Special interviews
Written by: Zhai Ziqian
2021-04-07 07:00
Last update date: 2021-04-07 07:00
In the classic drama "The Great Era", "the way of stocks lies in people forsaking me and taking them" has become a golden phrase.
However, the stock market is changing rapidly, and it is difficult for the human brain to deal with the thousands of stocks in the market alone. Perhaps the only "winning formula" lies in Algorithmic Trading (Algo Trade, program trading).
Algo Trade has become an indispensable tool in the financial market. Institutional investors almost use it.
With all kinds of algorithms fighting against each other, it is even more powerless and insignificant to rely solely on the human brain to analyze transactions.
"Hong Kong 01" recently interviewed Zeng Qibang, the co-founder of Finbot and the director of Ming Capital, and he shared his experience in developing Algo Trade. It turns out that program trading is not just a large-scale patent.
"Post-80s" Zeng Qibang graduated from Hong Kong University of Science and Technology in 2010, studying Business Administration (Finance).
The first job was to engage in index arbitrage trading at Societe Generale. The frequency of trading is as high as one thousandth of a second, which is of the nature of "high-frequency trading" and requires computer assistance.
"It was the first time I was exposed to using a computer to trade." Working at Societe Generale, Zeng Qibang understood that many stock trading strategies and routines require precise calculations from a computer.
In order to keep improving, Zeng Qibang went to the University of Hong Kong for further studies, continued to study the knowledge of pricing, and taught himself machine learning (Machine Learning), trying to combine the two aspects of professionalism.
After that, Zeng Qibang joined Shanghai Commercial Bank and was responsible for the "settlement" work, that is, the bank paid the bills, and he mainly managed the portfolio.
At that time, the main force relied on the "human brain" to trade, but Zeng Qibang himself used a computerized method to calculate, and that year brought 57% of the investment portfolio return.
Zeng Qibang, co-founder of Finbot and director of Fengming Capital.
(Photo by Luo Guohui)
Leave the investment bank to research Algo Trade
At that time, Zeng Qibang had an idea to automate this trading strategy, from automatic order placement to the establishment of a Robo Advisor.
Therefore, he resigned resolutely in 2018 and practiced his dream to establish Fangbo Technology.
The algorithm (Algo) that can make Zeng Qibang give up his high-paying job in investment banking has deeply fascinated him. It can learn automatically and find common ground from victory.
Fangbo Technology was established in 2018. Zeng Qibang initially invested more than 2 million yuan in funds, half of which was used to buy historical data.
"Investment is almost a few years of income at the time." The company currently has a total of 4 people in operation. "Everyone is doing it at Part-time. If Full-Time does it well."
He said frankly that the company has reached a balance of revenue and expenditure, and even reached a "negative cost" because it has received funding from the Science and Technology Park and it also has private fund sponsorship.
Fangbo Technology's AI platform services include Robo Advisor.
"Reverse the effect as the cause" predict the future from the trace
I believe a lot of readers are more interested in how Algo Trade "makes money".
Robo Advisor, researched by Zeng Qibang, uses the "reverse effect as a cause" method to find short-term buying opportunities.
"If it rises by 7% within 15 days, this is a "effect", so I have to go back to the previous "cause"."
Zeng Qibang invested more than one million yuan to buy the data over the past 10 years. It is precisely to inject the data into Algo, so that it can constantly calculate according to the required parameters, and then find the "victory equation."
"It must be 10 antennas to rise through 20 antennas and you must buy." On the contrary, it has repeatedly studied the stocks that can rise by 7% in the past 15 days. "Whether there is a pattern before the rise?" And Robo Advisor set 7% mainly because Since the short-term volatility of each main board Hong Kong stock in one month is about 8% to 10%, after the buy signal appears, the 7% gain has roughly earned most of the volatility; and the other reason is that it is convenient for users, and there is no need to respond one by one. Individual stocks have been changed.
Zeng Qibang bluntly said that most people who develop Algo Trade will not share it with others, but he hopes that more people will be able to contact it.
(Photo by Luo Guohui)
Algo Trade is a non-major account patent "I hope more people can use it"
Currently, Robo Advisor developed by Zeng Qibang is available for investors to pay for subscription. The monthly fee is 1,328 yuan. The computer will automatically send out buy or sell signals after the market closes every day.
Zeng Qibang said frankly that very few individuals currently do Algo Trade in the market, and most of them are institutional investors, because it requires a lot of data and requires a team or talent who is very familiar with computer science to be able to do it.
"Most of the research and development are for my own use, don’t talk to others." Therefore, one of the reasons for his self-developed Robo Advisor is that he hopes that more people will have access to this area of expertise. Principles, but they can still be used or used as a reference.
Virtual banks are gradually developing in Hong Kong, and Zeng Qibang hopes to cooperate with Virtual Bank to develop Algo Trade in the future.
(Profile picture)
Hope to cooperate with Xuyin to develop towards Algo Trade
When it comes to the development of the company and even the entire Algo Trade industry, Zeng Qibang said frankly that one thing he hopes to be able to do in the future is to cooperate with virtual banks.
He pointed out that Robo Advisor currently mainly provides investment advice, and only requires the Securities Regulatory Commission to approve the "Plate 4" for securities consulting; and the "Plate 9" is to manage investment portfolios for clients in the form of discretionary authorization.
He believes that one of the future development directions of virtual banks is to establish a virtual asset manager (Virtual Asset Manager) through Robo Advisor similar to his research and development to provide investors with direct buy or sell signals. Users only need to decide "buy Don’t you buy?” That’s it.
In the field of program trading, Zeng Qibang believes that there is a shortage of relevant talents in Hong Kong.
"There are so few talents in Hong Kong. The interviews are mostly from Southeast Asia, Singapore, and South Korea, but they all graduated from Data Science/Machine Learning at the University of Hong Kong. They have a degree, but they don’t know whether they are from Hong Kong." Zeng Qibang estimates, Hong Kong Most of the students tend to develop in financial real estate, believing that the financial industry can quickly make a fortune.
He said frankly that local companies of this type do not lack funds to hire talents, but there are not many suitable options. He suggested that if the government intends to promote the development of the industry, it should provide students with tuition subsidies for relevant subjects to attract more talents into the industry.
artificial intelligence
