Interview by Manuel Boeck, Editor at Cash.ch, translated from German (original article)
cash.ch: Mr. Kolm, you are a mathematics professor at New York University and were Quant of the Year 2021. What are the fundamental principles of quantitative investing?
Petter Kolm: Balanced diversification is crucial. Instead of putting everything into a single investment, one should spread their money across different assets. This means you should hold a portfolio of various stocks, bonds, or other securities.
This does not distinguish quantitative investing from other strategies…
The traditional approach relies more on practical fundamental analysis. It is important to understand in which business areas companies are active and whether there are growth opportunities in these areas. However, when we turn to the quantitative side, quantitative investing is based on the same principles. The difference is that we use automated quantitative techniques to collect data from various sources, not just financial reports, but also from the media, stock exchanges, and macroeconomic developments.
How are these data used?
These data can then be utilized on a large scale to select investments. This real-time automation enables companies like Aisot Technologies to analyze thousands of investments simultaneously and in real-time across various asset classes. Since we are talking about millions of data points, this approach far exceeds human capabilities, but it enables people to make faster and more accurate decisions.
Especially the quality of the data is crucial. How is this addressed?
In the fundamental investment style, where people analyze financial statements and similar reports, they face the same challenges. Financial statements can be prepared in different ways, hence data is always of great importance. In an automated process, filters need to be implemented. Instead of visually examining the data and assessing its significance, filters should be implemented that evaluate the data and check its relevance because, as mentioned, decisions are only as good as the information available. And data is what we are looking for.
You make predictions about the future. However, not all influencing factors such as political developments are considered because there are no data available on them. Isn't this a problem?
The basic idea of a modern investment process is that you don’t have to be right 100% of the time.
Rather, I need to be right more often than wrong. It's about probabilities…
Exactly. Predicting future stock prices is extremely difficult. To navigate this dilemma, the risk is spread by betting on many stocks at once. Instead of just betting on a handful of stocks, bets are placed on hundreds or thousands of stocks. Additionally, predictions are usually linked with quantitative risk optimization. This approach can often yield substantial benefits even for smaller portfolios. It's well-known today that individual investors do not always act optimally. In quantitative investing, we build risk models capable of measuring and quantifying various sources of risk. With the help of these models, we can then create diversified portfolios that meet the risk profiles sought by a client.
You address behavioral economics. Is eliminating the human element a major advantage of the quantitative approach?
Exactly. This approach takes a much more comprehensive perspective. Humans cannot consider all available information when making decisions. For analysts, it is a huge challenge to collect and process information such as market dynamics, financial reports, press releases, earnings calls, and social media for just a few stocks. However, quantitative approaches allow us to handle this task efficiently and in real-time for thousands of stocks and other assets.
But are there times when humans have an advantage? Take political developments, for example.
When there is political uncertainty in a particular country, humans have an advantage due to their intuition. This may be because human intuition often relies on a broader range of experiences, social understanding, and emotional intelligence, which enables better capture and assessment of complex situations. In times of political uncertainty, these human capabilities can help detect subtle signals and nuanced contexts that purely quantitative approaches might overlook.
Back to investment decisions: How does Aisot proceed?
It involves an offering to enhance human capabilities in investing through systematic, data-driven approaches. However, quantitative trading has been around for a long time. It's about using data with various algorithms and statistical techniques for decision-making. What's new is the use of artificial intelligence and machine learning. We have built a team at Aisot that is familiar with modern techniques and machine learning. This allows us not only to deploy traditional quantitative tools but also to use modern tools for statistical data processing and machine learning. Moreover, at Aisot, we prioritize the use of new data sources over traditional data. For example, we can offer our clients strategies that utilize various news sources, like newspaper articles, press releases, and more.
And what is now the advantage of Aisot?
We are a young company, and currently, we focus on the stock markets in the USA, Great Britain, Switzerland, the EU, and digital assets. With our quantitative approach, we can create personalized portfolios from these assets that have different risk profiles for various types of investors. We have tools and techniques that allow us to personalize and shape portfolios. And the best part is that this process can be fully automated. For instance, Aisot clients can use our tools to calculate individual portfolio solutions. And all this is done on a secure, cloud-based platform.
How quickly does this happen?
It happens in a flash, within seconds. It is not a lengthy decision-making process that takes days to calculate.
What decisions does the client make?
Certain decisions are made at the client level. When a client comes to us and says, "I want to invest in the US stock market," we don't have a model that tells them to invest in the Japanese market instead. So, there are certain decisions that are made by the people in investment management. Sometimes clients come to us with a specific investment style in mind, and we support them in that. Of course, we need to know the current stock prices and look at volume and prices, but anyone can see that. If I look at a stock price today and compare it with yesterday's to say whether it will rise or fall, that is of very little significance.
What role does momentum play?
Momentum is an important factor that works on different time scales and markets. Academic studies show that there is typically momentum in stock markets over a period of three to nine months. There are two types of momentum: time series momentum and cross-sectional momentum.
What is cross-sectional momentum?
In cross-sectional momentum, the assumption is that stocks that have risen, for example, in the last six months, will on average continue to rise, while stocks that have fallen in the last six months will on average continue to fall. Based on these statistical insights, one can put together a portfolio that includes both long positions in risen stocks and short positions in fallen stocks. Through the difference in performance, one can then make money. This is an important insight. However, this is not how one would proceed in the real world. Because you cannot trade with all the thousands of stocks, and you may not necessarily want to. Therefore, one might want to be a bit more selective with the stocks that have risen and those that have fallen.
How is this done?
One can use quantitative tools to screen stocks and say, "Okay, we focus on these 50 stocks and exclude these 50." Someone could also say, "Oh, I don't want any short positions in my portfolio." This means that all stocks with negative momentum must be excluded. Instead, those that show positive momentum are chosen. One might also decide to combine momentum with several other signals to select a group of assets for trading. This approach enables investors to not only benefit from momentum but also to diversify over other investment themes.
How important will quantitative investing be in the future?
When we discuss quantitative or systematic investing, we can say that it constituted a very small portion of investments back in the 70s and early 80s. Today, the share of systematic approaches has increased. What's special about this is that we can all see it in some way today. You are probably familiar with the concept of systematic indexing. Companies like BlackRock and Fidelity offer ETFs that not only replicate an index like the S&P 500 or Nasdaq but also indexing strategies that were formerly systematic strategies. There are ETFs that follow a momentum approach. Many of these approaches, used by hedge funds in the 70s and 80s, are now available to us as ETFs. What I mean to say is that the acceptance of a systematic and scientific approach to investing is now understood on a broader basis. And this trend will undoubtedly intensify. Machine learning and new global data sources already offer fantastic opportunities to develop innovative investment strategies for clients. Through automation, we can quickly make decisions about various assets and investment themes and tailor portfolios to the risk profiles and investment goals of clients. Investors can already benefit from machine-optimized investments today through readily available investment products like AMCs and ETPs, for example, with the technology of Aisot in the background.
About Petter Kolm:
Before his role as a professor and director of the Mathematics in Finance program at NYU Courant, Petter worked in the Quantitative Strategies Group at Goldman Sachs Asset Management. In 2021, he was named "Quant of the Year" for his contributions to the field of quantitative portfolio theory. Petter has co-authored numerous articles and books on quantitative finance, trading, and financial data science, and is a member of several corporate advisory boards, editorial boards for scientific journals, and professional associations. He holds a Ph.D. in Mathematics from Yale University, an M.Phil. in Applied Mathematics from the Royal Institute of Technology, and an M.Sc. in Mathematics from ETH Zurich.