When It Comes To AI, Capital Markets Has Barely Scratched the Surface

This image has an empty alt attribute; its file name is logo-cropped-11-news-300x66.png

With the current market uncertainty, many companies are looking into AI to create efficiencies and save on costs. However, one particular area of the fixed-income market is not leveraging the vast potential for AI and data science-corporate bonds. Through utilizing AI, corporate bonds market could benefit from increased liquidity, greater capital efficiency, enhanced risk management, and higher capital velocity.


Source: Traders Magazine

The current uncertainty in the market is pushing companies to rethink their technology stack and look for opportunities that create efficiencies and save on cost, such as Artificial Intelligence. Artificial Intelligence (AI) has evolved rapidly over the past few years and has achieved ‘prime time’ in applications like Netflix, Siri and Alexa. It’s rapidly demonstrating its value in many other industries including financial services, healthcare and manufacturing.

In fact, one market research firm forecasts that AI software will create $2.9 trillion of business value in 2021, a figure that is similar in size to the UK’s annual income. The revenue generator of the AI age is brilliant software that gleans insights from the world’s fast-growing mountains of data. Humans will generate an estimated 50+ zettabytes in 2020 alone, which is remarkable given that we only entered the zettabyte era in 2010.

Capital markets is the most data intensive segment of the financial industry, and one of the largest spenders on AI technology. Several firms are leveraging AI to generate actionable insights out of the avalanches of data generated by a range of processes, thereby increasing efficiencies and lowering costs. For example, firms are adopting machine learning models for credit scoring and risk management while using algorithms to trade securities. But these applications may just be scratching the surface of AI’s capabilities for capital markets firms.

Take the global fixed-income markets, which represent the largest subset of financial markets in terms of issuances and market capitalization. The Institute of International Finance reports that the size of the global debt market – which includes publicly traded securities, such as commercial paper, notes, bonds, and non-publicly traded loans — surpassed $250 trillion last year ─ which is a staggeringly large pool of capital.

The global debt markets are also complex, not least because of the multifaceted nature of securities: Every product has its own maturity, redeemability, and convertibility. This scale and level of complexity enables AI to unlock unique insights that can solve problems that were historically intractable.

One particular area of the fixed-income market that holds vast potential for AI and data science–corporate bonds. It is no secret that the corporate bond market is hugely complex, inefficient and illiquid. In fact, according to the BIS, around 75% of US high-yield corporate bonds are not traded electronically, which can be compared to FX options, forwards, precious metals, cash equities and futures markets where 60-90% of products are traded electronically.

The corporate bond market’s byzantine nature is precisely the reason why it has been slow to adopt electronic trading methods. Size, complexity and opacity make it difficult for dealers to quickly identify natural counterparties for their bonds, and hence execute a trade. The situation has worsened since the global financial crisis of 2008, with dealers reporting that it has become extra challenging to manage risk and temporarily hold bonds while they search for counterparties. Recent credit market turbulence has brought the spotlight back on these long-term issues.

We know that AI can kill two birds with one stone for dealers. Software can trawl through historical trade data to generate insights that will identify the natural buyers and sellers of securities. This will help bring liquidity to the corporate bond market, and also boost the share of products that are traded electronically. For example, a dealer might find few potential counterparties for a $100 million trade of high-yield bonds, but algorithms could identify likely counterparties who might buy $5-$10 million chunks. The algorithms could reshuffle their predictions and rankings when variables such as the security’s price or volume changed.

Broker-dealers and asset managers could execute trades through protocols that leverage this AI to dramatically increase the probability of getting ’difficult’ trades done at better pricing for their customers, while minimizing information leakage.  This will not only boost liquidity in the corporate bond market as a whole, but also offer dealers greater capital efficiency, enhanced risk management and higher capital velocity. We see the selective application of AI as ultimately creating mutually beneficial outcomes for broker-dealers and their customers.