AI is the most disruptive direction in the world in 2023, triggering the pursuit of funds in the A-share market, and has become an important direction for the active expansion of investment research in the public offering industry.
According to the reporter of China Fund News, at present, many public offering institutions are actively using AI technology to explore landing scenarios in multiple business fields such as capital trading and quantitative investment research, and AI can be described as empowering investment and research capabilities.
Not only that, fund companies also vigorously recruit machine learning researchers and engineers, take many measures to gather excellent AI talents, and even some companies offer millions of annual salaries to attract talents. The application and landing of a large number of AI technologies may give new interpretations to the market and industry in the future.
Attract AI talent at a heavy cost
A few days ago, a quantitative private placement "daily salary of 4,000 yuan to recruit interns" has become a hot topic in the recent market, reflecting that quantitative, AI and other technologies are increasingly valued by financial institutions. Looking at this year, fund companies are vigorously recruiting machine learning researchers and engineers.
The reporter noticed that on the recruitment website, many fund companies are heavily seeking machine learning talents, and many institutions offer an annual salary of one million. A job posting shows that a fund company is hiring machine learning development engineers, giving a salary of 900,000 to 1.12 million per year.
There are also companies that recruit under real names. For example, Chuangjin Hexin Fund is looking for machine learning development engineers. In terms of job responsibilities, in this position, he needs to be responsible for the research and application development of artificial intelligence technologies and algorithms such as machine learning, NLP, OCR, RPA, etc., and form competitive intelligent application solutions based on the actual application scenarios of fund business, and implement them; Track the latest research results and technology trends of artificial intelligence, and introduce relevant technologies to drive business innovation in a timely manner. At the same time, Penghua Fund is also recruiting machine learning engineers in real names.
Another large fund company is also hiring machine learning/deep learning, offering competitive salaries of 25-50K·23. There are also fund companies that offer higher chips of 60-90K·15 salaries.
In addition to recruiting talent on job boards, many companies also have diversified talent strategies.
According to relevant people from Bosera Fund, in terms of recruiting and reserving machine learning-related talents, we first establish cooperative relationships with top universities and research institutions to attract outstanding graduate and doctoral students through internships, project cooperation and campus recruitment, for example, we are jointly holding quantitative competitions with Peking University CCF to seek suitable talents.
"Secondly, Bosera Fund has also established a good reputation in the industry and continues to attract professionals who have already made achievements in the field of machine learning. Finally, the company provides continuous training and learning opportunities for employees, has established AI interest groups, and regularly conducts seminars to identify employees with learning ability and innovative spirit from within. The above person said.
AI technology empowers public fundraising, investment and research
Behind the active recruitment, it is optimistic that AI technology can empower the investment and research business of public funds.
According to the above-mentioned Bosera fund personnel, in investment research, the company mainly uses deep learning and tree models to analyze market data and fundamental factor indicators, capture complex laws in asset prices and trading behaviors, and assist investment researchers in stock selection logic generation and stock selection signal mining. Deep learning is mainly used for factor mining, and machine learning algorithms are mainly used to capture the quantitative relationship and change law between the expected return of assets and factor indicators on a rolling basis.
First, machine learning algorithms can handle large, high-dimensional data compared to human researchers, which is especially important when dealing with complex financial data.
Second, unsupervised machine learning can sometimes reveal complex relationships and hidden laws between data, provide many enlightening new perspectives and new perspectives, can give researchers and fund managers more inspiration, and finally machine learning can also replace some repetitive work, such as information extraction, risk monitoring, high-frequency trading and so on.
In fact, through the use of machine learning technology, some fund companies have already obtained practical benefits. Lei Jun, assistant general manager of Great Wall Fund and general manager of quantitative and index investment department, said that in the past, more traditional fundamental multi-factor frameworks were used, and at the end of last year, the company applied a self-developed factor mining system based on deep neural networks and machine learning frameworks.
"From the past application of private equity and other peers, for products such as small caps, factor mining and deep learning frameworks have shown strong competitiveness, which has also been verified in the internal simulation disks that we began tracking in 2021. Therefore, at the end of last year, this set of Alpha model began to be applied to products with small cap market capitalization, and in the breadth field, the ICInformation coefficient of this methodology has stronger ability to predict the return of factors and higher stability of forecasting. We applied this method to the Great Wall CSI 500 Index Enhancement and Great Wall Quantitative Small Cap, which performed relatively well this year. Lei Jun further said.
Talking about the advantages of machine learning in the research field, Lei Jun analyzed, "First, compared with the linear index analysis of traditional multi-factor models, deep learning models can process some nonlinear data, such as text, a paragraph of text or a research report. The second is the difference between the section and the time series, which has many models to do under the deep learning framework. Third, there will be more non-structural things in the final result, and it is a system, which can theoretically do different 'supplies' and expand the huge factor library. ”
In addition, some people believe that on the whole, fund companies use AI-related technologies to empower research, including the purification and processing of massive information; Research data combing and modelling of upstream and downstream links of the industrial chain; Sentiment analysis of industries, targets, etc. based on Chinese text training. Through AI technology, the potential of research can be greatly released and the efficiency of research can be improved. In the future, with the further landing of AI technology, it may bring many changes to the overall investment and research.
Or will gradually participate in investment research
More and more artificial intelligence technology has been applied to the investment and research process, which has become an irresistible trend.
"Traditional machine learning development has a high training threshold, high requirements for training data, and limited application scenarios for investment and research. With the development of new artificial intelligence technology represented by large models, it is believed that artificial intelligence technology will play a greater role in quantitative investment, data analysis, compliance digitalization and other fields. The above-mentioned Bosera fund person said that in the future, it will continue to pay attention to and follow up the latest artificial intelligence technology, and apply it in investment research and daily work.
According to a large fund company, the application of machine learning in financial investment mainly includes predicting the stock market, risk control, smart portfolio and other aspects. It can be used to improve the quality of investment decisions and effectively manage financial investments, promote investment success, and enable investors to achieve higher returns with less risk.
"A very important feature of financial data is the low signal-to-noise ratio. On the one hand, because there is too much noise, it will drown out the real valid information; On the other hand, the existence of trading behavior makes the low signal-to-noise ratio of the financial market sustainable for a long time. The above people said that a single learner often cannot accurately capture some important characteristics and effective signals of the market, resulting in large deviations. The main purpose of machine learning is to discover or reproduce patterns. Continuous improvements in algorithms may improve machine learning performance on low signal-to-noise ratio data. In the era of comprehensive registration system, machine learning has broader application prospects, which is conducive to improving investment efficiency and assisting investment decisions.
Talking about the specific impact on the market, Lei Jun also has an in-depth analysis, he believes that from the IC point of view, the current deep learning applicable market is still very extensive, bull market, bear market, shock market can be applied, but there will be some differences.
"Overall, the effectiveness of ICs is declining, and the effectiveness of the market is improving. From a long-term perspective, more and more people are using this technology or this data to participate in the market. Lei Jun said that the effectiveness of deep learning is reflected in the two levels of bulls and bears, and the bears are estimated to be slightly better than the bulls, and when the market is weak or volatile, this type of product will rank better in the market. In the strong offensive market of the whole market, it may not outperform track stocks, but it is expected to outperform the index.
This article is from China Fund News