It can monitor the outcome of those transactions and feed trade lifecycle information back into those models to help inform future deals. Such technology can also enable sell-side traders to open illiquid markets and accelerate voice trades, which still dominate the fixed-income space. And it can help firms automate their trading workflows by offering prices for even the most difficult RFQs.
Price Discovery
Establishing fair prices to offer a buy-side client can be difficult, especially in fixed income. With bonds sold over the counter, prices are not registered centrally on exchanges, making it tricky to assess a fair quote to offer a buy-side trade. Instead, prices are gauged by the last time that asset was traded or inferring a price from similar bonds.
That adds complexity because so few bonds are traded regularly. If a client requests a quote for a bond that hasn’t traded recently, the trader will need to find the most recent sale of a bond that most closely resembles it. In a global economy that encompasses millions of transactions each day, that’s an impossible task for a human to perform accurately on a consistent basis.
But not for AI. The power of the technology resides in its ability to crunch through enormous pools of data, identify trends and, more importantly, relationships within them and generate actionable insights from those computations. One of those pools it can be trained to manage is asset trade prices.
Cloud-based AI platforms have the computational power and economic scale to sift through hundreds of thousands of historical trades within seconds to identify the most recently traded assets with the closest match to that being sought.
More Accurate Pricing
The accuracy of that search is boosted by AI’s ability to use a broader set of parameters to identify more correlations between the proposed trade and previous transactions. In the past, traders would have narrowed their examination to a few simple characteristics of the requested bond. AI can apply a broader set of attributes to find more correlations, therefore offering traders the most accurate fit available.
The pricing benefits of AI-driven and data-led analytics can be most clearly demonstrated in illiquid markets. With so few bonds traded regularly, it has often been the case that RFQs for some trades have been missed or even gone unacknowledged because traders haven’t had the resources or capabilities to offer a quote in a timely way. The window for completing trades can be very short and, without the ability to swiftly find a fair and trusted offer, can close abruptly.
That no longer needs to be the case. AI analyses can produce quotes within seconds, ensuring that even the most thinly traded assets can be matched, bringing liquidity to some of the sparsest corners of the market.
Holistic View
While the last trade of a bond will largely determine the price of its next sale, this won’t be the only influence on its valuation. Intangible factors will also play a role, among them perceptions of an issuer’s creditworthiness. While credit ratings guide traders, technology enables further assessment through alternative data sets.
Natural language processing, a form of AI that can grab data from text and sound files as well as by trawling through social media posts, is increasingly being deployed to search for signals on how an asset or its issuer is regarded. These signals, particularly negative signals, can have an impact on the price of an asset. These insights can be fed into the pricing model to refine the eventual price offered to a client.
It’s tempting to think that digitalization aids only electronic trading. But the swift data processing that automation brings can be of benefit to voice traders too. For one, they get the benefit of faster prices on their screens. But while AI quotes can be shown to have closely predicted eventual trade prices in the past, dealers can also present quotes with the confidence that counterparties will have more trust in what they are being offered.
In the same way, fairly assessed AI prices can provide a more accurate basis from which traders can refine quotes offered to various tiers of clients. For the buyside client, knowing the price from the trader is based on data science can instil higher confidence in the price quality, ultimately increasing the likelihood of a successful trade.
Adding Value
The sell-side succeeds when it can seamlessly provide the buy-side with the assets and the services it needs. Automation is redrawing best practices for doing that, particularly when it comes to pricing, which is central to everything the sell side does.
By enriching pricing models with holistic data-led analyses, dealers can offer the best service to their customers while increasing value to their business.
Bloomberg
Source link