Key challenges include the fiduciary element of asset management, outdated information in some widely available models, and a preference for traditional machine learning approaches.
On the positive side, generative AI offers significant productivity benefits, helping to process vast amounts of data, make quicker decisions, or summarise bottom-up analysis based on historical reports and data.
Kev Toohey, principal at Atchison Consultants, noted that while large language models (LLMs) have recently gained attention, his firm has been using proprietary machine learning models as part of its investment process for more than five years.
“We focused on developing machine learning models that identify and isolate patterns between economic and market factors and outcomes of different investment strategies,” he told InvestorDaily.
“These models help with assessment of market conditions and the development of signals for our tactical asset allocation views.”
Moreover, Atchison Consultants uses separate machine learning models to assess the characteristics of individual assets and managers. These models aim to predict the return profile of the investment relative to economic and market conditions.
“We see data management and visualisation as a key tool for our team to effectively identify and stress test investment opportunities and have hired a data scientist who spends a significant amount of resources on internally coding our investment and reporting models,” Toohey said.
Although acknowledging the role of AI in multi-asset portfolio optimisation, Toohey emphasised his firm still relies on “deterministic models, not AI” for portfolio optimisation.
“We balance the advantage of repeatability of calculation for things like portfolio optimisation as remaining preferred,” he said.
“For less well-defined problems such as where markets are in the cycle or how specific assets may perform in the future, we utilise machine learning models as being informative for our analysts.
Similarly, Sebastian Mullins, head of multi-asset and fixed income at Schroders Australia, confirmed generative AI offers benefits in accelerating processes.
However, he emphasised that traditional machine learning remains an effective tool for modelling “more complex, multidimensional relationships in data”.
Mullins warned that generative AI is less useful for real-time decision making because publicly available language models often rely on outdated information, such as ChatGPT, to which its data only extends to 2021.
“This means generative AI is providing incorrect or out of date data or news headlines,” he said.
By contrast, machine learning, he noted, is valuable for tasks such as forecasting future market pricing, making economic variable predictions, as well as determining structural breaks or outliers in data.
“We currently use deep machine learning modes (using neural networks) to determine short-term interest rate predictions versus market pricing, clustering analysis to determine conditional interest rates and currency spot probability distributions, along with reinforcement learning to determine fair value for government bonds,” he said.
Mike Chen, Robeco’s head of Next Gen Research, highlighted that while AI will transform many aspects of how multi-asset portfolios are managed, certain areas, such as face-to-face client servicing, will continue to rely on human skills.
Chen also pointed out that data availability could be a limiting factor for some asset classes.
“We believe AI has a large role to play in all assets where there is sufficient available data. In certain asset classes where data availability isn’t great, the role of AI will be relatively less as AI algorithms require data to be effective,” he said.
Similarly, Toohey pointed out that asset classes with frequent market valuations, like listed equities, are generally better suited for machine learning models compared to illiquid assets, which rely more on subjective or opinion-based valuations.
According to Tom Boyle, CEO of derivatives-based investment manager Atlantic House Group, the biggest benefits of leveraging AI are productivity-related.
“We are not giving AI money to manage, but it allows [us] to take in all the different data points, where previously we might have taken 10, now we can take in 30. And it allows you a wider scope or allows you to explore different geographies with the same level of detail but not necessarily having to have experts in every single field,” he said.
Boyle expressed discomfort relying on AI for asset classes in general in the present moment, but said he sees potential for AI to enhance transparency in less liquid and less transparent markets by processing data more efficiently than humans.
“I think in the next 18 months that can change into running money, but from our perspective, it is certainly a data tool to enhance productivity rather than actually apply money to different markets,” he said.
