Larry Cao, author of CFA AI pioneer in investment management.
In 2017, the defeat of the top human go-to player by Alfago Artificial Intelligence (AI) revealed to the public the world of possibilities that AI scientists had quietly explored year after year. Driverless cars, AI doctors, and robo-advisors, among other innovations, everything seemed within reach.
Among such promises and possibilities, the verdict was like remembering my famous law:
“We overestimate the impact of a technology in the short term and underestimate its long-term impact.. ”
Indeed, the story we hear from the media three years later is one of the imperfect possibilities and frustrations of adopting AI at a slower pace than a revolutionary transformation.
For example, in EconomistIn a recent technology quarterly headline, “Artificial Intelligence and Its Limits,” the editors claim that “information can be less than you think and full of traps.” This limitation, among other reasons, led them to the conclusion that “some dreams of high summer will fade into the cold of autumn.”
This line of reasoning is as old as AI. But the global epidemic provides a rare opportunity to determine how well AI and Big Data applications have performed in terms of investment. Since the Covid-1 crisis was on a “once in a century” scale, researchers could not deceive their models with epidemiological data that had never been seen before.
So how did they manage during these turbulent months? Have AI programs completely failed investors? Or did they serve them well?
The simplest tests come from the trading model – an AI program that predicts trading signals that traders can use to determine when, where and how to trade. Due to their short-term nature, these models rely on very recent data and can adapt to rapid changes. David Wang, CFA, who works as managing director at State Street Bank to work with AI on expanding the trading program, confirmed. “The low-delay process that we prefer has done particularly well,” he said. They have powerful hardware to process data very quickly.
It becomes less straight from there. For machine learning models that require long-term data series, the new environment presents a challenge. Of course, this is rarely unique to AI programs. All quantitative models face that challenge. (Many years ago, when I was developing “quantum” models, I realized: My choices in the development process were influenced by my experience in the market, although I did not match my models with historical data. In that sense, such an epidemic ratio is really unknown to us all But this is another day’s story.)
So how should investors coordinate with new data puzzles? There are several options that stand out, which are actually consistent with our philosophy that future investment teams will follow the “AI Plus Human Intelligence (HI)” model. AI programs are not a substitute for portfolio managers and analysts but rather a source of good support. In times of crisis and uncertainty, investors will naturally rely on their experience and judgment as before.
The most important thing for investors to realize at such times is that uncertainty is the focus of this business. We must always be vigilant for changes in the market environment. Or as Ingrid Tyrens, a managing director at Goldman Sachs, put it, “All AI (and Quantum) models should bring a kind of health warning.”
If we detect changes, we need to restore our reliance on historical data. Since machine learning models are trained on data, if we do not believe that the environment from which the data was obtained is compatible with market models, we should try simpler models. These models will depend on fewer features, or variables that explain the output or results of the models. Reducing the number of features helps us understand what will still work in the new environment and whether or not we are less likely to be confused by the dataset in question.
We can also test whether the range of features is broadly similar to what we have previously tested. This may be a new environment but if the features are in the same range our models can still hold on. “While recent market behavior has been volatile, the features used by our machine learning models have not been at an unprecedented level,” said Anthony Ledford, chief scientist at Man AHL in London. In other words, our ML models did not find themselves ‘outside of the information’ on which they were trained. “
Nevertheless, Ledford added that they employ strict risk controls that reduce positions during higher volatility as observed recently. These are the best practices of common sense no matter which model or method we apply in managing our portfolio.
The CFA of Oktree Capital, Howard Marks, recently highlighted the critical importance of identifying changes in governance during his presentation at the CFA Institute’s 73rd Virtual Annual Conference. He believes that Oktri’s greatest achievement was the transition from one regime to another. This theme seems to be equally applicable to machine learning models. As Mark Ainsworth, head of data insights and analysis at Schroders, said, “If you can detect changes in your model regime, you should be rewarded enough for it.”
More encouraging for AI is that investors have gone beyond the “deal” strategies described above. They actively follow new applications, especially big data applications, which help capture more real-time or at least more timely information. Tierens, for example, reported an increase in demand for their services from investment parties during this period. “We’ve been using a lot more data than ever before in the last few months,” he explained. “Investors in this environment are understandably more concerned and they are all looking at alternative data because of the timeliness.”
Ainsworth confirmed, “The epidemic has really given us a chance to shine because investors are looking to us to explain what is happening in the market.” “We have followed the approach of a scientist and tried to explain different developments [using simpler models] Instead of using the classic machine learning model that matches the data, which is more common than the engineer’s method. ”
Promotions on AI by Alfago have been fading since 2018, according to Google Trends. This is a good sign, though, if we believe in the Gartner hype cycle. This simply means that some pioneers have moved from hype to action. Incredibly some have failed, but mainstream adoption will only happen when the “disillusionment” phase shakes up skeptics.
AI plus HI remains the maximum framework for receiving AI. As this epidemic has appeared, the importance of professional investors has only increased. And it shouldn’t be bad news or disappointing for anyone.
If you liked this post, be sure to subscribe Entrepreneurial investors.
All posts are the author’s opinion. As such, they should not be construed as investment advice, or the opinions expressed must not reflect the views of the CFA Institute or the author’s employer.
Photo Credit: © Getty Images / KTSDESIGN / Science Photo Library
Professional education for CFA Institute members
Members of the CFA Institute have the ability to earn self-assessment and self-reporting professional education (PL) credit, including content Entrepreneurial investors. Members can easily record credit using their online PL tracker.