AI is the most disruptive technology innovation of our lifetime. Business is embracing AI and leveraging a variety of data types for integrated processes across all lines of industry. Although firms in the alternative investment sector understand the importance and potential impact of AI, they often struggle to move from this knowledge to incorporating the resource in their business operations in a meaningful way. Top challenges that organisations must address to implement and scale AI initiatives are understanding, cost, access to useful operational platforms and technologies, access to adequate volume and quality of data and to address the correct governance pathway for AI to ensure regulatory compliance.

Whilst AI is banded around as a buzzword in many industries, most companies are actually using machine learning (ML) in practice. ML is the process of using mathematical models of data to learn and predict outcomes without direct human instruction. It enables algorithms to continue learning and improving on their own, based on experience. ML uses historical data to learn where signals exist, and outputs a predictive model based on a defined target. The benefits to investment management are clear.

ML algorithms are classified broadly into 4 mainstream types: unsupervised learning, which effectively learns through failure using unlabelled data; semi supervised learning, which uses a small amount of labelled and a large amount of unlabelled data and reinforcement learning where positive and negative actions are used methodically to reward or penalise certain behaviours. Lastly is supervised learning, which uses a training set of data that is labelled to learn.

A perceived benefit of using ML to build an end-to-end investment strategy is that it removes human bias from the decision-making process of a fund. Deep learning allows the strategy to be influenced by scale as it can look at thousands of stocks at once, across thousands of dimensions. This simply isn’t possible in a fund where the investment team need to review and categorise data ahead of making investment decisions. By removing human workflow and subjectivity from the decision-making process a fund can change strategy and adapt modelling at speed, meaning predictions can be made on a huge volume of stocks in just a few minutes. The high degree of accuracy of the ML output can lead to better returns on investments as well as reducing risk over time.

When considering how to implement ML into a fund strategy, consistent enterprise-grade security and governance are a significant requirement. The tremendous growth of available data in terms of volume, size, and complexity means it is imperative to also implement an institutional grade data strategy to address the funds growing needs. A realistic data strategy must incorporate a clear road map with milestones, so that strategy documents do not end up as digital assets with no real value. Unless a fund has an excellent strategy in place ensuring data security, data quality, data stewardship, and data governance, ML tools cannot contribute to superior business outcomes and support winning new mandates and generating alpha.

ML can contribute to the competitive value of a fund as part of the overall data modelling. This requires a sound technology architecture that can support database storage and access, data flow and sorting of large and granular data sets with the power to produce high quality analytics accurately at speed. Funds should consider their choice of third party vendor when considering implementing ML and what other specialisms the fund will require; as we move to automated workflows and ML strategies we see more data scientists in the financial sector, signalling a change in approach for funds both internally and externally.