Xinhua Finance and economics, Beijing, January 20 – Xiang Wei, deputy general manager of smart Equity Investment Department of Zheshang fund, visited today’s headline finance channel and Xinhua Finance online interview column “investment base methodology” on the 20th to share smart investment strategies.
Guest profile: Xiang Wei, PhD, Department of computer science and engineering, Hong Kong University of science and technology; In 2006, he entered the field of artificial intelligence and once served as the technical director of Baidu International Technology (Shenzhen) Co., Ltd. and the quantitative researcher of Shanghai Tongsheng Tonghui Asset Management Co., Ltd; Founded Bgi Genomics Co.Ltd(300676) elastic computing laboratory and Baidu machine learning Shenzhen Branch; Currently, he is the deputy general manager and AI principal of intelligent Equity Investment Department of Zheshang fund.
Q: what is the difference between AI stock selection and quantification? What is the specific strategy?
Xiang Wei: the operation mode of AI fund is not as magical as you think. At present, the common narrow quantitative investment in the market includes style factor stock selection, high-frequency volume price timing, etc. many quantitative investment players will also use a large number of historical data to train the model for future prediction. We believe that, whether it is active investment or traditional quantitative investment, their income sources can be adopted as the main income components of AI funds. When we have the ability to widely absorb the underlying assets of various income sources, the success rate of dismantling income risk objectives and building combination will be gradually improved.
At the same time, we are not inclined to use AI technology to make predictions, because historical data never represent the future law. Therefore, we use the probability distribution of historical gains and losses to help us consider what the goal of reasonable return threshold is for a certain type of underlying assets or a certain type of risk exposure, Adopting the strategic framework of decentralized investment, as long as we insist on taking only a reasonable return on any kind of assets and constantly rebalance the risk budget, we can also obtain a good profit loss ratio for a long time with the help of the improvement of capital turnover rate.
(photo source: Zheshang Fund)
Q: where does the excess return of AI strategy fund come from?
Xiang Wei: we have a wide range of sources of excess returns. For different betas at the bottom, we will have beta timing benefits according to their respective fundamental valuation emotions; Further, for the disassembly of beta income, we will also disassemble it into macro exposure income, growth income, valuation fluctuation income and emotional trading income. For different types of products, we will also combine and construct different excess return targets according to the differences in return, volatility and capital duration at the capital end.
For example, in the strategy of rotating exposure with style factor, generally speaking, among the reasonable returns we think, high profit quality accounts for 2%, growth accounts for 3%, transaction volatility accounts for 2%, valuation repair accounts for 2%, etc; Another example is the strategy of broad-based industry. If we disassemble the KPI of excess return into 30 primary industries, assuming that each industry accounts for 1% by referring to the winning rate of 70%. If we further disassemble it into more than 100 secondary industries, each industry accounts for 0.3%. The greater the difference in types of excess returns, the more dispersed the KPI dismantling objectives. On the one hand, it will be less difficult to achieve KPIs. On the other hand, if a pullback occurs, there will be more room for dynamic rebalancing of position exposure and the smoother the net value of the portfolio.
Q: what is the principle of using AI to learn the ability of fund managers? Will there be a lag?
Xiang Wei: each fund manager has its own investment logic and its corresponding long board and short board. The purpose of our research on human samples is to extract the risks and opportunities in the long and short boards of different excellent human beings in a quantitative way. For example, the difference between excellent value investors and mediocre value investors is whether they can eliminate the “crooked melons and cracked dates” in a pile of bargains; As another example, excellent growth investors vs mediocre growth investors can strike a balance between the valuation advance overdraft performance duration and the realization of growth options. Therefore, for our team, stock research is only the foundation, and the core competitiveness is to study the ability circle of human excellent fund managers. Such research will lag, but the research on high-frequency trading ability is only a small part of our research goal.
Q: how do you control the optimal sharp ratio and risk?
Xiang Wei: in the current era of fierce insider trading between buyers and sellers, any kind of return or excess return needs to be obtained through a certain form of risk exchange, such as volatility, pullback, or sacrificing a certain amount of capital liquidity, and we design a fund to present it to customers in a quantitative and transparent way, What kind of price you may exchange for what kind of income. Therefore, the net value of our AI product series is disassembled through the KPI described above, and then synthesized by using the permutation and combination of different types of risk exposures.
Q: what do you think is your biggest feature compared with similar fund managers in the market?
Xiang Wei: fund managers of traditional fund companies actually belong to traditional manufacturing industry in our eyes, because most fund managers focus on building portfolios and ignore the matching degree between capital end and asset end. We believe that the essence of wealth management service is a service industry, and the fund with the most growth is not the best fund. On the contrary, the demands of different capital end are actually thousands of people and thousands of faces. Only the asset portfolio scheme with the highest matching degree with the capital end attribute is a good fund.
Therefore, in order to achieve this goal, we can think of using AI technology to arrange and combine various types of risk exposures and income sources. On the one hand, the volatility of different types of asset portfolios is different, but each needs to have very distinct and stable volatility characteristics and relatively certain and transparent income sources, On the other hand, we are also arming our marketing team colleagues with our AI technology, so that with the support of technology, the matching degree, matching timing, matching satisfaction and matching persistence of capital end and asset end can be improved.
Q: the market structure has risen for three years. What do you think of this year?
Xiang Wei: the market beta rose in 2019. In fact, the main reason for the rise was the valuation repair after the valuation error of the overall downward market in 2018; The rise of market beta in 2020 is mainly due to the core assets benefiting from China’s epidemic prevention advantages, such as “Mao index” and Shanghai and Shenzhen 300.
In 2021, because of the burst of core asset valuation bubble and the pressure of economic fundamentals, it was reasonable to regress the value of the valuation. On the other hand, the subdivision circuit leader represented by the CSI 500 and the special new company rose sharply with the global growth rate of growth, so the growth was due to growth.
In 2022, from the perspective of macroeconomic fundamentals, China’s economic and employment data are under pressure in the first half of the year, so the loose state of liquidity is still expected to last for half a year to a year. From these two aspects, the left hand pays attention to the large financial and other value sectors and cyclical products with income and economic bottom recovery, and the right hand pays attention to the varieties with high growth certainty in the growth and technology sectors and whose valuation is wrongly killed by market sentiment at different stages, which can play a state of advance, attack and retreat. There is not much opportunity in the overall market beta, but there is a lot of operable space in the microstructure.
The fund is risky and investment needs to be cautious. The above is the actual record of guest interview Q & A, which only represents the personal views of the respondents, not today’s headlines and the views of Xinhua Finance and economics.