The use of AI in trading will further expose weaknesses in network and data governance in markets where liquidity is dispersed. Policymakers who desire competition in trading need to shift their focus to governing the network, and the data that flows through it, at a systemic level rather than relying on historic policies that were predicated on their national trading venues fulfilling that role.
MSP releases the final paper its “Markets Unstructured: The Importance of Connectivity in the Reinvention of Markets” Series. The previous two papers explained how liquidity in competitive markets is dispersing, causing data asymmetries and connectivity challenges that have implications for access to markets, competition and the ability to optimise investor outcomes.
Paper III, released today, examines how the adoption of AI and potential growth will be stalled, and may increase risks, in competitive markets where governance frameworks for data access, data integrity and connectivity infrastructure have not been addressed. It calls for industry action, including a rethink by policymakers as to what constitutes systemic market infrastructure in competitive markets.
Key Findings:
- Most firms are in the exploratory phase of embedding AI into trading workflows.
- The sell-side know they need to adapt but are struggling to meet the operational connectivity demands of today before investing for tomorrow. 63% of sell-side cite gaining transparency over network vendor costs, a historically opaque business, as a bigger focus than investing in future technology. 78% of sell-side interviewed report having zero transparency into these vendor costs.
- Asset managers increasingly recognise the need for greater technological independence. 75% of firms are already reviewing or rebuilding their connectivity infrastructure but, for many, this is a tactical change rather than one led from a strategic enterprise level. Only 37% are preparing their connectivity infrastructure to support AI tooling.
- Those at the vanguard are driving a shift toward requirements for cloud-native, API-first, interoperable architectures that support flexible, scalable, multi-asset trading. However, this is a challenge to traditional vendor models that have historically offered all-in-one systems with extensive contractual lock ins. Transformation is, therefore, slow and many firms are yet to treat vendors as potential strategic partners.
- As firms adopt more agentic, model-driven workflows, existing weaknesses in data semantics, latency, and interoperability will be further exposed. Without easy access to clean data and a holistic connectivity architecture, fragmented, cross-asset data cannot be consolidated into a unified, real-time, normalised event stream with full context, data integrity and traceability.
The report concludes that the industry is moving from execution through a venue to execution through a network:
- Reliance on a single trading venue’s use of memberships and rules is no longer sufficient to uphold the integrity of data in an entire market or to provide sufficient transparency on who can access the market and the terms on which they do so.
- CLOBs are no longer the gravitational centre of trading, and, in the future ecosystem, they will be one trading model among many, not the de facto dominant venue or trading model
- Orders no longer follow a consistent linear route to a market or flow through a single market model.
The right foundations must, therefore, be in place to give all market participants the same opportunities to transform and scale secondary market activity or there will be more distortions in market structure and greater threats to market growth and resilience. As AI adoption accelerates, the case for co-ordinated industry action has never been stronger.
The report recommends that policymakers wanting competitive markets must reframe their thinking and elevate governance of data and networks to a systemic level. Meanwhile, market participants must redefine their value propositions and place in the ecosystem, raising the issue of connectivity and data management to strategic decision levels.
Niki Beattie, CEO of MSP and one of the authors of the report comments “Policymakers introduced competition. Now they must finish the job. Policymaking that continues to treat national exchanges as systemic infrastructure is outmoded – these exchanges are now just part of a growing network of liquidity options, and they should be treated as such. Governance of the network, and the data that flows through it is now of utmost importance and must be elevated to systemic levels in market regulation. If left unaddressed more structural distortions and limits on growth are inevitable.
Any policymaker that doesn’t believe such action is needed should ask themselves the following questions:
- Should a trading venue be able to force participants to use its proprietary technology in order to access its market?
- Should a technology vendor be able to determine which firms can or cannot connect to other participants on its network?
- Should the requirements for pricing transparency of network connectivity be the same as the transparency requirement for trading venues?
- Should an appointed CTP provider be able to run the CTP at a loss?
- Why is market data free in some asset classes (such as crypto) and not in others? Which markets are growing faster?
- What are consequences of asking the market to adopt standards on a voluntary rather than a mandated basis?
- In a world of increasing speed and decision making how quickly should data be cleaned and corrected?”
Rebecca Healey, also one of the authors of the report said: “The challenge is no longer data scarcity but fragmentation and semantic inconsistency. FIX variants, custom APIs, uneven tagging, and siloed systems degrade model accuracy, slow iteration, and increase operational risk. The real bottleneck is the inability to transform scattered, cross-asset data into a unified, real-time, normalised event stream with preserved context and traceability. Without this foundation, agentic AI is brittle and difficult to supervise.”
A copy of the Paper III report is attached.
