个人征信机构自动化决策的风险识别与区分规制

个人征信机构运用自动化决策提供信用信息查询、信用评价与信用反欺诈产品或服务,可能因传导不实信息、复现社会偏见与歧视、弱化人类决策个性化空间等对人的权益产生风险。通过对个人征信机构自动化决策应用场景进行分析发现,信用信息查询类产品可引发“算法黑箱”风险,信用评价类和信用反欺诈类产品可同时引发“算法黑箱”“算法歧视”等风险。我国现有法律衍生出个体赋权机制、透明度义务机制和第三方主体外部监督机制,以应对自动化决策引发的各类风险,但单一的规制机制较难解决个人征信机构自动化决策的风险规制难题。对个人征信机构自动化决策的风险进行规制,需要综合考量规制对象、规制目的和规制实效,并以个人征信数据处理关系为分析背景,建构公私法规制机制相衔接的立体化风险规制框架。     

大型互联网平台算法歧视下的金融消费者权益保护问题研究

大型互联网平台得以运行的核心是算法的应用,一方面其促进了互联网经济的飞速发展,可作用于精准客户营销、智能决策、创新商业模式、提升效率等多个领域,另一方面也引发了金融公平公正价值缺失、金融消费者权益保护缺位等多个问题.目前,并没有关于大型互联网平台算法歧视下的金融消费者权益保护问题的细致研究.因此,本文梳理了金融算法数据源获取的3种方式,并将大型互联网平台算法歧视分为设计者主观歧视、机器学习主观歧视、客观系统算法歧视3种类型,归纳总结国内外算法歧视下的法律规制现状及存在的问题,最后提出大型互联网平台算法歧视下的金融消费者权益保护相关政策建议.     

数据正义、算法歧视与规制

大数据的蓬勃发展,推动着人工智能算法在工商业活动中的深度应用,此过程中,数据分析和算法内部演算带来了针对特定群体的歧视性对待。作为传统歧视在算法领域的一种技术表现,算法歧视主要根源于归纳式逻辑等技术陷阱,相较于传统歧视而言,算法歧视造成了更深层的非正义现象。对此,应通过数据开放、算法透明与优化、伦理审查等手段推动算法监管转型,逐步建立完善的算法监管制度,实现对算法歧视的法律规制,确保数字时代社会的安全和发展。     

算法个性化定价：经济原理、实施风险与法律规制

作为一种新型价格机制与定价秩序，算法个性化定价本质是一种经济实践：网络效应、长尾理论诠释了其实施基础和动力机制，价格歧视是算法个性化定价的经济学属性。算法个性化定价虽然没有超越价格歧视的实质范畴，但算法与差异定价的结合却可能产生对消费者知情权和公平交易权的普遍性损害以及不正当竞争和垄断风险。鉴于我国现行《价格法》《消费者权益保护法》《反垄断法》对算法个性化定价的规制都尚存缺漏，未来应当通过完善《价格法》中有关价格欺诈与价格歧视规制的条款，优化《消费者权益保护法》中涉及消费者知情权、公平交易权的相关规则，改进《反垄断法》中对于交易相对人与市场支配地位的判定，实现对算法个性化定价的协同规制。     

公共治理视域下自动化应用的法律规制

摘要技术飞跃与治理提升之间的关联通常被假定,也早已被吸纳进“数字中国”“智慧社会”等政策实践,但法律的习惯性滞后带来了监管套利的空间。公共治理领域的自动化应用经历了从基于数据库编码的计算机自动化到基于机器学习的算法自动化再到基于神经网络的超级自动化的逐步跃迁,从早先自下而上专业人士的“辅助/参考”和个别部门的“部署/应用”嬗变为自上而下的社会化“嵌入/集成”,公权力、私权力及私权利的关系发生了结构性转换。相应地,国际通行、日益趋同的个人隐私保护（事前同意）、算法可解释性（事中监测）和完整履责链条（事后追责）的监管范式也应依照宏观技术共治、中观价值位阶和微观权利保障三位一体的规制路径进行体系化重构。     

ChatGPT对数字金融的影响及其法律规制

ChatGPT作为人工智能发展新的里程碑，在数字经济时代与数字金融的融合发展，给金融业带来机遇的同时也伴随着相应的法律风险。ChatGPT对数字金融发展的影响包括：驱动金融科技研发，优化金融要素配置；拓展金融服务模式，促进金融产业数字化发展；完善金融风险防控，提高金融风险预测能力。但ChatGPT也衍生出了一定的金融法律风险，具体包括：金融数据法律风险、金融算法法律风险、金融算力监管风险和金融应用场景风险。为有效对其进行规制，应该采取四方面措施，一是增进ChatGPT数据规制，确保金融数据合法；二是优化ChatGPT算法，促进算法合理化；三是强化ChatGPT算力规制，推动算力资源共享；四是廓清ChatGPT应用场景规制，优化金融应用场景风险管理。     

论金融科技企业数据合规治理中的技术规则

对金融科技企业数据合规的治理，不能只依赖于单一的法律规则，还应发挥技术规则的优势作用。作为一种内在的自治性规范，技术规则在数据合规风险的识别、防范与应对中具有重要的地位和价值意义。由于技术规则与法律规则在作用方式、内容侧重点与强制力度上存在诸多差异，它们的应用场景也有所不同。技术规则主要应用于事前防范与事中规制，法律规则更多应用于对权利义务责任的明确与事后规制。为更好地发挥二者的不同功效，推动二者良性互动，可以从技术规则的“法律化”和法律规则的“技术化”两大路径展开。     

定向增发的法律界定与规制路径——兼谈对中小股东的保护策略

不同于上市公司公开发行股票或债券的融资模式,定向增发属于再融资途径,对于解决流动资金短缺、实现扩股增资、参与公司重组和并购等具有不可替代的价值,但定向增发往往牵涉到的高额利益输送、大股东压迫小股东等问题亟待法律调整和规制.在重构监管思路的市场观念之下,宜借鉴美国私募发行的法律规则,从发行人与认购人两个维度扩展法律规制的边界,实现对中小股东的有力保护.     

反垄断法下平台自我优待行为的规制路径探析

从行为表现看，平台自我优待行为可分为三类：通过不正当手段对竞争对手的交易设置障碍的行为，通过不正当手段对用户进行诱导的行为以及不对等的资源共享行为。从既有滥用市场支配地位的行为类型出发，差别待遇的规制路径更为合适，但面临“交易相对人”和“条件相同”这两个条件难以满足的挑战，可以通过调适解决：一方面，限定“单一经济实体原则”的适用范围，将自营卖家视为交易相对人；另一方面，在实质层面从平台对自营卖家和其他卖家是否存在歧视待遇进行考察，不应对“条件相同”进行过度要求。对于既有滥用行为类型不能涵盖的优待行为，可以通过兜底条款进行规制。在违法性认定方面，应当从竞争优势的来源、反竞争的效果评价以及正当理由的抗辩等三方面，综合对其判断。     

模型算法审计：理论内涵、国际经验与审计框架

模型算法是大数据、人工智能等数字化技术和数字经济发展的核心要素，对数字化转型起着关键性作用。模型算法的应用越来越广泛，在提升管理水平和自动化程度的同时，也对模型算法风险监管提出了严峻挑战。美国、英国、欧盟、加拿大等已陆续出台模型算法监管规定和审计指引，英国、荷兰等最高审计机关已开展模型算法的审计实践。本文基于以上分析，提出模型算法审计的概念内涵，尝试构建了适合我国的模型算法“五维”审计框架，回答了未来我国模型算法审计“谁来审”“审什么”“怎么审”“审计的条件”以及“审计的原则”等五个方面的问题，以期为今后开展模型算法审计实践提供理论参考。     

上市公司“信贷期限歧视”再分析

本文通过采用不同的债务期限结构计算方法和分年度回归,对“信贷期限歧视”提出了一些思考。在保留静态债务期限的同时,本文采用增量数据,即用“新发生长期借款占新发生总借款的比例”来衡量企业的动态债务期限。本文发现在新增银行债务里,私有企业与国有企业的债务期限结构没有显著差异。     

TSFDC: A trading strategy based on forecasting directional change

Directional Change (DC) is a technique to summarize price movements in a financial market. According to the DC concept, data is sampled only when the magnitude of price change is significant according to the investor. In this paper, we develop a contrarian trading strategy named TSFDC. TSFDC is based on a forecasting model which aims to predict the change of the direction of market's trend under the DC context. We examine the profitability, risk and risk‐adjusted return of TSFDC in the FX market using eight currency pairs. The results suggest that TSFDC outperforms the buy and hold approach and another DC‐based trading strategy.     

Trading performance and market efficiency: Evidence from algorithmic trading

In India, National Stock Exchange directly identifies algorithmic trading participation. Algorithmic traders possess intraday market timing skills. Results are not motivated by extreme short-term signals or transitory price trading. Magnitude of market timing performance in cross-sectional group of traders shows that they earn profit across all the cases, and maximize while providing liquidity. Volume-weighted-average-price decomposition analysis reports algorithmic traders earn profits through intraday market timing performance for five-minute and one-minute intervals, and it is higher compared to short-term market timing performance across all trader groups. Order imbalance and price delay regressions show that algorithmic trading significantly improves price efficiency.     

Does high frequency algorithmic trading matter for non-AT investors?

The extant literature has typically measured the impact of high frequency algorithmic trading (HFT) on short term outcomes, in seconds or minutes. We focus on outcomes of concern for longer term non-algorithm investors. We find in some cases HFT increases volatility arising from news relating to fundamentals. Furthermore HFT is associated with the transmission of that volatility across industries, and that transmission is based on short term correlations. Finally, we find that the period since the introduction of algorithmic trading (AT) has seen increases in both the variances and covariances of return volatility in most industries. However increases in the variances has not been uniform in that it has fallen sharply in a few industries. The magnitudes are such that, overall, AT has coincided with reduced return volatility variance.     

Does high-frequency trading actually improve market liquidity? A comparative study for selected models and measures

The increasing volume of messages sent to the exchange by algorithmic traders stimulates a fierce debate among academics and practitioners on the impacts of high-frequency trading (HFT) on capital markets. By comparing a variety of regression models that associate various measures of market liquidity with measures of high-frequency activity on the same dataset, we find that for some models the increase in high-frequency activity improves market liquidity, but for others, we get the opposite effect. We indicate that this ambiguity does not depend only on the stock market or the data period, but also on the used HFT measure: the increase of high-frequency orders leads to lower market liquidity whereas the increase in high-frequency trades improves liquidity. We hypothesize that the observed decrease in market liquidity associated with an increasing level of high-frequency orders is caused by a rise in quote volatility.     

Automated trading with boosting and expert weighting

We propose a multi-stock automated trading system that relies on a layered structure consisting of a machine learning algorithm, an online learning utility, and a risk management overlay. Alternating decision tree (ADT), which is implemented with Logitboost, was chosen as the underlying algorithm. One of the strengths of our approach is that the algorithm is able to select the best combination of rules derived from well-known technical analysis indicators and is also able to select the best parameters of the technical indicators. Additionally, the online learning layer combines the output of several ADTs and suggests a short or long position. Finally, the risk management layer can validate the trading signal when it exceeds a specified non-zero threshold and limit the application of our trading strategy when it is not profitable. We test the expert weighting algorithm with data of 100 randomly selected companies of the S&P 500 index during the period 2003–2005. We find that this algorithm generates abnormal returns during the test period. Our experiments show that the boosting approach is able to improve the predictive capacity when indicators are combined and aggregated as a single predictor. Even more, the combination of indicators of different stocks demonstrated to be adequate in order to reduce the use of computational resources, and still maintain an adequate predictive capacity.     

High-frequency trading and stock liquidity: An intraday analysis

This paper studies the impact of high-frequency trading (HFT) on intraday liquidity of CAC40 stocks listed on Euronext. Spreads display an intraday L-shaped pattern, while quoted depth follows an inverse pattern: low at the open and increasing towards the end of the trading day. When liquidity demand is particularly high, there is a high rate of order cancellations attributable to high-frequency traders who use frequent order cancellations to strategically manage their limit orders and close positions near the market close. Using the generalized method of moments estimator, we generate strong evidence that greater intensity of HFT is associated with lower spreads and higher depth. The positive effect of HFT on liquidity is due mainly to decreased adverse selection costs arising from asymmetric information among market participants.     

Algorithmic market making for options

In this article, we tackle the problem of a market maker in charge of a book of options on a single liquid underlying asset. By using an approximation of the portfolio in terms of its vega, we show that the seemingly high-dimensional stochastic optimal control problem of an option market maker is in fact tractable. More precisely, when volatility is modeled using a classical stochastic volatility model—e.g. the Heston model—the problem faced by an option market maker is characterized by a low-dimensional functional equation that can be solved numerically using a Euler scheme along with interpolation techniques, even for large portfolios. In order to illustrate our findings, numerical examples are provided.     

The information content of a limit order book: The case of an FX market

In this paper we examine the question of whether knowledge of the information contained in a limit order book helps to provide economic value in a simple trading scheme. Given the greater information content of the order book, over simple price information, it might naturally be expected that the order book would dominate. Using Dollar Sterling tick data, we find that despite the in-sample statistical significance of variables describing the structure of the limit order book in explaining tick-by-tick returns, they do not consistently add significant economic value out-of-sample. We show this using a simple linear model to determine trading activity, as well as a model-free genetic algorithm based on price, order flow, and order book information. We also find that the profitability of all trading rules based on genetic algorithms dropped substantially in 2008 compared to 2003 data.     

算法控制视域下平台用工劳动关系认定的困境与出路

“算法控制”描述的是平台应用算法控制劳动力与劳动过程的用工事实。对“算法控制”进行从属性检验,人格从属性的检验要素可从指示与服从、监督与惩罚、考核与薪酬方面考察;经济从属性的判断可从经济依赖性和劳动力与生产资料结合的角度展开;对于利用算法实质实施劳动管理的平台,应认定其用工关系具备组织从属性。以“专送骑手”为分析样本,“算法控制”揭示其存在“共同雇主的劳动关系认定困境”。对此,规制重心应从“劳动关系”转向“劳动权利”,将共同雇主管理下用人单位如何归属的难题转化为共同雇主内部的用工责任如何分担的问题。平台处于平台用工生态的关键环节,控制核心劳动条件,应承担“守门人”责任。“从‘劳动关系’到‘劳动权利’”之提倡,强调对具有一定程度从属性的平台从业者予以相应劳动保护,对其他类型平台用工的劳动保护研究亦具有方法论意义。