risks for organizations. Traditional detection methods - such as manual audits and red-flag indicators - struggle to keep pace with the growing volume and complexity of financial data. In contrast, artificial intelligence technologies, including machine learning, anomaly detection, and natural language processing, offer scalable, realtime solutions to identify suspicious activity more efficiently. This paper compares conventional fraud detection techniques with AI-driven approaches, highlighting their respective strengths and limitations in terms of accuracy, efficiency, scalability, and adaptability. While AI enables faster and more comprehensive analysis, it also raises challenges related to data quality, algorithmic bias, and transparency. Ethical and legal considerations, including data privacy and compliance with regulations, are crucial for responsible implementation. The paper concludes with strategic recommendations for adopting AI-based fraud detection systems - emphasizing AI readiness, robust data governance, and human oversight. With a thoughtful approach, AI has the potential to significantly enhance the detection and prevention of accounting fraud. |
| Keywords: | Artificial Intelligence, Fraud Detection, Machine Learning, Anomaly Detection, Natural LanguageProcessing, Data Quality, Financial Fraud, Auditor Oversight, Transparency, AI Implementation |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:iubhbm:321858 |
| By: | Zeqiang Zhang; Ruxin Chen |
| Abstract: | The application of Reinforcement Learning (RL) to economic modeling reveals a fundamental conflict between the assumptions of equilibrium theory and the emergent behavior of learning agents. While canonical economic models assume atomistic agents act as `takers' of aggregate market conditions, a naive single-agent RL simulation incentivizes the agent to become a `manipulator' of its environment. This paper first demonstrates this discrepancy within a search-and-matching model with concave production, showing that a standard RL agent learns a non-equilibrium, monopsonistic policy. Additionally, we identify a parametric bias arising from the mismatch between economic discounting and RL's treatment of intertemporal costs. To address both issues, we propose a calibrated Mean-Field Reinforcement Learning framework that embeds a representative agent in a fixed macroeconomic field and adjusts the cost function to reflect economic opportunity costs. Our iterative algorithm converges to a self-consistent fixed point where the agent's policy aligns with the competitive equilibrium. This approach provides a tractable and theoretically sound methodology for modeling learning agents in economic systems within the broader domain of computational social science. |
| Date: | 2025–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2507.18229 |
| By: | Victor Klockmann (JMU - Julius-Maximilians-Universität Würzburg = University of Würzburg [Würsburg, Germany], Goethe University Frankfurt = Goethe-Universität Frankfurt am Main, Max Planck Institute for Human Development - Max-Planck-Gesellschaft); Alicia von Schenk (JMU - Julius-Maximilians-Universität Würzburg = University of Würzburg [Würsburg, Germany], Goethe University Frankfurt = Goethe-Universität Frankfurt am Main, Max Planck Institute for Human Development - Max-Planck-Gesellschaft); Marie Claire Villeval (GATE Lyon Saint-Étienne - Groupe d'Analyse et de Théorie Economique Lyon - Saint-Etienne - UL2 - Université Lumière - Lyon 2 - UJM - Université Jean Monnet - Saint-Étienne - EM - EMLyon Business School - CNRS - Centre National de la Recherche Scientifique) |
| Abstract: | In the field of machine learning, the decisions of algorithms depend on extensive training data contributed by numerous, often human, sources. How does this property affect the social nature of human decisions that serve to train these algorithms? By experimentally manipulating the pivotality of individual decisions for a supervised machine learning algorithm, we show that the diffusion of responsibility weakened revealed social preferences, leading to algorithmic models favoring selfish decisions. Importantly, this phenomenon cannot be attributed to shifts in incentive structures or the presence of externalities. Rather, our results suggest that the expansive nature of Big Data fosters a sense of diminished responsibility and serves as an excuse for selfish behavior that impacts individuals and the whole society. |
| Keywords: | Artificial intelligence, Big data, Pivotality, Distributional fairness, Experiment |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:hal:journl:hal-05165240 |
| By: | Patrick Cheridito; Jean-Loup Dupret; Donatien Hainaut |
| Abstract: | In this paper, we introduce a model-based deep-learning approach to solve finite-horizon continuous-time stochastic control problems with jumps. We iteratively train two neural networks: one to represent the optimal policy and the other to approximate the value function. Leveraging a continuous-time version of the dynamic programming principle, we derive two different training objectives based on the Hamilton-Jacobi-Bellman equation, ensuring that the networks capture the underlying stochastic dynamics. Empirical evaluations on different problems illustrate the accuracy and scalability of our approach, demonstrating its effectiveness in solving complex, high-dimensional stochastic control tasks. |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2505.15602 |
| By: | Haochen Luo; Yuan Zhang; Chen Liu |
| Abstract: | Sparse portfolio optimization is a fundamental yet challenging problem in quantitative finance, since traditional approaches heavily relying on historical return statistics and static objectives can hardly adapt to dynamic market regimes. To address this issue, we propose Evolutionary Factor Search (EFS), a novel framework that leverages large language models (LLMs) to automate the generation and evolution of alpha factors for sparse portfolio construction. By reformulating the asset selection problem as a top-m ranking task guided by LLM-generated factors, EFS incorporates an evolutionary feedback loop to iteratively refine the factor pool based on performance. Extensive experiments on five Fama-French benchmark datasets and three real-market datasets (US50, HSI45 and CSI300) demonstrate that EFS significantly outperforms both statistical-based and optimization-based baselines, especially in larger asset universes and volatile conditions. Comprehensive ablation studies validate the importance of prompt composition, factor diversity, and LLM backend choice. Our results highlight the promise of language-guided evolution as a robust and interpretable paradigm for portfolio optimization under structural constraints. |
| Date: | 2025–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2507.17211 |
| By: | MINAMI, Koutaroh |
| Abstract: | This study explores the potential of machine learning, Long Short-Term Memory (LSTM), to detect asset price bubbles by analyzing prediction errors. Using monthly data of the Nikkei225 Index, I evaluate the performance of LSTM model in forecasting prices and compare with the GSADF test. I find that LSTM’s prediction accuracy significantly deteriorates during periods associated with asset bubbles, suggesting the presence of structural changes. In particular, the LSTM approach of this paper captures both the emergence and collapse of Japan’s late 1980s bubble separately. In addition, it can also capture structural changes related to policy changes in the 2010s Japan, which are not identified by the GSADF test. These findings suggest that machine learning can be used for not only identifying bubbles but also policy evaluations. |
| Keywords: | Bubbles, Generalized Supremum Augmented Dickey-Fuller test (GSADF), Machine learning, Long Short Term Memory (LSTM) |
| JEL: | G10 G17 |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:hit:hcfrwp:g-1-30 |
| By: | Christopher Clayton; Antonio Coppola; Matteo Maggiori; Jesse Schreger |
| Abstract: | Geoeconomic pressure—the use of existing economic relationships by governments to achieve geopolitical or economic goals—is a prominent feature of global power dynamics. This paper introduces a methodology using large language models (LLMs) to systematically identify the application of and response to geoeconomic pressure from large textual corpora. We classify which governments apply pressure to which foreign targets, using which instruments, firms, and products. We demonstrate that firms affected by tariffs respond primarily with price changes whereas firms affected by export controls respond disproportionately by investing in research and development. We document significant heterogeneity in how firms respond to pressure based on whether their home government is applying the pressure, whether their home country is the recipient of the pressure, or whether they are based in an affected third party country. Finally, we quantify the degree of measurement uncertainty generated by the LLM-based analysis by comparing the classifications across multiple open-weight models as well as considering a wide range of variations of our prompts. |
| JEL: | C4 F3 F4 G3 |
| Date: | 2025–07 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:34020 |
| By: | Abdullah Karasan; Ozge Sezgin Alp; Gerhard-Wilhelm Weber |
| Abstract: | In this study, we propose a novel machine-learning-based measure for stock price crash risk, utilizing the minimum covariance determinant methodology. Employing this newly introduced dependent variable, we predict stock price crash risk through cross-sectional regression analysis. The findings confirm that the proposed method effectively captures stock price crash risk, with the model demonstrating strong performance in terms of both statistical significance and economic relevance. Furthermore, leveraging a newly developed firm-specific investor sentiment index, the analysis identifies a positive correlation between stock price crash risk and firm-specific investor sentiment. Specifically, higher levels of sentiment are associated with an increased likelihood of stock price crash risk. This relationship remains robust across different firm sizes and when using the detoned version of the firm-specific investor sentiment index, further validating the reliability of the proposed approach. |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2505.16287 |