Why Fintech machine learning

Machine learning technology is of interest to world fintech companies and financial organizations whose business is somehow connected with investment, lending, consulting and security solutions. We at PayOnline , a company specializing in automating the acceptance of online payments, have decided to consider international fintech cases of machine learning technology.

Computers appeared in the 80s, and gradually we observed how their use for storing and processing information became the norm for most companies. In the 90s, we witnessed an Internet boom that truly changed the world. Collecting information about anything today is nothing. In the middle of the last decade, social networks appeared and entrepreneurs noticed that customers began to spend as much time in them as they hadn’t spent on any other site before. As a result, businessmen around the world began to invest in social media to increase the reach of the audience and for marketing purposes. When Android and iOS were introduced to the general public, a paradigm shift occurred. People began to spend more time with their smartphones than personal computers. Over time, consumers began to use smartphones for making decisions, making purchases and even payments. Today, realizing that smartphones have become an integral part of the consumer decision-making process, companies are striving to provide them with omnichannel interaction experience. In this regard, the question arises: "What other innovative tools exist that can change the market?" Companies should probably pay attention to the use of machine learning algorithms.

Let me illustrate the benefits of algorithms with a small example. Dash Financial , one of the leaders of the FINTECH market, recently won the Waters Technology Awards in the category “Best Algorithmic or Providing Direct Access to the Market Product for Investors”. Let's look at the functionality that the algorithms of this company offer. Dash developed a toolkit aimed at optimizing performance and reducing commission costs. The ability to customize the behavior of the system at any level and in any category is one of the distinguishing features of the product. In conjunction with the desire to provide full transparency in real time for all operations, the company's clients receive a full-fledged, investor-oriented trading infrastructure.

A set of Dash touch algorithms has its own customization wizard, a patented company development that allows you to adjust the parameters, behavior and techniques in accordance with the goals and needs of the investor without having to rewrite the program code or work through long development cycles. The results of any changes are instantly and clearly displayed in the control panel - a detailed, real-time web interface that allows access to each component of the application and execution process. Dash customers get the opportunity to gradually monitor where their applications are sent, as well as the state of the market and the depth of quotations up to a microsecond during their execution. This tool is a key element of algorithmic identification, analysis and further optimization and adaptation to the investor’s goals.
')
Over the past 10 years, companies have managed to collect a lot of data using a variety of channels and now it’s time to apply algorithms to this array of information. They will help companies to go beyond traditional reporting and to get deeper into the essence of the information collected. They will also be useful in analytical forecasting, helping companies make decisions instantly. Data collection, identification of patterns, smart classification and machine learning will change the situation in a variety of industries in the next five years. And the Fintech market will definitely not stand aside. Today, FINTECH segments such as stock trading and lending have already integrated algorithms into their activities to speed up decision making. It is safe to say that the application of algorithms by FINTECH companies has just begun and will definitely reach its highest point in a few years.

Below is a list of companies that use in their work an extensive and one of the most popular varieties of artificial intelligence today - machine learning.

18 fintech players using machine learning

Machine learning is a kind of artificial intelligence that provides computers with the opportunity to learn without any strictly programmed behavioral scenarios. Specialists in this field are engaged in the development of computer programs capable of learning independently, growing and changing based on new data. Young FINTECH companies use this technology to adapt to constantly changing market conditions.

Here are some of these companies:

Kensho : The company's developments combine the use of machine intelligence, work with natural languages, graphical user interfaces and secure cloud computing and represent a new class of analytical tools for investment professionals. Kensho's clever computer systems are able to answer difficult financial questions asked in plain English and, according to information on the company's official website, "are able to solve the most complex analytical tasks of our time." The company was founded by graduates of Harvard and MIT, it employs experienced former employees of Google, Apple and the US Federal Reserve, and among its investors are names such as Google Ventures, Goldman Sachs, In-Q-Tel (CIA venture division).

Affirm : A financial company whose technological tools collect huge amounts of data for effective use in estimating credit parameters. Machine learning is used to protect against fraud and credit data collection.

Lending Club : The world's largest online marketplace for borrowers and investors. The platform uses machine learning to predict potentially bad loans.

Kabbage : An online company from Atlanta specializing in financial technology and data collection. Kabbage offers direct financing for small businesses and consumers through an automated lending platform. The Kabbage team specializes in developing next-generation machine learning tools and a set of analytic tools for creating models for assessing credit risk and analyzing existing loan portfolios.

ZestFinance : The company uses machine learning techniques and large data analysis to make more accurate credit decisions. According to the company's website, the traditional approach to credit rating assessment uses only 50 parameters, which is only a small part of the number taken into account by the ZestFinance algorithms.

BillGuard : Provides personal financial security services. Billguard protects users from the theft of personal data during financial transactions, as well as erroneous write-offs and gray transactions. The company specializes in the use of such technologies as data mining, machine learning algorithms, security, and design of convenient web interfaces.

LendUp : The company specializes in microcredit, including allowing other organizations to provide similar services using their own API. LendUp uses machine learning and algorithms to accurately determine the 15% of borrowers who, according to statistics, will be able to most likely return loans.

Bloomberg : One of the leading providers of information for professional financial market participants. Bloomberg promptly provides accurate business and financial information, news and expert opinions from around the world. Using the methods of statistics, natural language processing and machine learning, the company offers analytical solutions for the financial community.

AlphaSense : A financial search engine that solves the fundamental problems of excess information and its fragmentation. The target audience of the company are professionals from various areas of the financial sector. AlphaSense effectively employs its own patented algorithms for processing natural languages ​​and machine learning, providing users with a powerful and highly differentiated product with an intuitive interface.

FinGenius : A platform focused on working with banks and insurance companies. A set of FinGenius technologies is a combination of various artificial intelligence techniques, including machine learning, processing natural languages ​​and modeling human logic in order to simplify the processing of arrays of complex data.

Dataminr : The company provides information retrieval services for clients from the financial sector. Dataminr tools in real time "comb" social networks and other open sources of information using machine learning algorithms in search of important information elements and their subsequent transformation into useful recommendations and practical advice.

Binatix : A stock broker using modern patented artificial intelligence algorithms that take into account the factors of human biological behavior and use them for large-scale data analysis.

Brighterion : The company offers one of the world's largest sets of artificial intelligence and machine learning technologies, capable of collecting and analyzing information from sources of any type, complexity and volume.

Feedzai : The company uses machine learning and working with "big data" to ensure the security of its clients' business. Feedzai self-learning models are able to recognize fraud 30% earlier than traditional methods.

Nymi (formerly Bionym) : The company has developed and is promoting a biometric authentication device using an electrocardiogram, using machine learning algorithms among other things.

EyeVerify : EyeVerify's proprietary software uses the so-called eye prints — vascular patterns in the eye proteins — as an identity identifier, also using machine learning technology.

BioCatch : A leading provider of behavioral biometrics, authentication and malware detection solutions for mobile and web applications. Banks and online stores use Biocatch to avoid conflicts associated with risky transactions and to protect users from cyber threats, such as seizing accounts, browser Trojans and attacks with remote access.

KFL Capital : The company specializes in the analysis of financial patterns and price fluctuations and the search for the most profitable ones. For these purposes, KFL Capital applies machine learning, forecasting algorithms, statistical methods and powerful computing computers.

MasterCard Machine Learning Network that broke ATM attacks

Separate attention is given to the news about the machine learning technology of the payment giant MasterCard, which made it possible to quickly take control of 3 separate cyber attacks aimed at the ATM network, limiting the total damage to about 100 thousand dollars in each of the cases.

The operations monitoring system, which also includes visualization tools, recorded three attacks that occurred, according to MasterCard, in January-February 2016.

The global Safety Net, launched last year, analyzes more than 1.3 billion transactions involving MasterCard debit and credit accounts, merchants and ATMs per day. To do this, the system uses algorithms for evaluating customer behavior in real time.


The 2013 cyber attack scheme, which began in one of the banks of the United Arab Emirates and in an American company that processed credit and debit card data.

The fraud detection system MasterCard Safety Net has recognized activity that spans more than 300 ATMs in 26 countries. Then within 11 hours, the criminals tried to withdraw more than 40 million dollars. ATMs are marked in red circles, attack intensity is marked in yellow.

In the three previously mentioned attacks against two US banks and one bank in South America, Safety Net identified anomalous behavior, such as withdrawing a large amount of cash or performing transactions outside of the geographic area familiar to those accounts. As Ajay Bhalla, the company's president for corporate security solutions, described the situation, MasterCard notified banks and rejected operations, limiting losses to less than $ 100,000 in each case. The company declined to name the banks to which the attack was directed, citing a non-disclosure agreement.

Opinions about the possibilities of machine learning in Fintech

Data analyst in a crowd expert project for Sentifi financial markets:

Among other things, machine learning is used to “crowd-collect” information on the state of financial markets. The idea is that the “crowd”, represented by experts from various fields of knowledge and by amateurs, can provide valuable information through this diversity of opinions. The goal of this approach is to get analysts what the general population thinks about certain companies and their actions, the stock market and other items related to finance.

Useful ideas are extracted from the "crowd consciousness" by large-scale "mining" of information from social networks, blogs and newspapers. Due to the natural disorder and lack of structure in the data thus obtained, specialists need to apply machine learning, processing natural languages ​​and image recognition to extract benefits from them.

Most often, for making decisions, investors use technical and fundamental types of analysis. Crowd-gathering technologies, on the other hand, add a third component — a social analysis of financial markets — and use all possible data to improve the decision-making process.

Moreover, this method makes it possible to democratize the information flow, since not only a group of individual experts (who are often not experts at all) often have a good understanding of the situation.

Fintech-entrepreneur specializing in roboconsulting and artificial intelligence:

Today, in the industry of robotic financial advisors, risk assessment and a client’s personal profile are based on a 10-point questionnaire with several possible answers. Now compare this approach with another, in which we use “big data” for image recognition, for example, recurrent neural networks to identify objects and compare them with averaged graphic images, or processing natural languages ​​to understand how people express their individuality texts on social networks. Imagine that all this will be used to compile personal characteristics of clients, allowing you to get a holistic view of their personal financial needs and goals. This is the potential of new technologies that attracts the industry so much and can be equally applied to opening accounts, CRM or financial planning.

IT Specialist Housing.com , behavioral finance:

How can the rise or fall of the market be the result of rumors? Why is there a “ January effect ” on the stock market? How can natural disasters and news affect the price movements of stocks? For answers to these questions, researchers turn to a scientific discipline called behavioral finance. It assumes that all individuals act irrationally, and anomalies are nothing but the result of the accumulation and addition of behavioral features of rational and irrational agents. Therefore, experts use Bayesian networks and other techniques to simulate decision-making processes and identify systematic errors.

Professor Kin Lam of the Baptist University of Hong Kong, together with a group of scientists, presented a pseudo-Bayesian quantitative approach to modeling investor behavior. Scientists have created a model that takes into account investor sentiment and compares their practical solutions with shock from changes in the stock market.

An Amazon.com employee specializing in programming and machine learning:

• Determining the credit rating of a borrower who wants to apply for a loan or mortgage based on his activity in social networks, history of financial transactions and other similar factors. Here, the most commonly used technologies are managed learning, such as neural networks.
  
  
• Portfolio optimization. Here, linear and non-linear programming, stochastic gradient descent, and genetic algorithms are used.
  
  
• Providing the most relevant answers to questions related to financial planning. It uses natural language processing and knowledge graphs.
  
  
• Automation of the process of redemption and placement of securities (the so-called underwriting)
  
  
• Automation systems for the classification and distribution of financial documents.

How algorithms will help companies win the competition:

Today, investors and borrowers from all regions of the world are looking for technologies that can make decisions in a fraction of a second, as the acceleration of this process guarantees victory in the race for primacy in the market. The use of artificial intelligence, machine learning and other similar technologies will allow companies to interpret the huge amounts of data accumulated over several years in a matter of seconds. Regardless of whether it is a matter of trading in stocks, lending or fraud detection, the recognition of complex patterns is one of the keys to success for any fintech company. To reduce risks, algorithms can be very useful for preliminary experimentation and evaluation of scenarios.

New algorithmic technologies are already here, and thanks to them, companies have been able to radically change their value propositions for customers. Fintech companies that have used algorithms in the near future have every chance of surpassing their competitors.

Continue to follow PayOnline processing blog updates and stay up-to-date on technologies that change financial markets.