Commentaries and insightful analyses on the world of finance, technology and IT.

« July 2016 | Main | September 2016 »

August 17, 2016

The 'artificially intelligent' hedge fund rises

- By Sweenie Dabas and Mayur Bansal

Right now, artificial intelligence (AI) is certainly the buzzword in financial services. From robo-advisors (automated financial advisors and planners) and humanoid robots in branches, to chat-bots (chat robots for customer service) and machine learning algorithms for fraud detection and credit scoring, AI is inundating the world of banking, payments, and wealth management. And the big, secret world of hedge funds is actively exploring this technology as well.

The traditional systematic trading techniques used by hedge funds have to rely on human beings to develop the mathematical models to find trading opportunities. Now, AI is enabling the development of self-learning programs where the initial software application is created by humans, but after which it can develop itself, learn through real-time experience, and adjust its strategies accordingly. The shift from discretionary to systematic hedge funds is fast taking place and is visible in the number of hedge funds being launched. In 2014, more than 40 percent of new hedge funds launched were systematic funds, i.e., they utilized computer systems to select investments, which was the highest ever till that time. AI-based hedge funds have also justified their existence by outperforming average industry returns consistently since 2008, with 2012 being the only exceptional year in which they under-performed.

Artificial intelligence is being used for natural language processing (NLP), sentiment analysis, big data analysis, and deep learning to explore investment opportunities. In addition to processing vast amounts of financial data to extract hidden trends, AI techniques enable the analysis of non-financial data like news articles, tweets, pictures, videos, etc., and recognize patterns though human-like inferences. Advanced techniques such as Bayesian networks and evolutionary computation are also being used to build machine learning models for hedge funds.

Artificially intelligent systems are being explored and developed by well-established hedge-funds like Bridgewater Associates, Two Sigma Investments, and Renaissance Technologies, as well as new AI-focused firms including Sentient, Rebellion Research, and Aidyia, with the total managed capital running into trillions of dollars.

Some examples include:
1) Two Sigma manages assets worth US$35 billion and uses advanced AI technologies to discover new investment opportunities. One of the applications it uses is NLP to analyze the Federal Reserve (FED) minutes of meeting in order to gain insights into focus areas of the FED
2) Hong Kong-based Aidiyia is another flag bearer which has developed a trading robot inspired by genetics. Aidiya started trading in US equities in 2015
3) San Francisco-based startup, Sentient Technologies, has also been trading using AI-based systems since last year
4) Numerai is another AI-based hedge fund which has been developed by a community of anonymous data scientists. It makes use of a monthly tournament to get AI based-models from various data scientists to predict the stock market movements. Numerai received a funding of US$1.5 million from Renaissance Technologies

Since its inception, AI has witnessed contrasting hype cycles. Currently, however, the industry seems primed for disruption and innovation and AI's evident broad-based applications and benefits make it a potential game-changer. But it will not be devoid of challenges - AI requires not just skilled human resources, but also technology infrastructure that can work on extremely low latencies. It will also need extremely supportive reconciliation, reporting, and various other operation support applications. In addition, the regulatory side of use of AI in hedge funds remains hazy. But as AI becomes more mainstream, more robust regulations will surface that will require financial institutions (FIs) to develop stringent controls and risk assessment frameworks.

Like any other disruptive technology, there would be some AI techniques that fail or lead to unexpected results. But overall, it could lead to some big systemic changes in the hedge fund industry.

Risk management - Evolving challenges and models

- By Mayur Bansal and Ashima

The post-crisis era has witnessed a slew of compliance regulations in the finance industry. New products, increased government scrutiny, and a strong focus on compliance, has brought greater risks and a larger set of rules and regulations. Financial institutions (FIs) must now review their compliance practices and the technology infrastructure that supports them, and pursue a broad range of compliance and risk initiatives. According to predictions, the global IT expenditure by FIs on risk IT services and systems is expected to be US$70 billion in 2016. This expenditure has been mainly driven by investments in the areas of governance and systems, and process integration. However, expenditure in additional areas, such as compliance, stress test reporting, data aggregation, enterprise crime and fraud, are also important for the growth of risk IT spending.

Industry participants and regulators are focusing on managing risks due to various costs associated with enterprise fraud, money laundering, market and credit positions. In addition to US$12.4 billion in monetary fines (till 2014, as per CEB TowerGroup), there are various hidden associated costs such as fraud remediation and ongoing monitoring costs, suspension of licenses, opportunity costs, reputation damages, and higher risk premium costs. FIs face various challenges when attempting to improve their risk management practices. One big challenge involves identifying, developing, and adapting to new technologies. The other is adopting proactive risk management and compliance methodologies. Availability of good quality data, ever increasing pressure to maintain / increase profitability, reducing margin of error, and growing complexity in impact analysis, are key factors that make it necessary for FIs to understand risk with a fresh perspective.

Risk management approaches adopted by FIs are being enabled by technology in many ways. Their current approaches utilize advances in the fields of big data, analytics, and storage technologies to make risk management more futuristic and predictive as compared to the traditional and backward-looking methodologies. The approaches are being fine-tuned to make them more integrated and holistic for real-time analysis and reporting, as per regulatory requirements. Risk management in financial services is continuously evolving with the expansion of the role of the Chief Risk Officer (CRO) or Chief Finance Officer (CFO) being introduced to take care of capital utilization, cash management, and regulatory compliance. CROs have now become the first line of defense with the responsibility of overall risk management. In addition, it is now crucial to invest and innovate in risk platforms. For instance, innovations such as blockchain technology and crypto-currencies have removed various intermediaries in the transaction processes. The innovations must focus predominantly on risk reduction with efficient and predictive technology.

Looking ahead, the business models of FIs will need to be risk-based with a focus on just-in-time risk management and analysis. The technology infrastructure will need to be reinvented to bring improved controls and compliance while decreasing operational costs. Further, holistic risk management frameworks will need more investments, as a fragmented approach will adversely affect the complexity, efficiency, and sustainability of FIs and their systems.

August 2, 2016

Bank(ing) on big data - Banks or fintechs?

- By Chetna Narayanan and Prasanna Sekar

Today Data is omnipresent and is one of the most talked about topics across the table - anywhere and everywhere. But if you think that only the availability of data makes sense to create better offerings or to generate better value in the financial services (FS) industry, then the answer is a big 'NO'! Why? Because only the proper channelization of data will create value-based offerings for customers.

Though banks have huge amounts of data available to them, they, inadvertently, do not utilize this data to the extent they are supposed to in today's highly competitive marketplace. Banks have in their possession a significant amount data, which can be represented as the voice of the customer (VoC) on different social media platforms to understand customer needs and behavior. Unfortunately, they are not technically competent to utilize this data to create a more engaging customer service

Banks own and enjoy the advantage of data volumes, which comes in all shapes and sizes: horizontal, vertical, structured, semi-structured, and unstructured. Despite banks and financial institutions being such significant custodians of financial data, they do not use data points innovatively to create value-based offerings to customers as they are still struggling to be technically competent. However, they are powerhouses in terms of data collection and this is where fintechs firms like PayPal or TransferWise - are playing an active role in building customer-centric products and taking the financial innovation concept to banks

Among fintechs, big data analytics is one of their favorite roundtable discussions as they are utilizing and putting data into action by creating value for traditional banks through increased efficiency and cost reduction.

Banks provides plethora of customer data but it's of no use if it cannot be retrieved, validated, disseminated, and effectively converted into actionable insights. Today, fintechs are helping banks to disseminate the customer information so they can use it effectively to do deep dive analysis of consumer's financial behavior, enabling the delivery of much more personalized and comprehensive range of services.

For example, Barclays uses big data and Hadoop to come up with services called the Local Insights and Smart Spend. This is their attempt to stay technologically competent to understand and engage with their clients.

Fintech firm Trulioo has created a first-of-its-kind database for rural African population, which will help in searching for personal data across regions, which was not even previously considered by big banks.

Fintechs 1010Data and Xignite have created new, cloud-based data-sharing models, which will be useful for countries like Luxembourg, which is stringent on privacy regulations.

As fintechs continue to evolve their big data and analytical offerings, banks will establish and define each customer's experience according to individual's requirements, much before s/he can request a specific service. Though banks have started to adopt big data more and more over the last few years, there are still a lot of areas in which it could be leveraged and hopefully we will see more innovation in the use of big data techniques by banks.

Financial innovation is no longer a game of isolation. The owners of data (banks) and the kings of innovation (fintechs) are collaborating with each other and they need to as the former possesses the scale of data and the latter has the ability to convert that data into real-time actions - to take the FS industry to greater heights of customer innovation and deliver a truly engaging experience.