Note: This article was originally published in Singapore Lianhe Zaobao. Babbitt Information was released with the author's authorization.
Fintech is mainly a comprehensive application of a variety of new information and communication technologies, including artificial intelligence, blockchain, cloud computing, big data and other technologies. These technologies have had a significant impact on financial business. Driving financial innovation has brought a variety of financial products and services, intricate business dependencies, and the complexity of risk transmission paths and mechanisms, making it possible to effectively identify risks and formulate targeted regulatory rules. The difficulty of taking timely supervisory measures has increased accordingly.
Public opinion has rarely discussed the application of fintech in financial supervision. According to the author's observations, fintech has also made progress in multiple application scenarios of the regulatory level of the capital market, improving aspects such as the effectiveness, timeliness, deterrence, credibility, and regulatory efficiency of supervision. Here are some examples.
- Li Lihui, Former President of Bank of China: We should step up efforts to improve the technical standards and security specifications of blockchain finance
- National University of Singapore Visiting Professor Bai Shifang: FinTech Promotes Tech Finance
- Long Baitao Comments: Comparing the Development and Supervision of Large Chinese and German Technology Companies
- Article overview of fintech infrastructure change trends and market panorama
- Global Fintech Financing Report 2019: Financing Amount Exceeds 261.9 Billion, Blockchain Financing Leads Amount
- Central Bank: Supports Pioneering Pilot Fintech Innovation Supervision Pilot in Beijing
First, the market surveillance system . Driven by interests, there are various illegal and illegal trading behaviors in the capital market, such as illegal use of off-site allocation of funds to increase leverage, and the use of a large number of linked accounts for market manipulation, which have seriously disrupted market order and fairness. In order to conceal their violations, the perpetrators often use various methods to conceal them. In addition, the number of transactions in the capital market has reached hundreds of millions a day, making it difficult to identify these behaviors by relying only on traditional data statistics monitoring methods Higher and higher.
In order to promote the application of regulatory technology and enhance the front-line supervisory capabilities of the exchange, the stock exchange has developed a new generation of monitoring system that comprehensively uses a stream computing engine, a full-text retrieval engine, and an offline data storage and analysis platform that is highly dynamic and real-time. Big data technologies such as MPP (Massively Parallel Processing ) massively parallel processing database, realize the calculation ability of second-level real-time data analysis, and have the data processing ability of more than one million transactions per second, supporting cross-product, cross-market linkage analysis. In terms of monitoring methods, investigative methods, supervision scope, etc., substantial improvements have been made to make market supervision more complete, monitoring angles more comprehensive, and monitoring methods more in-depth.
Second, intelligent compliance risk control . The intelligent risk control system uses big data and artificial intelligence technology to perform pattern matching on the data and then classify it. According to the transaction styles of different accounts, transaction size, position, operating frequency, operating period, operating source and other behavior styles, with static characteristics, etc. Factors to match and categorize them, thereby digging out potential connections in massive accounts and detecting illegal accounts. Among the large number of suspicious accounts screened out, according to their trading volume and the intensity of the market fluctuations, the accounts are classified according to their impact levels, which facilitates the follow-up of high-risk subjects.
This technology has been put into use by some Chinese securities companies with an accuracy rate of over 80%. After accumulating more and more successful cases, incorporating the cases into the artificial intelligence learning library will further promote the improvement of accuracy, thereby forming a virtuous circle of technology to promote business improvement and business results feedback technology.
Third, intelligent financial audit . Information disclosure, as an extremely important part of the capital market, has always been a key area of focus for supervision. For example, listed companies are often heard about financial whitewashing.
Intelligent financial audit combines expert experience and artificial intelligence technology to classify financial anomalies and form a database of financial anomalies, including revenue and cash flow deviations, sales expenses and income deviations, inventory balances and depreciation preparation deviations, and revenue and cost deviations. , Abnormal quality of credit, abnormal costs and other financial whitewashing issues. Through the analysis of vertical historical fluctuations, horizontal industry trend comparison, cross-checking relationship verification, linkage trend testing and other methods, the financial data is comprehensively analyzed in multiple dimensions, and the financial report quality assessment report is output.
Fourth, intelligent public opinion analysis . Individual investors in China's capital market account for 40.5%, far exceeding 10% of overseas developed capital markets. Because retail investors have a weaker professional judgment on investment value than institutional investors, making China's capital market more susceptible to the influence of public opinion, Internet public opinion analysis has become an important perspective to observe China's capital market.
The public opinion analysis system uses big data and artificial intelligence technology to first perform data collection, data cleaning and processing, summarize data from various sources, and use different collection frequencies for the source according to its importance and credibility. Then use deep learning and natural language processing technology to perform sentiment analysis on the data to determine whether it is positive or negative, label the data, and quantify the impact of public opinion based on the empirical database. Finally, based on the inter-industry relationship, analyze the public opinion transmission route, and predict that the company may be affected.
Fifth, natural language processing . Natural language processing (NLP) includes natural language understanding (reading) and natural language production (writing). The former is to enable the machine to analyze the language structure and understand the meaning of the language, laying the foundation for the next automated processing. The latter is where the machine outputs the existing information into text that humans can understand.
In the financial field, there is a large amount of unstructured data, including regular and irregular information disclosures, news information, fundraising prospectuses for stocks or bonds, investment research reports, legal documents, etc. Due to the huge amount of these data, it is difficult to analyze and process it manually. As an important branch of artificial intelligence, natural language processing technology is a technology in which humans "read" or even "write" human language and text by machines.
Supervisors can apply NLP technology to corporate portraits, investor portraits, public opinion analysis, and information disclosure compliance inspections to improve regulatory efficiency.
From the above examples, we can see that strengthening the research and application of fintech in supervision and guiding the technology to exert its maximum value can protect the financial market's safety and stability while promoting the healthy development of the financial market.
The application of fintech in the regulatory field can bring improvements to the supervision methods and effects, especially the powerful analysis capabilities produced by the combination of big data and artificial intelligence, which makes many more comprehensive, deeper and more timely in the past that could not be achieved by manual or traditional data statistics technology. Analysis becomes possible.
First, improve the effectiveness of supervision. The data channel sources and data types that can be analyzed through technical means have been greatly enriched, the market monitoring scope has become more comprehensive, the monitoring level has been deepened, and the cross-correlation analysis ability of various data has also been greatly improved. This gives regulators more comprehensive insights into the market.
Second, improve the timeliness of supervision. In the past, a large amount of regulatory information was reported by the industry afterwards, and some even required the supervisor to obtain on-site inspections. With the maturity of regulatory technology and the significant enhancement of system processing capabilities, regulators and market institutions can conduct technology docking, extend regulatory tentacles to the forefront of the market, and upgrade post-event reporting to real-time or quasi-real-time reporting . This will effectively improve the timeliness of supervision and avoid the possibility of timeliness of data submission after the fact.
Third, enhance regulatory deterrence. Before the use of regulatory technology, due to the limitations of regulatory methods and the intensive amount of information generated by the huge market, the success rate of mining clues from illegal data to illegal laws and regulations was not high, which helped to a certain extent Speculative. The use of regulatory technology has greatly improved the intelligent analysis capabilities of massive data, and has also formed an effective psychological deterrent to the regulated.
Fourth, enhance the credibility of supervision. With the application of more and more scientific and technological supervision methods, the rigorous calculation of machines will increasingly replace the subjective judgment of humans. Make regulatory standards more uniform and stable, thereby improving regulatory consistency and credibility.
Fifth, improve the degree of inheritance of supervision. The regulatory results and regulatory experience will be fed back to the artificial intelligence system through case methods, so that these knowledge and experience are solidified into the system and can be effectively passed on in the organization. It not only avoids the knowledge fault caused by the changes in the working relationship of employees, but also enables artificial and intelligent to form a virtuous circular relationship that promotes each other.
Sixth, improve regulatory efficiency. The use of semi-automated or automated supervisory technology can provide supervisors with a wide variety of analytical tools and investigation methods. While improving the efficiency of supervision, it is also expected to reduce the overall supervision cost. At the same time, supervisors can be freed from a large number of repetitive tasks, and regulatory resources can be promoted to improve the design of regulatory mechanisms.
In the end, the ideal state of "one foot high and one foot high" was realized.
Bai Shihuan: Visiting Professor, National University of Singapore, Dean of the Li Bai School of Finance, and former Dean of the Monetary Authority of Singapore
Huang Bingxiong: Senior Manager of Shenzhen Stock Exchange