AI in Finance: New Applications for High ROI in 2020

Note: This article was originally published on October 16, 2017 and has been migrated from our previous blog. Some details — tools, libraries, benchmarks, industry context — may be outdated. For our latest perspective, see our recent posts.

Artificial intelligence (AI) in finance is taking the industry by storm. According to the 2020 Business Insider Intelligence report, 75% of respondents at banks with over $100 billion in assets say they’re currently implementing AI strategies, compared with 46% at banks with less than $100 billion in assets. The size of the cost savings opportunity is $199B for front office (conversational banking), and as much as $217B for middle office (anti-fraud). According to McKinsey Global Institute, applying AI technologies in banking could generate more than $250 billion in value across the sector. Considering the significant savings opportunities, the number of companies implementing AI will be growing. In the next decade, AI will help financial services companies maximize resources, decrease risk, and generate more revenue, in trading, investing, banking, lending, and fraud detection verticals.

Use Cases

• Trading
• Investing
• Banking
• Lending
• Fraud Detection

Trading: Better trading through algorithms

AI in finance can help process data and create the algorithms managing trading rules. It is especially useful in quantitative trading. AI-powered devices analyze large and complex data faster and more efficiently than people, which leads to saving time and money.

Case Study: KAVOUT

• Problem: Spending vast amounts of money and time on analyzing tons of data.
• Solution: With machine learning, Kavout can quickly process large amounts of unstructured data and generate signals and identify real-time patterns in financial markets. Their solution K Score analyzes data such as RSI, SEC filings, or trading volume data and then condenses the information into a numerical rank for stocks. Applying deep learning methods allows for capturing nonlinear relationships and correlations. The client gets all the essential data instantly.
• ROI: According to Kavout, for as low as 0.02% of fund profit, the annual subscription gives managers daily data-feed and an estimated alpha of 4.84%.

Investing: Fintech companies offer investment insights

In the wealth management arena, B2C robo-advisors augment portfolio management and rebalancing decisions made by humans, often analyzing a person’s portfolio, risk tolerance, and previous investment decisions to offer advice.

Case Study: KENSHO

• Problem: Spending hours on researching answers to financial questions.
• Solution: With its AI-platform, Kensho allows you to answer the question, “What happens to a given set of assets when a certain set of conditions is true?”. For example, you can ask the system what happens with Apple stock if Apple launches a new iPhone or what happens to South African mining stocks immediately after the resolution of labor strikes. These questions are relatively simple, but finding an exact answer may often take a lot of time. Kensho’s platform analyses historical patterns and gives you a quick response, even to complex financial questions, in plain English.
• ROI: According to Forbes, Kensho helped investors predict an extended drop in the British pound in the days before Brexit.

Banking: AI in finance enhances efficiency, offers data insights, and manages risk

Chatbots help banks serve customers more efficiently, even though they aren’t advanced enough to handle support cases autonomously. Powered by natural language processing, bots can listen in on agents’ calls, provide accurate answers quickly, and suggest best practice answers to improve sales effectiveness. Neural networks help agents respond to common customer service queries by sorting and labeling metadata and generating three potential responses, each with a level of certainty attached.

Case Study: Kasisto

• Problem: High bank customer care costs
• Solution: Kasisto’s conversational AI platform KAI was created to improve customer experiences in the financial industry, and to reduce call center volume and customer care costs by filtering and triaging customer inquiries. The AI-powered chatbot provides customers with self-service solutions, predicts their financial needs, and makes calculated recommendations.
• ROI: Bots handling 82% of customer requests and inquiries without a live agent’s involvement.

Lending: AI for credit lending

Machine Learning is a game-changing technology for lenders, lowering compliance and regulatory costs, and helping with robust credit scoring and lending applications. Credit decision-makers can use AI in finance to achieve faster, more accurate risk assessment, using machine intelligence to factor in the character and capacity of applicants.

Case Study: Zest AI

• Problem: A lot of potential borrowers look good on paper and are classified as creditworthy, but are in fact risky. On the other hand, a lot of those who could turn out to be creditworthy get rejected by the traditional credit model.
• Solution: Zest AI provides transparency which helps lender assess borrowers with little credit history. Their model assesses hundreds of applicant data points, up to 10 times more than the client’s credit model had used before. For example, a history of discount-store shopping will boost an applicant’s chances of getting a personal loan, while an applicant writing the full legal name of an employer on a loan application will lower it. Inclusion of more data and more sophisticated math reduced default rates, setting the stage for potentially tens of millions of dollars in credit savings.
• ROI: According to Zest AI auto lenders using ML underwriting cut losses by 23% annually, more accurately predicted risk, and reduced losses by more than 25%.

Fraud Detection

According to KPMG’s Global Banking Fraud Survey, cyber-related fraud risk is the most significant challenge for banks today. Over half of the respondents of the survey said they recover less than 25% of fraud losses, which demonstrates that fraud prevention is key. On the other hand, a common problem for banks is false positive, so when a legitimate transaction is for some reason marked as suspicious, and the user is incorrectly identified as a fraudster. This leads to shutting down payments or even locking accounts completely. Kount, a digital fraud prevention company, reported a cost of $2 billion for e-commerce merchants in the USA alone due to false positives. According to Forbes, a study found that 40% of consumers in Europe say they won’t do business again with a merchant which declined the card when it was a legitimate purchase.

Case Study: Vectra

• Problem: Traditional signature-based firewalls, IDS and IPS miss advanced threats and cannot distinguish between benign anomalous behavior and high-risk attacker behavior.
• Solution: Vectra’s AI-driven Cognito cyberattack-detection and threat-hunting platform enable banks to identify threats in real-time, automatically triage alerts, and respond quickly to hidden attackers in data center workloads and user and IoT devices. Combining human expertise with a large set of data science, machine learning techniques, and behavioral analytics, Cognito automates manual, time-consuming threat hunting, and response.
• ROI: Cognito condenses days and weeks of threat hunting into minutes, which reduces the security operations workload by 36 times.

How to get started with automation for AI in finance?

Picking the right solution provider is essential for your AI in finance journey. At Sigmoidal we have implemented numerous solutions with our financial industry partners, with high RoI, for example:

Investment opportunity discovery

Discovering (sourcing) investment opportunities is key to make sound investment decisions, and the most precious opportunities are often hidden. Our client aimed at expanding their investment horizon without spending vast resources on information sourcing and research. Before leveraging the AI in finance solution, they employed a team of research analysts looking through thousands of web articles, tweets, and social media posts, trying to find information on recent changes in companies’ structures. Our solution automated that work by scraping all the articles and tweets, providing the client only with the essential pieces he was looking for, all in one place. Moreover, we extracted the vital information, using named entity extraction, so for each piece, the client would have the name, position, and the reason, the article is valuable to them. Then, our algorithms would put all the information into an executive summary. This solution, apart from preventing our client from spending a vast amount on the research team, gives them all the information instantly.

Learn more here: Use Case – Investment Opportunities

Choosing the most profitable portfolio

Another client needed help with his investment strategy. We designed a system, which predicted the expected return for every asset in a portfolio.

Learn more here: Use Case – Portfolio Analysis