Extrapolating the recent past

The last two years have been full of one-off surprises. A global pandemic, supply chain disruptions, the Ukraine war. No wonder investors have had a hard time forecasting what is going to happen next.

In fact, in such an environment of repeating surprises, forecasting errors are larger than normal not just because it is harder to forecast something where there is little to no historic precedent, but also because humans are prone to overreact to such one-off events. One natural tendency of analysts, economists, and humans, in general, is to overreact to the most recent events and project them into the future. Over the last twelve months, I have seen people claim that we will never leave the virus behind and will always have to live with social distancing restrictions, that supply chain disruptions will create a recession, and that the Ukraine war is only the beginning of World War III. Well, social restrictions are behind us (in North America and Europe) or soon will be (in Asia). Supply chain disruptions are sorting themselves out and freight costs between Asia and Europe/North America are falling rapidly, and we still haven’t had a recession. And as for World War III, the jury is still out, but at the moment, it seems an escalation of the Ukraine war to a NATO country is unlikely.

So why do people constantly forecast the recent past into the future? It is simply availability bias. Because the recent past is fresh in our memory, it is readily available when we make predictions about the future. And the more volatile and “random” a process is, the more we rely on this kind of availability heuristic.

Imagine a stock that goes up in one month and then down the next and alternates in that fashion every month. Then, one month, the stock goes up three months in a row. Most people will have a hard time keeping track of the random ups and downs of that stock, but a 3-month winning streak will stick in their memory and lead many investors to forecast that the stock will continue to rally over the next couple of months or even the next year. But when the stock then returns to its usual one up, one down pattern, these forecasters have to revise their forecasts down and eat humble pie.

If you want to see this kind of forecasting in action, all you have to do is watch any sports punditry on TV. These guys are utterly useless in their predictions mostly because they have no clue about stochastic processes. Thus, they talk about a hot streak and try to change managers after a series of losing games to turn the fortunes of the team around. Yet, after an initial honeymoon period when the team does a little better, the average points won by a team simply go back to where it was before the new manager was appointed. Changing managers after a string of bad results is simply a costly error made by the team owners.

The same happens with economists and analysts when they try to forecast company sales, earnings, or GDP. The more random (i.e. the less autocorrelated) the quarterly sales of a company are, the more analysts focus on the recent past and project that into the future. Even worse, the more random quarterly sales are, the more willing analysts are to project the latest sales trends far into the future. After a couple of sales distorted by supply chain disruptions, analysts are willing to go out on a limb and claim that sales will slump for the next 12 or even 18 months and that the economy overall will go into recession. After 30 years of peace in Eastern Europe a Russian invasion of Ukraine means that Putin will march right through to Poland and the Baltics, China will invade Taiwan next year and Iran will get into a war with Saudi Arabia.

Nonsense. The likelihood of any of these events happening is extremely small. It is not zero, which is why doomsayers can make a living and if these events happen, no matter how unlikely they are, these doomsayers will become global stars and make a fortune from speaking gigs and advising investors.

This brings me to 10 rules of forecasting that I emphasise all the time because they have made me a better investor. They won’t make you famous (in fact, they will actively keep you from becoming famous in today’s attention economy) but they will make you money.

In particular, I emphasise the following three rules:

2. Don’t make extreme forecasts. Predicting the next financial crisis will make you famous if you do it at the right time. It will cost you money and reputation at all others. Remember that there are only two kinds of forecasts: Lucky and wrong.

3. Reversion to the mean is a powerful force. In economics as well as politics, extremes cannot survive for long. People tend toward average, and competitive forces in business lead to mean reversion.

5. We rarely fall off a cliff. People often change their habits in the face of a looming catastrophe. But for that behavioral change to occur, the catastrophe must be salient, the outcome certain, and the solution simple.