Lei Feng.com AI Financial Review reported that on January 20, 2018, it was sponsored by China Cyberspace Security Association, Financial Technology Innovation Alliance, Industrial and Commercial Bank of China, China Financial Computer Magazine, and jointly sponsored by Tencent, Baidu and IBM. The 2nd China Financial Technology Innovation Conference was held in Beijing.
According to the AI ​​Financial Review of Lei Feng.com, the conference jointly issued a special report, “Intelligent Financial Joint Report: Advancing with AI, Winning the Futureâ€; at the same time, domestic mainstream financial institutions such as Industrial and Commercial Bank of China, Agricultural Bank of China, Bank of China, and Guotai Junan Securities, Research institutes such as the Intelligent Finance Technology Center of the Interdisciplinary Information Institute of Tsinghua University have released the latest achievements in financial technology and application innovation cases.
Among them, the intelligent investment management system developed by the Intelligent Finance Technology Center of Tsinghua University's Interdisciplinary Information Research Institute - high-profile investment engine system.
The underlying architecture for building financial market modeling decision-making optimization
Tsinghua University's cross-information research was led by Yao Zhizhi. According to Lei Feng's AI financial review, Tsinghua University's Financial Technology Center was established in April 2016. In December 2017, the financial technology center was upgraded to the Financial Technology Research Institute. It is its own financial technology company, Fortune Engine Technology Co., Ltd., which is used to carry out industrial cooperation in the technology of financial technology centers.
At the same time, the center spokesperson said that the Cross Information Institute of Tsinghua University established the world's first financial technology laboratory with Ant Financial in October 2016, and on this basis established a four-school alliance, including: Tsinghua University University, Princeton, Korea Advanced Institute of Technology, Risk and Management Center of the Higher Business School of Northern France. “Before, we have implemented industry-university-research projects in Europe and the United States, and have done research and development and system research on financial market simulation and financial optimization algorithms. The other is to build a system for TowersPerrin.â€
Tsinghua University's Interdisciplinary Information Institute said that to build a top-level structure for the wealth management community, professional investors are more concerned about risk management. In the United States, the underlying technology and underlying algorithms cover more than 85% of financial institutions. The biggest difference between professional and professional is that professional investment means that regardless of value investment, quantitative investment, and fundamental investment, the significant difference is professional investors. Can know why they are losing money, why make money. "So, the introduction of a scientific investment and investment system is to solve the problem of "what to vote, why to vote, how to vote."
He introduced,
“The technical architecture we are doing is to be a platform for big data. All the underlying asset data are all integrated and modeled and analyzed. We combine the methods of traditional statistics and financial measurement that can be displayed, and now New machine learning methods—machine learning is very competitive in large-scale style mining. On the other hand, statistical and econometric analysis have advantages in financial model interpretability, so the two combine.â€
Tsinghua University Financial Technology Center said that from the technical framework, the entire investment management problem is solved:
The first is the simulation of financial markets, which solves the problem of joint risk modeling in financial markets. In fact, it is to answer "what is it"
The second is the decision-making optimization engine problem of the financial market, to find an optimized solution for all problems, this part uses the operation optimization method to make an optimized decision engine.
Regarding “what to vote forâ€, it is a system that quantifies the keynotes. It conducts a very in-depth analysis of all the underlying assets, and at the same time, the user-level modeling process from the label system to unsupervised learning, that is, the user portrait modeling system. .
Modern brokerages need a full-service risk management engine
To build an investment management infrastructure and the underlying capabilities, at the application level, including the risk management of securities companies, the optimization of various public funds, large institutional investors, bank insurance companies, pension fund assets and liabilities. "The so-called intelligent wealth management, we actually want to use the power of financial technology to empower China's financial institutions, which can drive business growth."
For example, every broker in the United States will have a risk engine - one day without this risk engine, you can't live. There may be a broker in the morning, and it will be gone in the afternoon. There used to be a precedent. There was a brokerage in KCG in the morning, but there was a big bug in the program. In the afternoon, the company’s ownership was lost, and it was acquired in the afternoon.
A spokesperson for the Tsinghua University Financial Technology Center said that a very powerful engine is a very core thing for modern brokers. It can be used to estimate the overall risk of multiple business lines across the company. The implementation of control is very necessary. With the development of China's financial market, each financial institution will need such a core engine system in the future.
“That is to say that financial market modeling decision-making optimization requires someone to do the underlying architecture. Tsinghua University is the best carrier for doing this.â€
For example, the system products developed by the Tsinghua University Financial Technology Center currently include a “MSG Financial Market Simulatorâ€, which is used to first understand the financial market and conduct joint risk modeling of all assets in the financial market, especially Like last year's regulatory level, new regulations were introduced to break the rigid redemption and introduce risks into the market.
However, risk is a matter of probability distribution. “Our approach is to really go to a scientific quantitative analysis and conduct an in-depth analysis of the risk components inside the assets, such as analyzing the nutrients in a dish, how many trace elements, how many proteins, how much starch, etc., instead of floating Modeling methods for characterization."
For example, what kind of risk drivers are internal to assets, and how many macro, fundamental, technical, industry, and company-level risk drivers are affected. The smart wealth management engine first uses the program to figure out these drivers. "In this regard, we have summarized the pool of thousands of risk drivers."
At the same time, based on the machine learning method, we will screen the risk drivers that have an impact on the market in the near future. According to reports, the system combined statistical measurement method to jointly model the income risk of the entire financial market to better understand the final driving factors of the financial market.
“80% of risk changes in the market are quantifiable.â€
decision making
According to Lei Feng's AI financial review, in the investment decision-making process, this decision-making optimization engine, on the one hand, optimizes investment management, on the other hand, it is risk control optimization, and the third aspect is user portrait modeling.
However, Tsinghua's investment engine system has another innovation: the entire user image is divided into multiple levels. According to reports,
On the one hand, based on the user's basic information, flow information, and behavioral data and other information for a comprehensive assessment;
However, labeling users has no direct driving effect on the business. Therefore, the second aspect is to combine the unsupervised learning technology to directly model the user's behavior, directly predict and explain its micro-behavior, and thus carry out investment management and service management.
Finally, it is to integrate the assets of the entire market for quantitative analysis and solve the problem of “what to vote forâ€.
Dongguan Guancheng Precision Plastic Manufacturing Co., Ltd. , https://www.dpowergo.com