For decades, post-trade operations have been dominated by manual, monotonous, and inefficient processes. To mitigate operational errors, investment managers have often implemented risk management procedures and automated processes. Despite these efforts, few would argue that post-trade operations have truly been optimised.
AI aims to replicate human-like cognitive skills. This encompasses machine learning, natural language processing, pattern recognition, problem solving, and decision making. Notably, one of the recent areas of accelerated development is generative AI (GenAI). GenAI encompasses advanced machine learning techniques and focusses on creating models to generate new creative outputs such as images or text.
Over the last decade, interest in AI has steadily grown—though adoption is still in its earliest stages. While much of the buzz around AI is focused on algorithmic trading and operational error reduction (among other use cases), there has been less conversation on its post-trade applications. While we’re still in the innovation phase of the AI adoption curve, it’s worth broadening the conversation to AI’s numerous potential post-trade benefits such as improved efficiency, error reduction, and informed decision making with deeper insights into trade-data.
Some post-trade service providers are exploring using natural language processing (NLP) to extract relevant trade information from unstructured data to create a digital version of a trade confirmation. Use of pattern recognition models in trade matching and settlement process holds the promise of quickly identifying anomalies and irregularities, resulting in faster resolution of the identified issue. GenAI models could also predict potential delays in the trade confirmation and settlement process by analysing historical data patterns which improves resource allocation and risk management.
As settlement windows contract, AI could play a crucial role in identifying and addressing the complex issues underlying settlement failures or operational inefficiencies, leading to faster resolution. Recent studies indicate that financial services firms have faced regulatory penalties and resolution expenses exceeding $900 billion over the course of the last decade.1 Improving the effectiveness of matching and settlement procedures is crucial as the European market moves closer to achieving T+1 settlement, with the prospect of settling on T itself becoming a near-term reality.
Automation based on pre-defined rule sets will result in matching of records between two data sets automatically. Use of machine learning algorithms can enhance this process by learning from historical patterns and suggesting potential matching records. Pattern recognition models will enable identifying irregularities and anomalies thereby improving the accuracy of reconciliation. With just the exceptions left for the users to research and solve for, use of AI models creates a much more efficient process.
In terms of cash reconciliations, we are rapidly approaching the ability to conduct them in near-real-time, which is a significant advancement compared to the current market standard of T+1 reconciliations. This advancement will provide fund managers with real-time insights into their cash positions, facilitating faster decision-making and enhancing cash flow management and forecasting capabilities.
Since the 2008 financial crisis, there has been continuous demand for various regulatory reporting needs. The task of gathering necessary data from multiple sources, conducting validation checks, and formatting it to meet regulatory standards remains a challenge. While automation can alleviate some of the workload, managing exceptions still requires manual intervention. This is where the implementation of AI could be particularly beneficial, as it holds the promise of proactively identifying exceptions and address them based on past incidents, thereby reducing the risk of regulatory breaches.
AI could be particularly useful in regulatory change management, as it can efficiently analyse the impact of new regulations or amendments on reporting requirements. This analysis results in smoother adoption of revised regulatory standards.
The integration of AI in data management and analysis has the potential to significantly enhance the operational framework of post-trade operations in the future. AI holds immense potential in areas such as enhancing data quality, detecting anomalies, and ensuring data privacy and security, thereby driving significant benefits.
Identifying and rectifying data errors and inconsistencies at the trade entry stage enables a more seamless downstream processing environment. Given the myriad third-party dependencies, maintaining high-quality inbound data is paramount for facilitating the trade lifecycle. AI algorithms exhibit proficiency in detecting anomalies and outliers within trade data, potentially indicating discrepancies, fraudulent activities or compliance issues.
Asset management organisations are not shying away from adopting the latest innovative technology solutions that enable them to generate alpha. Whilst some asset managers may not be totally ready to deploy AI at an enterprise level, with only individual employees opting to transcribe and summarise calls, others are beginning to adopt AI solutions that help them make better investment decisions and generate alpha—but many are missing the value that they could gain from implementing AI in post-trade processing.
Perhaps portending wider adoption to come, the European Union formally adopted world’s first AI regulation which seeks to establish a comprehensive framework to govern the development, deployment, and use of artificial intelligence across various sectors. The law adopts a risk-based approach, categorizing AI systems into four levels of risk: unacceptable risk, high risk, limited risk, and minimal risk. Developers and deployers of high-risk AI systems must adhere to strict requirements, including data quality, transparency, documentation, human oversight, and robustness. While non-compliance with these regulations may result in penalties for companies, it is important for firms to integrate these technologies in alignment with organisational policies, regulatory requirements, and industry best practices to ensure the integrity and security of the operational processes.
Though it is still early days for many AI use cases, exploring AI and Gen AI now will set the stage for reducing errors and increasing efficiency in the inherently complicated world of post trade processing. For asset managers looking to gain any advantage in a competitive market, it’s worth keeping a close pulse on AI applications and technology maturity over the coming years.