AI Progress in Trading and Settlement

With the successful implementation of T+1 settlement of financial securities in the US on 28th May 2024, all eyes are on European T+1 Settlement next as ESMA proposes that the European Union (EU) moves to T+1 only in October 2027.

This is due to the fact that the EU with its numerous countries and

clearing platforms such as Euroclear and Clearstream, as well as country- specific settlement infrastructures, has a higher complexity in trading and post-trade processes.

In the US, the DTCC (Depository Trust & Clearing Corporation) took the lead on preparing for T+1 settlement by training all market participants settling trades in the US, with a clear roadmap and successfully launched in 15 months only. The relatively simpler infrastructures in the US and governance by regulators along with pricing aligned for earlier settlement all helped for the T+1 to successfully launch on schedule end may 2024. Some of the benefits so far observed are improved liquidity of 23%and reduced fails of 10% in the first days after the transition from T+2 settlement.

In fact, underlying the faster processing in securities settlement are a myriad of AI systems already deployed by various settlement infrastructure providers. These cover different parts of the client compliance and trade processing chain. For example, Clearstream has several AI tools combiningGenerative AI, LLM and Machine Learning to provide efficient processing for its clients account management, especially shortening the KYC (Know your Customer) procedures. AI tools also participate in the actual trade order booking, refinement and risk management, as in the case of collateral management of traded products.The model and system choices driven by such AI systems in which client data and sensitive information is shared with service providers and 3rd parties are pushing towards a trust- based system where fewer authenticated certified entities with partially segmented or secluded repositories share data and process it using AI, yet protecting the privacy of these sensitive data.

In reality T+1 settlement already exists in European large volume transactions such as bond trading, although it is not generalized and not across all asset classes. The current T+1 trading is managed thanks to strict compliance procedures, clearance for certain clients to engage in higher volume and faster trading settlement schedules, with streamlined processes thereby building a system of trust. So in order to achieve Pan-European T+1 securities trading entitlement, for the EU banks, brokers, custodians, the priorities will be to make sure their different systems, from front office trade booking systems to trade affirmation, settlement and post trade back office processing are 100% fraud proof,

efficient and agile with architecture that will fit with the AI systems of clearing platforms, and able to communicate client data in segmented and vault-protection methods. This seems obligatory before migrating to a pan European T+1 processing and to safeguard themselves from data leaks of AI processing, for keeping their competitive advantage, and avoiding large volume problems related toTrade processing and settlement.

Aparajita Sood D. is a Finance and Investment Specialist.