The best artificial intelligence technology provider category is three years old. The first two winners were Sentieo and AlphaSense, both of which are startup specialists. This year’s winner, however, is nearly two decades old: MarketAxess. And its win shows that the field of AI is spreading well beyond the buzzy startup community and filtering into the traditional capital markets.
MarketAxess won this year’s award thanks to the incorporation of machine learning for its Composite+ algorithmic pricing engine for corporate bonds. The research platform allows for pre-trade price discovery, liquidity provision, transaction-cost analysis, auto-execution, and crossing. The automated engine algorithmically generates prices for over 24,000 corporate bonds from public Trace data and MarketAxess’s proprietary TRAXdata platform. Composite+ is updated every 15 to 60 seconds and generates nearly 20 million levels per day.
Composite+ uses a machine-learning technique known as a gradient boosting model to help produce more accurate prices in the corporate bond market. MarketAxess was sitting on a treasure trove of good data but it was too complex for a human to create a hand-built set of rules to aggregate, analyze and make sense of the data in a meaningful way, says David Krein, head of research for MarketAxess.
“If the trade occurred on MarketAxess, we have insight as to what actually transpires and the algo can effectively navigate this new insight and make sense of it,” he says.
For example, if a trade is struck at 96 bps, that could be one client asking one dealer for a price, or it could be one client asking 10 or 20 dealers. Composite+ would give the former a lower weight and the latter a higher weight, because it can piece together the context around the trade. “The AI can make sense of the data and value those inputs quite a bit differently,” Krein says.
Julie Lerner joins to talk about the hemp market and PanXchange's launch of a hemp exchange.Subscribe to Weekly Wrap emails