On May 19, 2017, IEEE Fellow, the chief scientist of Microsoft's artificial intelligence, revealed that he had left Microsoft and joined the US fund company Citadel as the Chief Artificial-Intelligence Officer.
In the field of financial investment, technology has been involved in the intervention of quantitative hedging. There are also many scientific and technological talents in computer science or the combination of finance and IT to enter the financial industry to promote the development of financial technology, but it is so important. An expert in the field of artificial intelligence has joined the giants of fund companies, and it is indeed a move that cannot be ignored in the transformation of AI and financial business.
Quantify the value and difficulty of hedging
The arrival of artificial intelligence is not a sudden technological change for financial investment. The development of investment has always been led by technology. For example, before the 1970s, the transaction was mainly conducted by telephone. The administrator judged through his own experience. Later, when the computer appeared, the ability to process and analyze the information accelerated, and bulk transactions began to appear, that is, hundreds of stocks were traded at a time. .
With the development of technology, we know that although the future is unpredictable, the risks can be predicted. As a result, hedge funds have emerged – through calculations, rigorous analysis, and large risk hedging, to achieve more robust benefits. This may not be obvious in China, but the practice of robots doing macro hedging has emerged abroad.
Finance is a better allocation of social resources. Many resources can be prioritized to become "resources". If you can know some valuable information faster than others, you can know the opportunity ahead of time and have a competitive advantage over others. Investment is a highly competitive industry, and whoever leads the way to make effective decisions can get a high return. The prerequisite for discovering these opportunities is more powerful computing and forecasting capabilities.
Therefore, not only Citadel, but also the world's leading hedge funds such as Man Group, Winton, and Aspect Capital are enriching their team of machine learning experts.
As far as Citadel is concerned, its housekeeping skills are high-frequency trading. For example, in general, when the company releases its earnings quarterly, if the performance exceeds analysts' expectations, the stock price will rise; if it is lower than expected, the stock price will fall. "So simple as a logic, what needs to be compared is that the organization can use the machine to interpret the financial information more quickly, making investment decisions in the first place. The machine interpretation of financial reports is a typical machine learning application, such as RNN network technology," Dr. Wang Wei, the co-founder of the investment, said. Dr. Wang Wei worked at Bloomberg and was responsible for quantitative transactions.
Therefore, based on high-frequency transaction quantification, the company's R&D system requires a lot of manpower and resources. It is necessary to hire professional mathematicians and computer experts, pay for expensive hardware, rent specialized microwave communication lines, and even build the system specifically for trading. There are all at the door.
However, the hardware keeps up, what about the software?
Even in the financial industry where stock data is extremely large, and machine learning has been used for some time, Wang Wei said to Lei Feng.com (Public No.: Lei Feng) AI Financial Review that there are still many "mistakes" in the machine:
For example, when FaceBook released its earnings last week, it released earnings reports under different accounting standards, namely, US GAAP and non-US GAAP earnings. The first line of the financial report confuses two different accounting standards, and judges that the mistake is considered to be lower than expected, so the FB stock is sold immediately; but the actual FB is up to expectations, after the wrong kill, the FB stock will be the next day. Go back.
“So, there is still a need to improve the accuracy and efficiency of machine learning.â€
At the same time, according to Lei Feng's AI financial review, Deng Li's research direction is mainly applied to deep learning and machine intelligence methods for big data, speech, text, image and multi-modal processing. Therefore, in combination, Deng Li The research may first break through the technical threshold in the module of machine interpretation of text data, improve quantitative investment, and let AI make a greater breakthrough for financial investment.
New opportunities brought by information + machines
As mentioned above, basically, the investment will conduct a performance survey based on the financial report of a company and pre-determine the stock price, but this is only the use of second-hand data; including the use of futures, bonds, fund trading information, etc. Predicting the trend of the investment market and quantifying the allocation of assets is more of a three-handed data.
Therefore, Dr. Zheng Yu, head of the urban computing field at Microsoft Research Asia, said in an intelligent financial training that the market should pay more attention to the performance of the results of the first-hand data before this link. “First-hand data actually affects the company's operations and financial reports, as well as the macroeconomic situation.†Being able to get information from first-hand data means getting closer to opportunities.
Jiang Long, the chief scientist of Tonglian Data, pointed out in his previous sharing that the information on big data products used for investment is mainly divided into these categories:
Information on the Internet – the Internet of Things, there are some cases in the world that monitor crop yields to predict crop yields and make large-scale investments. If you monitor the climate, moisture, etc. in real time, you can know the harvest of this year's agricultural products in advance, and then have an investment advantage for related products.
People's information - wearable devices, people and objects interact, where people go, what they see, what they think, this presents our society.
Environmental information - low-orbit satellites, low-orbit satellites are tested at a low altitude of 10,000 meters, not only can detect the production areas of agricultural products, but also detect the number of cruise ships parked in the port, parking lot parking in supermarkets, etc. You can know if it is in line with expectations and you can judge the future earnings.
For example, in the modeling process, the whales will add satellite remote sensing data to predict the changes in the output of agriculture, forestry, animal husbandry and fishery, and then make investment decisions. “For example, we can predict the cotton production in a certain place with a relatively accurate (85% rough measurement), and this action can be two months earlier than the official government figures, so we can consider investment decisions.â€
Using these alternative data to make a profit on investment, Citadel does not know this. It is also conceivable that the process of modeling based on the above data requires the machine to learn and analyze the satellite image and the geographical environment. A wider range of data, more massive data, longer data cycles... The opportunities brought by these new information resources are artificial intelligence, deep learning and glowing places.
At the same time, Wang Wei explained: "Go is doing well under the machine because the rules of Go are clear, the machine can have complete knowledge from the beginning to the end. The machine is still difficult to use in the financial field, because no one and no machine is at this stage. There can be complete knowledge, which is why the machine often makes mistakes that some people seem to be mentally handicapped."
“Therefore, getting more information than others, getting a more accurate interpretation, being able to respond more quickly to decisions, and avoiding 'low-level mistakes' on machines is a big challenge that needs to be addressed.â€
How to use the cutting-edge artificial intelligence academic breakthroughs to cope with these current opportunities and challenges - this is a new topic that Deng Li will face, and the next highland that Wall Street will compete for.
SSMB Cable Mount Connectors,SSMB Bulkhead Mount Connectors,SSMB Flange Mount Connectors,SSMB Pcb Mount Connectors
Xi'an KNT Scien-tech Co., Ltd , https://www.honorconnector.com