MDT’s investment approach is quite different to many other active managers you may come across in the market. There are collectively a few things that I think really differentiates MDT from our peers.
First off is our experience. We’ve been managing our investment process now for over 30 years. We’ve been able to evaluate how our process has worked in many different types of market environments over that time frame, and we have very strong experience in utilising predictive modelling and advanced modelling techniques to help us identify a broad range of companies to consider for our portfolios.
We have a great team here at MDT, and everyone contributes to the success of our franchise in different ways. Many of the people that were on the investment team when I started are still here, so there’s an added familiarity among many of us and the people on our team who are newer have brought so much value and new ideas to the process – it’s a great mix of longevity and freshness.
In the modeling that we employ, we have insisted on maintaining a transparent and accountable process. The machine learning technique that we have been using for decades involves decision tree algorithms, an analytical tool that helps us to understand and explain the rationale behind every trade.
Simply put, a decision tree is a series of yes/no questions that lead to a specific outcome. It’s a straightforward tool that many people are likely to be familiar with.
In our quantitative investment application, the questions are of companies, and relate to company data – like valuations, price trends, and corporate structure. The outcome we aim to predict is the return – specifically, whether and how much a company is likely to outperform or underperform its benchmark. And with this machine learning technique, we’re able to construct daily relative return forecasts for every stock in the US equity market, which is the information we use to build our portfolios.
We manage highly diversified portfolios built to limit unintended risks, which can lead to stronger portfolio resilience through market cycles and the potential for improved risk-adjusted returns.
We have portfolio constraints at four levels: company, sub-industry, industry, and sector. These constraints limit the benchmark-relative exposure to any single company or group of companies. Notably, we avoid taking large overweights or underweights in any single sector. That helps to reduce active exposure to macroeconomic risks that can drive a lot of tracking error in volatile markets.
When we analyse performance, we look at how indicative our factors, and key combinations of these factors, were in predicting performance.
By considering what drives the model’s decisions and where the strategies are positioned within that we get a much better picture of why the strategies performed as they did.
Where MDT differs from quants of old is our use of advanced algorithms to develop superior insights from our data, and the speed at which we can process it. In our ever-more complex world, we believe this data-driven approach is increasingly valuable.
