At AIR Summit our mission is to highlight the latest in cutting-edge technology that can help investment managers find a path that will allow them to generate alpha and thus justify their existence (and fees). The companies we find, a category we call AlphaTech, will be instrumental to that journey.
Artificial intelligence (AI) is a notable component of virtually every company I speak to that is involved in building innovative technology for investment managers and other market participants. It is imperative that businesses think strategically about how they can better leverage data and AI to build a real competitive advantage; time to adapt to this technology is going to run out sooner than most think.
Setting a framework that extracts more insights to drive better human decision making is a growing necessity for those organizations that want to deliver exceptional investor returns and survive.
According to Pitchbook, AI startups raised $31.8 billion over 2,899 deals in 2019, yet AI adoption and capabilities at large corporations remains relatively low for a technology with so much potential.
According to a recent survey of 500 CFOs and senior executives conducted by Genpact, just 13% of finance executives are using AI, analytics, and automation in any meaningful way. In my conversations with large asset managers, the same seems to be true for those executives leading investment strategies. Some might say there is a level of complacency or lack of imagination when it comes to re-inventing the investment process using AI and other advanced technologies. Others will just admit that it is hard to implement such a change and they are not exactly sure how or where to start. I think it is both.
AI is THE Major Trend
While I am not a trained AI expert, my partner Morgan Dunbar and I have focused our efforts at the AIR Summit to highlight AI as a major trend that is reshaping the financial industry. It will, in fact, eventually disrupt all aspects of investment management, including portfolio construction, risk management, stock selection, trading, operations, and distribution. On top of that, there will be a massive shift of skillsets across all these functions, as data science becomes blended with operational know-how. The future survivors will figure all this out.
There are many companies that have presented at the AIR Summit who have built proprietary AI to squeeze additional alpha from the market. Some notables include CausalityLink which is using AI to collect and analyze millions of documents to extract and aggregate knowledge; FactSquared which uses proprietary AI to identify who is speaking on a corporate conference call, in real-time, and to understand changes in tone or non-natural speech; and TruValue Labs which uses AI to determine and score real ESG behavior that can have an impact on company value and performance.
Risk Forecasting using AlgoDynamix AI
In 2019, the AIR Summit invited Dr. Jeremy Sosabowski from AlgoDynamix to present at both the London 1.0 and AIR Summit 5.0 events. The work he and his team are doing is based on their years of academic research at Cambridge University. It is interesting that their AI uses what is known as ‘unsupervised AI’ because it is identifying patterns in data sets that are neither classified nor labeled. This means the system can spot any type of market event anomaly, regardless if it has occurred before or not. A prime example is the current pandemic caused by COVID-19 where AI models trained on normal behavior failed. Betting on normalcy in financial markets can be expensive!
AlgoDynamix is using its unsupervised AI to do directional price forecasting, which is basically reporting what everyone else is doing in the market. It is finding patterns in primary market data by clustering it and spotting anomalies in asset prices, then using that to predict future prices. It isn’t a model that a supervised AI algorithm might utilize, so it makes no assumptions about the past to predict the future, which is part of their secret sauce. You’ve probably heard people talk about in-sample vs. out-of-sample results. But with AlgoDynamix there is no back-test or model: it is ‘just’ continuously reporting – in real time! – what everyone else is doing.
You might think of this as a ‘black-box’ which makes some active investors nervous when they are introduced to something they don’t fully understand. However, I think AlgoDynamix signals are meant to augment human decision making, not replace it. While the signals are generally short-term (days to weeks), they may also be predictors of major directional movements for most asset classes. Think of AlgoDynamix as another arrow in the quiver that creates market insights that should have some weight in the timing of investment decision making. To me, managing risk in a portfolio is job number one, so an early-warning signal such as what AlgoDynamix has created, can be both an alpha generation and preservation tool.
In my recent conversations with Jeremy, clients are now requesting factor model forecasting signals, using the same unsupervised machining learning technology that is core to AlgoDynamix. The use cases may be even more interesting when applied to the timing of tactical allocation strategies, so stay tuned!
Embracing AI for Alpha
There are so many reasons to embrace AI in the investment process, mostly because it will add alpha if an asset manager can assemble the right talent, data, analytics, and perhaps most importantly, culture. As we continue to engage with AlphaTech companies, it is paramount to understand if or how they utilize AI as part of the offering. While AI is still in its infancy when it comes to adoption and capability, it is the future of business and certainly investment management.
As you might expect, future editions of AlphaTech Spotlight will likely have an AI thread that weaves throughout as we continue to uncover and highlight innovations in the capital markets.
AlgoDynamix Alert (example)