Harnessing GenAI: How Tradeweb is Unlocking Data to Enhance Liquidity Discovery
Carsten Just, Director, Automated Workflow Solutions, Tradeweb
What are the core components of a great deal? At the base level, moving the most volume at the lowest possible transaction cost is certainly important, but speed and scalability also play heavily in the equation. There’s also the relationships – the ‘X-factor’ that incorporates trading history, predictability, and first-hand knowledge of what the other side is hoping to accomplish. In essence it’s a liquidity play.
Now, thanks to the power of Generative Artificial Intelligence (GenAI), Tradeweb has made it possible to incorporate all of these core components – including the relationships and dealer compatibility – into an intelligent automated optimization solution for interest rate swap (IRS) trading.
Introducing AiSNAP
Revolutionizing the dealer selection process, AiSNAP bridges and unites historical trade data and GenAI algorithms to produce a list of dealers that are most likely to offer the best terms for a given trade. Essentially it improves liquidity discovery by matching inquiries to dealers that are not only wanting to price an enquiry, but also best placed to price. Unlocking the depth and quality of Tradeweb data, the AiSNAP tool analyzes dozens of different variables associated with each transaction, including the specific assets being traded, historical and real-time market data, the current market environment, past performance and more. This is stripping out the need for extensive human data analysis, research and assumption, which can be time-consuming. Traders are able to manually select dealers as they normally would, and with a single click of a button bring the total dealer count to 5, complementing original dealer selection with the AiSNAP dealers.
Playing on equal parts market psychology and data science, AiSNAP allows clients to optimize the dealer selection process based on all the available information, without having to disclose every detail of their own positions. First, the tool offers an integrated solution for an enhanced dealer panel. Secondly, the tool works in conjunction with Tradeweb’s traditional Request-for-Quote (RFQ) protocol and its newer Request-for-Market (RFM) protocol, which allows clients to ask dealers for a two-way market, rather than a price based on one direction. That means, even in cases where clients do not want to reveal the direction of a trade, they can still receive a list of dealers optimized for that transaction based on the underlying trade data being analyzed by AiSNAP. This makes the tool ideal for large trades where clients will benefit from complete market transparency without revealing their own trading intent.
Layering GenAI, data, and trader intuition
During the pilot phase, AiSNAP consistently created more liquidity and resulted in significant reductions in transaction costs. On average, over a sample of 15,000 transactions, the AiSNAP tool delivered estimated transaction cost savings of 0.055 basis points on GBP-denominated RFQ swaps transactions and 0.075 basis points for GBP-denominated RFM transactions under 25k delta. For larger trades above 25k delta, the tool generated even more significant estimated savings of 0.096 basis points for RFQ trades and 0.089 basis points for RFM trades.
Ultimately, what AiSNAP is really doing is breaking each transaction down into its core building blocks, aggregating all available information on each of those variables, and assimilating that data to recommend dealers who are in the best position to offer the best terms. It’s doing what seasoned traders have always done intuitively, but doing it based on petabytes of data processed in split seconds to take the guess work out of the equation.
This intuitive dealer selection methodology is yet another example of how Tradeweb is continually evolving and shaping our platform offering by developing new, innovative technology- and AI-driven solutions and strategies that make the lives of traders easier.
Related Content
Finding the Fastest Path from Idea Generation to Trade Execution