Stock trading in AI era Part1

From the discussion such as” Legal issues of online securities business “(“Shinkyo reports 5 (3), 17 – 31, 2001 – 08) , securities transactions have completely changed their appearance for 15 years. it will be one of the most changed fields by networking society.

Online transactions have become generalized and the securities market has become fragmented (a dark pool is developing, which is a place of business where a large number of markets are established and a private trading system is also opened and transaction information that does not disclose sign information is made) . In addition, autonomous algorithmic trading is developing very closely to the majority of the market.

Algorithmic trading can generally be defined as referring to transactions that decide the timing and quantities of stock trading and make orders, according to the program formulated in advance. Among them, the whole transaction made at high speed and high frequency is called high frequency transaction (HFT). Institutional investors and others are conducting such transactions and are beginning to show an extremely large presence in actual securities transactions.

It is said that such high-frequency trading has merits such as efficiency of price finding function and liquidity. On the other hand, it seems to be pointed out that there are various problems. IOSCO’s “Regulatory Issues Raised by the Impact of Technological Changes on Market Integrity and Efficiency” (outline version in Japanese) are risks of market efficiency, risks on market fairness and robustness, We describe risks to market elasticity and stability. Besides these, it is pointed out that issues regarding best execution, narrowing the scope of market participants, program operation by market, regulation cost problems, etc. are pointed out.

By the way, let us consider this algorithm transaction and the current “Stock trading using AI”.

In the first place, AI is a too generic name, it is a concrete technique, and it itself can not be said to lead something productive conclusions. Among my friends, AI is a childhood name and talks about “AI grown up theory” that when AI grown up  in “the formal clothes”, AI will be  given “Grown Up” name such as natural language analysis or image recognition among the university professors, there are many people who surprisingly support.

Aside from that, when discussing with AI in the form of stock trading, I think that it is possible to think that there are two directions. One is to use artificial intelligence to advise customers on asset management and the other is to make it possible to use the results of analysis by artificial intelligence regarding the behavior of securities trading algorithms .

Regarding the former, it is called a so-called robot adviser / service, and its legal issues are also being studied (Hiroyuki Hasegawa “FinTech and its legal issues in the securities field” NBL 1081, page 71 (Commercial Legal Affairs, 2016)).

Regarding the latter, in terms of using analysis results by artificial intelligence, specifically speaking, from the point of view of what artificial intelligence technology I can do it.

As being published in business
Stock price rise prediction system
A system that analyzes time series stock price data by RNN (recurrent neural network)
And others have been announced.

Also, academically,
Application to momentum trading strategy
Application of event-based stock price prediction using natural language analysis
We will be in the category of this AI technology such as.

Let’s examine in detail what kind of thing can be mentioned as a specific legal problem concerning the application of securities trading of such AI technology.