Artificial Intelligence and Finance

liu, tempo Date: 2021-07-20 15:30:08
Views:56 Reply:0

There are currently four mainstream gameplay in financial institutions:
1. Investment banks and sellers conduct research and try to automatically generate reports;
2. Financial intelligence search;
3. Public and private equity funds are assisting quantitative transactions through artificial intelligence;
4. Wealth management companies are exploring the direction of robo-advisors;

1. Investment banks and sellers conduct research and try to automatically generate reports

In the investment banking business and securities research business of investment banks, a large number of fixed-format reports are involved, such as some chapters in the prospectus, research reports, and investment intent. The writing of these reports requires a large number of junior investment bank employees to do long and tedious data listing, sorting, and repeated Copy-Paste work.

 financial institutions
Currently, automatic report generation mainly uses two techniques in natural language processing (NLP):

Natural Language Understanding (NLU): Digest and comprehend everyday speech, and transform it into a structure that can be processed by the machine;

Natural language generation (NLG): The structured data that has been split by the machine is expressed in natural sentences that people can understand.

We can think of these two technical understandings as the separation and processing of the raw material of daily conversations and the assembly into understandable natural sentences-the final product.

However, to actually generate a report, you need to use the above techniques to complete 3 steps:

1) Industry report

Processing massive amounts of heterogeneous data digests the annual reports that investment bank analysts need to read, real-time news and data from Bloomberg News, industry analysis reports, and legal notices. Among them, technical analysis such as OCR (Optical Character Recognition) is required for pictures and tables in the text.

2) Analyze the data

This process involves the use of commonly used knowledge extraction and entity associations in the knowledge map to extract its key logic backbone, combining factors such as the location of the event, and embedding key information into a pre-designed report template.

3) Article generation

After processing massive amounts of heterogeneous data and analyzing data, news can be produced, brokerage analysis and research reports, listing prospectuses, corporate annual reports, fixed increase announcements, and even investment proposals for fund researchers to hold daily morning meetings. It can be generated in a similar way. Users only need to select a template that meets their needs, determine the subject and key information, and report presentation form, and then generate basic content. Moreover, investment bank analysts can perform proofreading and manual secondary editing, add valuable opinions and conclusions, and improve the accuracy of the report.

Automatic report generation has been widely used in the news industry. The representative technology company Automated Insights invested by the Associated Press has automatically generated more than 1 billion articles and reports for the Associated Press. The French company Yseop can produce 3,000 pages of content per second and supports multiple languages such as English, French, and German. Its products are widely used in customer service departments of banks and telecommunications companies and financial news websites. But some technology companies are not just satisfied with providing automatic report generation services for the news industry.

Narrative Science was co-founded by the Department of Journalism and Computer Science at Northwestern University, aiming to automatically generate article reports through data analysis on a given topic. The company’s well-known data analysis platform Quill can analyze structured data, integrate artificial intelligence and big data technology, understand the importance of these data, and produce short textual statements or structured report content. Quill’s main object-oriented approach is financial service providers.
Frankel, CEO of Narrative Science said, “Our goal is to replace humans to do most of the basic work, and let machines process data and information.”

2. Financial search engine

When conducting research, brokerage/private equity fund researchers need to collect massive amounts of information, and then sort and analyze the content, such as upstream and downstream analysis, benchmarking company research, competitor research, corporate highlights/risk points analysis, and so on.

However, at present, the auxiliary research software used by most securities analysts, such as Bloomberg data terminal, only solves the basic data problem without considering the problem of information overload. This makes it impossible for researchers to find the most accurate and valuable information when faced with a large amount of basic data and explosive information, nor can they improve their work efficiency.

The core technology behind the financial search engine is a high-quality knowledge graph and a large number of business rules to help realize association, attribute search, and short-range relationship discovery. Exploration engines, such as faceted browsers, also provide a human-computer collaboration interface based on the knowledge graph, so that the process of data exploration can be easily recorded, iterated, and reused. In addition, the recommendation system and push system are also very useful, helping financial users focus on key data, and save time and effort for pre-investment discovery and post-investment tracking.

Among them, semantic search is to provide different types of queries (such as companies, funds, events, etc.), such as the impact of the Chilean earthquake on copper futures, and the impact of the Middle East crisis on the overall currency market. Then the information is sliced and then aggregated to provide visual elements for overview, such as the average market value of fixed increase in film and television media and the financing price-earnings ratio.

3. Public and private equity funds are using artificial intelligence to assist quantitative transactions

Semantic search leaves complex queries to users to complete, such as looking for upstream companies in VR. When the search fails to provide accurate upstream information, companies with cameras will be recommended to users, and a convenient interactive interface will be provided to users to perform some complex tasks of filtering. Alphasense is such a financial search engine that is lightweight at the data level, and leaves complex logical judgments to users to complete, focusing on solving professional information acquisition and fragmentation problems. Alphasense is oriented to the financial investment field. It gathers all investment information from documents/news and research and conducts semantic analysis, and conducts trend analysis in global company data. Its mission and vision is to find valuable information from a lot of noise and focus on the basic issues of information richness and fragmentation, thereby greatly improving the work efficiency of financial professionals and saving working time.

Quantitative trading has been using machines for auxiliary work from a long time ago. Analysts write simple functions, design some indicators, and observe data distribution, and these only use the machine as an arithmetic unit. Until the rise of machine learning in recent years, data can be quickly and massively analyzed, fitted, and predicted. People have gradually connected artificial intelligence and quantitative trading more closely. It can even be said that the three sub-fields of artificial intelligence (machine learning, natural Language processing, knowledge graph) run through quantitative trading.

Machine learning: Quantitative trading analysts model financial and trading data from digital speculation models, analyze their salient features, and use traditional machine learning algorithms such as regression analysis to predict trading strategies. There are two main disadvantages of this approach. One is that the data is not rich enough and is limited to transaction data. More importantly, it is limited to the selection and combination of features (Feature Engineering). The quality of the model depends on the analyst’s understanding of the data sensitivity. Another approach is to imitate the behavior of experts, select specific experts in a certain field, copy their decision-making process, and import a repeatable calculation framework.

Bridgewater Asspcoates, the world’s largest hedge fund, opened a new artificial intelligence team as early as 2013. The team has about six employees and is led by David Ferrucci, who once worked for IBM and developed Watson, a cognitive computing system. According to Bloomberg News, the team will design trading algorithms to predict the future through historical data and statistical probabilities. The program will change as the market changes, constantly adapting to new information, rather than following static instructions. The founder of Bridgewater Fund has also stated publicly that its funds hold a large number of long and short positions, invest in 120 markets, hold more than 100 portfolios, and considered investment portfolios using artificial intelligence. Rebellion Research is a quantitative asset management company that uses machine learning to make global equity investments. Rebellion Research launched the first pure artificial intelligence (AI) investment fund in 2007. The company’s trading system is based on Bayesian machine learning, combined with predictive algorithms, and constantly evolves in response to new information and historical experience. It uses artificial intelligence to predict stock fluctuations and their relationships to create a balanced portfolio risk and expected return. , Using the rigor of the machine to surpass the trap of human emotions, and effectively complete the transactions on stocks, bonds, commodities and foreign exchange in 44 countries around the world through self-learning.

Castilium, a hedge fund institution in London, was founded by big financial players and computer scientists, including former Deutsche Bank derivatives experts, former chairman and CEO of Citigroup, and professors from MIT. They interviewed a large number of traders and fund managers, copied the reasoning and decision-making processes of analysts, traders and risk managers, and incorporated them into the algorithm. Artificial intelligence startups in quantitative trading include Japan’s Alpaca. Its trading platform Capitalico uses deep learning technology based on image recognition to allow users to easily find foreign exchange trading charts from the archive and help with analysis. In this way, Ordinary people can know how celebrity traders do transactions, learn from their experience and make more accurate transactions. At the same time, Alpaca also launched AlpacaScan as a K-line chart tool for real-time reflection of the U.S. stock market, abandoning the limitations of binary filtering to provide traders with a daily necessary tool to identify potential market trends. Aidyia, located in Hong Kong, is committed to using artificial intelligence to analyze the U.S. stock market. It relies on a mixture of multiple AIs, including genetic evolution and probabilistic logic. The system will analyze the market and macroeconomic data, and will do it later. Develop your own market forecasts and vote on the best actions. Similar companies include Point72 Asset, Renaissance Technologies, and Two Sigma.

4. Smart Investment Advisor

Traditional investment advisors need to stand from the perspective of investors to help investors manage their portfolios that are in line with their risk appetite characteristics and market performance in a particular period. These tasks need to be completed in a large amount of expensive manual methods, so wealth management services have invisibly raised the barriers to entry and are only open to high-net-worth individuals.

But now, intelligent investment advisors (robot advisors) are carrying out portfolio management with minimal human intervention. You can manage your assets by a row of computers, and you don’t have to be a high-net-worth individual. And robo-advisors are using more powerful computer models to use artificial intelligence technology to profile a large number of customers, and use artificial intelligence algorithms to provide tailor-made asset management investment solutions for each customer.

Wealthfront is a very representative robo-advisory platform. With the help of machines and quantitative technology, it provides tailor-made asset investment portfolio recommendations for customers evaluated by questionnaires, including stock allocation, stock option operations, debt allocation, and real estate assets. The configuration aims to provide an automated investment management service to maximize the return on investment. There are 5 steps in Wealthfront’s automated investment management:

Determine the ideal asset class for the current investment environment

Represent each asset class with the lowest cost ETF (trading open-end index fund)

Determine risk tolerance and create a suitable investment portfolio

Diversify risks with modern portfolio theory (MPT)

Regularly monitor and rebalance the investment portfolio

This investment method has also been affirmed by the market. Wealthfront’s management fund scale has increased by nearly 64% from 2015 to the end of 2016. As of the end of February 2016, Wealthfront’s asset management scale has reached nearly US$3 billion.

Behind the recognition of the market is the confidence in robo-advisors. Robo-advisors can overcome human nature, prevent investors from being irrationally emotionally affected by market changes, and make machines strictly implement pre-set strategies. And robo-advisors have more transparent and open information disclosure than traditional wealth management institutions and private banks, and provide timely risk warnings, which greatly reduces the information communication barriers between asset custodians and managers.
Betterment is also a financial technology company focusing on intelligent investment management. It uses Markowitz asset portfolio theory and various financial derivative models to apply to products. It solves various data operations in the cloud at low cost, quickly and in batches, and then according to users Tendency and set risk preferences, personalized asset allocation portfolio plan. Its founder, Jon Stein, worked as a senior investment consultant in a financial institution on Wall Street, and is committed to building Betterment into a robo-advisor that makes investment more convenient and accurate. In March 2016, Betterment received USD 100 million in Series E financing. And FutureAdvisor, founded by two former Microsoft employees, is a robo-advisor company that focuses on the pension wealth management market. FutureAdvisor serves customers who have many different financial accounts, pensions, savings, stocks, and even some CDs or bonds but do not know how to make the right choice. FutureAdvisor uses intelligent algorithms to monitor wealth management accounts in real time, find tax-saving opportunities and adjust multiple accounts. In addition to providing free portfolio optimization and same-source integration of investment data, FutureAdvisor also provides a paid version of investment agency services. Currently FutureAdvisor is acquired by BlackRock, the world’s largest fund management company, with a valuation of US$200 million. (Schwab intelligent portfolios portfolio income chart)

In the face of the unpredictable financial market, the intelligent investment advisory product schwab intelligent portfolios launched by Charles Schwab (Charles Schwab) can use Monte Carlo to simulate the performance of the portfolio in the dynamic market for post-investment tracking. At the same time, when the investment portfolio is losing money, the machine will automatically harvest tax losses, which means that a portion of the capital gains tax will be reduced on the loss of securities sold. When the investment portfolio deviates from the preset risk tolerance and asset allocation construction, the machine will automatically adjust the balance of assets through a series of buying and selling behaviors.

Robo-advisors make artificial intelligence technology no longer far away from the crowd, and truly enable every ordinary person to enjoy the benefits of smart financial technology companies, and also make many people who once thought “artificial intelligence is out of reach” realize Smart finance companies not only serve professionals in the financial industry, but can also create value for the entire business community.

When artificial intelligence is no longer a new thing, the four mainstream gameplay methods of combining artificial intelligence and finance, namely, investment banking and securities research and automatically generate reports, artificial intelligence-assisted quantitative trading, financial search engine securities research, and intelligent investment advisor wealth management. Seeing that in the future, the combination of finance and artificial intelligence will become infinite possibilities for smart finance.

And smart finance is providing a large number of auxiliary decision-making tools in a way of combining man and machine, so that investors can more easily obtain data and analysis support in the process of forming a logical chain, so that they can use more energy to discover The machine is not good at completing the work, thus greatly improving the work efficiency.

Leave a comment

You must Register or Login to post a comment.