7 decision errors in investment management

Yesterday, I wrote a long post on 7 decision errors made by business leaders and entrepreneurs. As promised, today, I will apply the same methodology to investment decision making and give some tips on how to mitigate them. If you are not familiar with the seven types of errors, read yesterday’s post first, please and then come back. Don’t worry, I’ll be waiting right here…

…all good? Ok, so let’s talk about type I and type II errors in investing. In my experience, these are by far the most common mistakes made in finance. In particular, the error to assume a correlation where none exists is incredibly common in the investment world. The vast majority of smart beta funds suffer from these errors, in my view. Before Christmas, I wrote about a study that showed that publication bias in economics is extreme with studies that find a significant correlation three to four times more likely to be published than studies that find no such correlation.

But reducing the likelihood that you commit a type I or type II error in investing is not so hard. Out of sample testing of a hypothesis is, in my view, the best way to do that. Use one data set to develop a theory that there is a correlation between two variables, then use a completely different data set to test if the theory still holds. This out of sample data set doesn’t necessarily have to be data that was published after a theory was developed. It can also be data from a different country than where the effect was first described. Often an effect is first described in the US, but when you test it in other regions, you may find it is probably just a spurious effect in the US data. Or you can look for data that is older than the data you used to test your hypothesis and outside of commonly used databases like the CRSP data in the US. That might show you if a hypothesis holds true out of sample as well like in this case.

Type III errors can be tackled in the same way they are tackled in the business world: increase the diversity of experts providing input to a problem. Hedge funds with more diverse investment teams outperform their more homogenous peers. In particular – and I know I am talking about my own book here – you need to have someone in the investment team that gives you a broad overview of the world. That can be an economist or a strategist, but the important thing is that it is a person who understands more than just equity markets and sector earnings or the like. A good understanding of fixed income markets, fiscal and monetary policy and geopolitics is absolutely crucial in today’s markets. The year 2022 should have made that clear. Don’t rely on so-called strategists that do nothing but draw a few lines on charts or try to forecast earnings growth and valuation multiples for the next 12 months. There is very little added value to that kind of service, in my experience because it misses the bigger picture.

Type IV errors are more difficult to overcome. It is incredibly easy to identify the right problem and then choose a suboptimal solution. In my experience, this happens particularly often when people look for textbook solutions based on the rational behaviour of investors or political actors. More than 50% of market action is driven by human behaviour. If you do not understand behavioural finance, you will have no chance to ever outperform an index investor in any market. Obviously, becoming an index investor and forfeiting the chance of outperformance is a valid option and in my view the best option for most investors. But if you are a professional money manager and you are trying to manage a fund without deeply and constantly thinking about how human behaviour influences share prices and markets overall, I don’t think you have much of a chance.

Type V errors together with type VI errors are in my view best managed by using stop loss and re-entry rules for investments as I have described in my book. In essence, a stop loss limit tells you to act when you might be pre-disposed not to act, while a re-entry limit does the same just in the opposite direction. On the other hand, if a stop loss signal has not been triggered, it is simply not time to act, whether you like it or not.

This brings me to type VII errors. Just like in business, I think there is no real protection against such error cascades. The only action I can think of is to introduce regular after-action reviews in a team to analyse where mistakes have been made in the past and come up with changes to the investment decision-making process that can help avoid making the same mistakes again. But in particular, when it comes to protecting against Black Swan events, which often trigger type VII decision errors, I think there is no viable hedge. Nassim Taleb might disagree with me, but I think all tail hedging strategies suffer from the same flaw. They require the investor to suffer recurring small losses for a long time in the hope that eventually, the strategy will pay off big time and make up for all the past losses and more. That sounds nice in theory, but in practice, it is an invitation for investors to commit type V errors and act when they should not change anything. If you think about tail hedging strategies over the last 20 years, they have worked in 2008 and then again in 2020. In between, you had to suffer constant losses for 12 years and now again for 2 years in the hope that the next catastrophe will show up in time before your profits from these two years have been eroded completely. There are very, very few investors that can successfully implement an investment strategy with that kind of payoff structure.