AI in Asset Management
The introduction of artificial intelligence (AI) technology in asset management is being heralded as an opportunity to streamline the creation of more targeted and bespoke outcomes for clients, but the reality is there is still considerable disparity in the interpretation of what AI can deliver versus its use in practice. While increased use of technology can improve customer engagement and data can be mined for information on clients and potential clients; sub-sets of AI can empower asset managers to streamline processes to optimise investment decisions and processes (see Chart 1).
Understanding how AI can be incorporated into investment strategy workflows to deliver value, rather than representing an obstacle to overcome in terms of cost and resources, is slowly becoming more apparent. As third-party technology providers become more ubiquitous in asset management, there is the opportunity to learn from other industries and rethink current workflow processes.
One example from the oil industry is tracking robots on the seabed to monitor for maintenance requirements. Predicting the likelihood of an event before it takes place not only can prevent oil supplies from being disrupted but can also improve the pre-trade interpretation of the impact of including certain instruments in a portfolio.
Another example is the use of natural language processing (NLP) to systematically detect when CEOs ‘duck’ questions on earnings calls to provide predictive analysis on future earnings; or the use of clustering algorithms to identify fake news by examining the content for the inclusion of ‘clickbait’ wording – the higher the content, the higher the probability of the article being fake news and information that needs to be discarded rather than acted upon.
The level of adoption of AI has up to now depended on where on the technology curve an individual firm, team or portfolio manager stands, but while systematic funds with quantitative analysts are more likely to dedicate the technology resources necessary to uncover signals, many discretionary fund managers have yet to embrace any form of AI technology.
Historically, the greater the level of standardisation in extracting information for a fundamental investment strategy, the lower the alpha opportunity. However, the ability to extract value from data in a repeatable and reliable format is one example where decision making processes can be enhanced and investment strategies improved.
Modelling predictions based on enriched datasets can be used to facilitate the automation of certain tasks – whether that is an underlying decision to trade, uncovering latent liquidity or adjusting an investment strategy. By creating more efficient workflows on a scalable basis, firms can review a greater number and more diverse range of datasets such as credit card transactions to track company earnings pre-estimates, company solvency statistics, satellite imagery to track shipping or sentiment analysis for consumer goods. Then, through using enhanced analytics together with machine learning and NLP, decision trees can be improved and sped up by removing any unnecessary information.
Accessibility and portability of data will be crucial to the implementation AI technology. For many active management firms the incorporation of new alternative datasets remains on the periphery for a subset of funds or particular managers given the expense of the individual dataset, the level of accuracy and cost of extracting value. The time it takes to onboard and evaluate datasets differentiates how far individual firms are on the adoption curve – with sophisticated funds acting more as software development shops, consuming data to build a strategy, versus more fundamental active managers remaining wedded to traditional datasets to validate an existing investment idea. However third-party aggregators are now providing cost-effective means of aggregating company research, foreign news websites or more in-depth environment, social and governance (ESG) profiling.
Visualisation tools and user-friendly interfaces will better facilitate data consumption across a broader sector of industry participants and use cases. But to improve the portability of data, firms need to address the persistent siloed nature of data storage within organisations. This can translate into individuals being unable to access certain data sources, such as fixed income teams walled off from equity market data, and prevents the combination of all data sources. Firms are starting to address this issue through the introduction of centralised golden source datasets with data curators, which ensures there is a central point of access which holds clean, accurate, catalogued and documented data.
As the next generation takes up the mantle of building cleaner and more accessible data through APIs, their expectation levels and knowledge of what is now possible for asset management from their day to day interaction with technology will ensure the continued evolution of AI. For example, currently consumer datasets are difficult to reparse without breaching cross-border regulation; however, the replication of consumer data in an anonymised synthetic form can then be repurposed and modelled. The proposed use of video game engine sampling will enable firms to build and experiment with hypothetical alternative datasets from already cleaned and tagged data. This has the potential to translate into seemingly unlimited modelling scenarios which can form part of a portfolio construction and re-adjustment in real time.
As further innovation takes place, in whatever form it may take, AI will continue to assist in ways which large datasets can be consumed and repurposed in a manner that suits a particular user, asset class or geographical focus beyond the capability of the human brain – but a successful outcome will require technology to be used in tandem with human input as well as be fully integrated throughout a firm’s investment process and organisational culture. The threat of artificial intelligence wiping out the asset managers is overdone – AI will not replace asset management, but those who invest in technology will undoubtably replace those who do not.
Rebecca Healey is head of market structure and strategy, EMEA at Liquidnet.
Recommended for you
Advice businesses that directly contract offshore workers are exposed to legal challenges in light of a recent Fair Work Commission decision, writes Danielle Cornelissen, CEO and founder of 5 ELK.
Referral arrangements with other professional advisers, known as Centres of Influence, can help financial advisers to build client relationships, engagement and trust over time.
One of the apparently happy outcomes of QAR Tranche 1 was the introduction of relief from having to provide a Financial Services Guide but it turns out this was not all it is cracked up to be, writes Samantha Hills.
With more women aged 35-50 engaged in their finances and investments than ever, the cohort is a growing demographic for financial advice firms to work with, writes Nina Kazmierczak.
