[科研交流] 机器学习与人工智能在商业与经济学中的应用专辑（ISE 2024年第4期） [推广有奖]

International Studies of Economics (ISE)

Volume 19, Issue 4

Special Issue on Machine Learning and Artificial Intelligence in Business and Economics

Guest Editor: Prof. Ye Luo, the University of Hong Kong

Special Issue on Machine Learning and Artificial Intelligence in Business and Economics

Ye Luo

机器学习与人工智能在商业与经济学中的应用专辑序言

Finance research over 40 years: What can we learn from machine learning?

Po-Yu Liu, Zigan Wang

金融研究 40 年：我们能从机器学习中学到什么？

Palm as Decentralized Identifiers: Mitigate scrounging of platform economy

Hanmo Wang, Shuqi Wang, Dong Zhang

Palm作为去中心化标识符：缓解平台经济中的“薅羊毛”行为

Factor timing in the Chinese stock market

Yuxiao Wu

中国股市中的因子择时

A new era of financial services: How AI enhances investment efficiency

Zhiyi Liu, Kai Zhang, Hongyi Zhang

金融服务的新时代： 人工智能如何提高投资效率

Investigating the profit performance of quantitative timing trading strategies in the Shanghai copper futures market, 2020–2022

Hongyu Tian, Wei Wang, Mengxin Yang, Ali Yilmaz

2020-2022 年上海期铜市场量化择时交易策略的盈利表现研究

Topic modeling of financial accounting research over 70 years

Mengxin Yang

70 年来财务会计研究的主题建模

Finance research over 40 years: What can we learn from machine learning?

Abstract: We apply machine learning models to a universe of 20,185 finance articles published between 1976 and 2015 on 17 finance journals, and objectively identify 38 research topics. The financial crisis, hedge/mutual fund, social network, and culture were the fastest growing topics, while market microstructure, initial public offering, and option pricing shrank most from 2006 to 2015. We also list each topic's most cited papers, and present the fastest-growing topics among the universe of 130,547 SSRN working papers. Moreover, we find a bibliometric regularity: the number of researchers covering n topics is about twice the number of researchers covering n + 1 topics.

金融研究 40 年：我们能从机器学习中学到什么？

摘要：我们将机器学习模型应用于1976年至2015年间在17种金融期刊上发表的20185篇金融文章中，客观地确定了38个研究课题。从 2006 年到 2015 年，金融危机、对冲基金/共同基金、社交网络和文化是增长最快的主题，而市场微观结构、首次公开募股和期权定价则缩水最多。我们还列出了每个主题被引用次数最多的论文，并介绍了 130,547 篇 SSRN 工作论文中增长最快的主题。此外，我们还发现了一个文献计量学规律：覆盖 n 个主题的研究人员数量大约是覆盖 n + 1 个主题的研究人员数量的两倍。

Palm as Decentralized Identifiers: Mitigate scrounging of platform economy

Abstract: This study investigates a novel challenge within the digital economy: scrounging, which entails the exploitation of false identities to capitalize on benefits offered by digital platforms, such as sales promotions and new user incentives. As a form of fraud, scrounging can substantially affect platform efficiency and has emerged as a critical issue in the contemporary digital economy and Web 3.0 sector. This underscores the necessity to model scrounging behavior and implement strategies to mitigate it. In this study, we develop a model that characterizes scrounging behavior in digital platform promotions and provides a theoretical explanation. We also introduce Palm as Decentralized Identifiers (DIDs), offering Proof-of-Human(PoH) to reduce fake identity prevalence. Unlike traditional blockchain technology, Palm as Decentralized Identifiers (DIDs) utilizes the Human Chain approach, ensuring equitable treatment of all users within the system. We demonstrate the effectiveness of our system in mitigating scrounging and explore its prospective applications in tackling real-world challenges in the digital and Web 3.0 economies.

Palm作为去中心化标识符：缓解平台经济中的“薅羊毛”行为

摘要：本研究针对数字经济中的一种新兴挑战——“薅羊毛”现象展开调查。“薅羊毛”指通过虚假身份利用数字平台所提供的优惠措施（例如促销活动和新用户激励）以获取不当利益。作为一种欺诈行为，“薅羊毛”会严重影响平台效率，已成为当代数字经济与Web 3.0领域中的关键问题。这一问题的凸显强调了对“薅羊毛”行为进行建模以及提出有效缓解策略的必要性。在本研究中，我们构建了一个模型对数字平台促销中的“薅羊毛”行为进行特征化描述，并提供相应的理论解释。同时，我们引入Palm作为去中心化标识符（Decentralized Identifiers,DIDs）的概念，并提供“人类证明”（Proof-of-Human, PoH）机制，以减少虚假身份的存在。与传统区块链技术不同，Palm作为去中心化标识符（DIDs）采用“人链”（"HumanChain"）方法，确保系统中所有用户的公平待遇。我们展示了该系统在缓解“薅羊毛”行为方面的有效性，并探讨了其在应对数字经济与Web 3.0现实挑战方面的潜在应用价值。

Factor timing in the Chinese stock market

Abstract: I conduct an exploratory study about the feasibility of factor timing in the Chinese stock market, covering 24 representative and well-identified risk factors in 10 categories from the literature. The long–short portfolio of short-term reversal exhibits strong out-of-sample predictability, which is robust across various models and all types of predictors. This predictability is significant both statistically and economically, with a simple investment strategy obtaining its return three times higher than the buy-and-hold return in the sample period and a significant annualized 20.4% CH-3 alpha. Portfolio historical volatility and market volatility measurement predictors play crucial roles in the reversal factor premium's robust predictability. However, such results are not evident in predicting all other factors' long–short portfolios as well as all factors' long-wing and short-wing portfolios, and this failure cannot be attributed to their exposure to unpredictable market returns.

中国股市中的因子择时

摘要： 笔者对中国股票市场中因子收益率择时预测的可行性进行了探索性研究，涵盖了共10类24个文献中主流的风险因子。短期反转因子的多空组合表现出很强的样本外可预测性，且该结果基于各种模型和各类预测因子都十分稳健。基于短期反转因子可预测性所构建的简单投资策略，其收益率是直接持有该因子多空组合的收益率的三倍，且CH-3 α可以达到 20.4%的年化收益率水平。投资组合历史波动率水平和市场波动率水平对于预测短期反转因子收益率发挥了关键作用。然而，在预测其他所有因子的多空组合以及所有因子的多头组合和空头组合时，可预测性并不明显，且该现象不能归因于它们在难以预测的市场因子之上的显著暴露。

Abstract:The wide application of AI infinancial investment has significantly enhanced the efficiency of interconnected financial entities and markets within the financial ecosystem, injecting new vitality into the financial sector. AI has injected boundless vitality into financial technology, significantly enhancing the efficiency of financial investments, optimizing industry services, and increasingly emerging as a key force for future changes in the financial industry. Looking ahead, as AI technology continues to advance and innovate, its application in financial investments will become more extensive and profound. As integral components of the financial complex system, traders, financial institutions, investors, and financial market regulators must collaborate closely to address challenges such as over-reliance on AI, algorithmic spoofing, model hallucinations, and legal and ethical risks. By exploring solutions to these risks, they can promote the healthy and sustainable development of the financial investment sector and usher financial services into a new era.

金融服务的新时代：人工智能如何提高投资效率

概要：人工智能在金融投资领域的广泛应用，显著提升了金融生态系统中相互关联的金融实体和市场的效率，为金融业注入了新的活力。人工智能为金融科技注入了无限活力，显著提升了金融投资效率，优化了行业服务，日益成为金融业未来变革的重要力量。展望未来，随着人工智能技术的不断进步和创新，其在金融投资领域的应用将更加广泛和深入。作为金融复杂系统中不可或缺的组成部分，交易员、金融机构、投资者和金融市场监管者必须密切合作，共同应对过度依赖人工智能、算法欺骗、模型幻觉以及法律和道德风险等挑战。通过探索这些风险的解决方案，他们可以促进金融投资行业的健康和可持续发展，并将金融服务带入一个新时代。

Investigating the profit performance of quantitative timing trading strategies in the Shanghai copper futures market, 2020–2022

Abstract: In conducting an extensive examination, we scrutinize the efficacy of algorithmic trading strategies applied to FuturesCopperMainContinuous in the Shanghai Futures Exchange, utilizing a comprehensive data set spanning from January 2020 to December 2022. To mitigate the potential risk of data-snooping bias—the probability that any favorable results may inadvertently arise from random events rather than the inherent value of the strategies employed to generate these results—our study prudently conducts a reality check and advanced assessments. Throughout the evaluated period, the benchmark demarcation between the in-sample and out-of-sample stages is established in February 2022. Regrettably, our meticulous exploration fails to identify any successful or advantageous algorithmic trading strategies within these categories, particularly following the systematic elimination of data snooping bias. These results underscore the intrinsic challenges inaccurately identifying and implementing profit-generating algorithmic trading strategies within the volatile and intricate futures market.

2020-2022年上海期铜市场量化择时交易策略的盈利表现研究

摘要：我们利用2020年1月至2022年12月的综合数据集，对应用于上海期货交易所铜主力连续期货的算法交易策略的有效性进行了广泛的研究。为了降低数据窥探偏差的潜在风险--即任何有利结果可能无意中来自随机事件，而非产生这些结果的策略的内在价值--我们的研究审慎地进行了现实检查和高级评估。在整个评估期间，样本内阶段和样本外阶段的基准分界线于2022年2月确立。遗憾的是，我们的细致探索未能在这些类别中发现任何成功或有利的算法交易策略，尤其是在系统性消除数据窥探偏差之后。这些结果凸显了在动荡复杂的期货市场中准确识别和实施可创造利润的算法交易策略所面临的内在挑战。

[url=https://doi.org/10.1002/ise3.88]Topic modeling of financial accounting research over 70 years[/url]

Abstract: Iutilize latent Dirichlet allocation and dynamic topic model that are machine learning algorithms across a data set encompassing 25,990 financial accounting articles issued from 1956 to 2023 in 16 accounting journals, and impartially ascertain 20 research topics. The topics of mergers and acquisitions, disclosure and internal control, and political connection exhibited the most rapid expansion, whereas management control systems, earnings management, and valuation experienced the greatest contraction from 2014 to 2023. I also catalog the most referenced papers for each topic and highlight the most swiftly expanding and contracting topics within the realm of 21,620 SSRN working papers. Additionally, my analysis reveals a declining trend in the concentration of research interests within published articles over the preceding seven decades. This research on topic classification itself will aid accounting investigators in bypassing superfluous efforts and fostering increased interdisciplinary research.

70年来财务会计研究的主题建模

摘要：我利用机器学习算法中的潜狄利克特分配和动态主题模型，对16种会计期刊从1956年到2023年发表的25,990篇财务会计文章进行数据分析，公正地确定了20个研究主题。从 2014 年到 2023 年，并购、信息披露和内部控制以及政治关联等主题的扩展速度最快，而管理控制系统、收益管理和估值等主题的收缩速度最快。我还对每个主题中被引用次数最多的论文进行了编目，并在21,620 篇SSRN 工作论文中突出强调了扩展和收缩最迅速的主题。此外，我的分析还揭示了过去七十年中发表文章中研究兴趣集中度下降的趋势。这种对主题分类本身的研究将有助于会计研究人员避开多余的工作，促进跨学科研究的发展。