What does the future role of AI look like in terms of execution?
We see the pursuit of prediction following a similar trajectory as the quest for speed. First, we expect predictive capabilities to become table stakes for most participants in electronic markets. For example, market data systems will be expected to be not only fast and content-rich, but also predictive, delivering the expected direction and timing of price changes synchronously with the latest prices.
This mirrors the evolution of low latency trading technology. Early movers gained an edge by automating and collocating their trading systems, then optimising them using purpose-built hardware logic in FPGAs and wireless network links between data centres. This is now de rigueur for most quantitative trading firms, agency brokerages, and exchanges.
Second, we expect firms increasingly will develop their AI-driven trading systems by identifying their differentiator or their “secret sauce”, focusing their internal development on capturing that edge, and buying the remainder of the capabilities from a trusted partner. Firms will recognise that they will develop more sophisticated quantitative capabilities more quickly by leveraging foundational signals, datasets, and tools from specialist providers. Many quant teams still suffer from “not invented here” syndrome, but we see this changing across the industry as competition heats up, talent remains scarce, and the need to accelerate development increases.
How important is data when considering AI?
How important is thrust when building a rocket? Data is both the fuel and purpose of AI much like thrust is both the driving force and purpose of a rocket. We discover, build, and train new AI models using vast datasets. Once ready for use, their purpose is to automate the synthesis of data at scales and speeds that were previously unimaginable.
Therefore, data quality is imperative. The ability of a firm to build AI models that are both effective and trustworthy is directly tied to the quality of datasets used in the discover, build, and train process, as well as operation.
In capital markets, data quality is judged at minimum by its timeliness, granularity, consistency, and completeness. Ensuring that the data is reliably delivered to the model, even in the presence of faults in other parts of the trading infrastructure, is an equal consideration to the accuracy and recall of a predictive model.
In terms of automation, what are the priorities for the buy-side, sell-side, exchanges specifically?
On the buy-side, we are helping an increasing number of firms push more of their automated trading strategies into purpose-built hardware. Firms see an advantage or necessity in having more of their tick-to-trade logic operating in the nanosecond realm. Encouragingly, we also see quantitative firms consuming more of our predictive datasets to use as inputs and building blocks for their algos.
On the sell-side, the dominant theme is deploying low latency market data and brokerage systems to more collocation data centres globally. We typically begin relationships with sell-side firms by addressing their high volume, low latency market data needs in US equities and options. We then work in partnership to deploy a tech stack to equities and futures markets in Canada, Europe, and Asia Pacific. Across the capital markets ecosystem, there are common requirements for reliably fast and resilient market data, especially under stressful market conditions.
What role can/do providers play in capital market democratisation?
The table stakes for competing in today’s electronic, automated, and data-driven capital markets are raising relentlessly. Specialist providers promote competitive and resilient capital markets by enabling a diverse community of firms to participate successfully. Lowering barriers to entry helps prevent brittle markets dominated by a continuously consolidating club of the largest trading firms.
Good providers help their customers achieve business results that they otherwise could not achieve on their own. In Exegy’s case, we help our customers grow revenues by expanding to new markets faster and more cost-effectively, improve trading returns with faster systems and more intelligent data, expand margins by reducing data centre footprints, or a combination thereof.
The process works best when firms view their providers as an extension of their own technology and operations teams. Tighter partnerships yield better results in our experience. This has held true across a diverse community of firms in terms of their roles in the ecosystem (buy-side, sell-side, trading venue), number of markets and asset classes traded, size and sophistication of their technology teams.