Using Analytics and AI to Manage Investment Risk

I attended a fascinating virtual keynote event provided by the Financial Times in partnership with Moody's. It encompassed the topic of data analytics and AI to manage and allocate investment risk to increase the effectiveness of portfolios; spearheaded by Jon Eggins (Jon E.), Michael Hunstad, Tatiana Segal, Matt Seymore at Moody's, and Will Schmitt.

AI modelling on Climate Risk

Essentially, general AI can process vast data sets more efficiently and perhaps more accurately. Therefore, building custom portfolios and risk calculations to include climate risk can help shield investor portfolios, particularly real estate investors, against catastrophic weather events.

This is a well known fact because residential and commercial real estate are closely linked to weather events by likely being the first forms of property affected.

In the event of a hurricane such as Hurricane Milton, increased wind speeds, and flooding etc., it would probably be too late to protect your real estate investments against losses in value.

As climate risk becomes much more of a factor in threatening both tangible and intangible assets, alongside data environments evolving to be far more complex, it may prove crucial to use AI to see direct and indirect exposures across individual and multiple sets of portfolios.

Geopolitical Risk

Importantly, the financial and tech industry does not possess sufficient data for general AI to create results in line with regulations and compliance but this may be possible in the next 1-2 years.

However, AI can be used in a 'chain of thought' style model to evaluate risk relating to geopolitical escalations, sanctions, regulations and tensions. Going even further, it is said that the next generation of AI regarding geopolitical risk, will likely be able to give answers of who the winners and losers are in geopolitics increasingly faster. Due to AI being able to monitor investment portfolios in real time and flag certain assets that may be subject to sanctions in the near future. Which is super important given the current geopolitical instability within the Isreal/Palestinian conflict and the Russian occupation of Ukraine.What does this mean? Put simply, investors will be able to predict the future of geopolitical tensions with increasingly more accuracy and adjust their portfolios accordingly to maximise profit protection and returns.

Investor Concerns

One investor asked: How do you ensure analytics are reliable and free from hallucinated data as much as possible?

This is a great question because we are normally led to believe that AI does not make human-like errors but it is never addressed on how this is the case since AI is designed to mimic that very same human-like intelligence. The consensus was that to mitigate hallucinated data there would be a lot of testing but it was acknowledged that the general AI research assistant had a relatively low hallucination rate.

Perhaps a more conclusive answer is that there is no escaping hallucinations because the AI research assistant is designed to be human-like so any errors made are no different to a qualified analyst. However, if you take a mature approach and understand your data sets alongside using AI then you can maximise your accuracy.

How Moody's Approach General AI

穆迪分析 house a processor that absorbs 27,000 news sources for managers to understand the direction of sentiment for their assets and portfolios.

Obviously, an individual investor may not always be able to house a behemoth of a computer to process such vast quantities of data but Moody's approach to AI modelling could be replicated on a smaller scale and even smaller for start-up investors with limited capital.

AI is being used by major financial firms to manage portfolios for their clients and it is time for individual investors to do the same.