In December, JPMorgan predicted that investment in US stocks would gain 5% this year, economists expected US 10-year bond yields to remain around 2%, and Goldman Sachs raised the possibility that Bitcoin would reach at $100.000. However, six months later, investment in US stocks is down 20%, the 10-year bond yield is at 3% and bitcoin has lost more than half its value, standing at $21.000. . The truth is that Wall Street professionals have a terrible track record of forecasting, and all we need to improve is to follow five simple steps inspired by the book Super Forecasting by Philip Tetlock.
Step 1: Define what we are trying to forecast
Let's imagine we come across a headline that says: "Stock investing is significantly overvalued and about to crash." The first thing is to understand what “stock investing is about to crash” really means. What type of actions are we talking about? What percentage loss is considered a “crash” and over what time period? To evaluate the accuracy of this statement, we first have to define what exactly we are predicting. So, in this case, you might decide to rephrase the problem as: "What is the probability that investment in energy stocks will be down more than 20% within a year?"
Step 2: Break down the problem
To calculate the probability of this problem, we have to recognize that it is made up of a couple of parts. By simplifying the problem into smaller parts we can quickly formulate a surprisingly accurate forecast in stock investing. The original statement consists of two parts: «markets are significantly overvalued y are about to fall«. Therefore, there are two elements that can be analyzed. The probability that the investment in stocks is "about to fall" and the probability that the investment in stocks will fall when they are known to be overvalued.
Step 3: Strike the right balance between internal and external opinions⚖️
Tetlock discovered that superforecasters see things in two ways, known as “internal” and “external” points of view. Let's start by raising a question of external vision. That is, they aim to eliminate emotions and observe hard, cold data. They gauge the frequency with which outcomes of this type occur in situations of this type by observing the facts. Returning to our market example, we could look at how often the S&P 500 has lost more than 10% over a one-year period. This tells us that stock investing has lost more than 10% only 15% of the time since 1996. The problem is that this case ignores important information that makes this situation somewhat unique. A single factor, in this case, could be how high the valuations are at that time. If we take these valuations into account by looking at the relationship between forecast P/E ratios and one-year returns (bottom left), the probability of a 10% loss rises to 30%.
However, the relationship is quite weak and there are almost the same number of times that investing in stocks ended up much higher. So an estimated range for the probability of "outside vision" is actually between 15% to 30%. Of course, no situation can be completely summarized with a number. The potential impact on the economy if Russia withdraws European gas, or if the US Federal Reserve changes its tone on interest rates, requires a judgment call. This part of the analysis, which involves using judgment to evaluate the specific aspects of each case, is what Tetlock calls "insight." But there are two problems with internal vision:
(i) It is prone to biases as it depends on the main market narrative and is driven by factors such as mainstream news.
(ii) Our brain naturally leans towards internal vision, full of attractive details and where it is easier to create a good story. To combat these biases, Tetlock explains that it is important to use the "outside view" as the primary anchor and use the "inner view" to adjust it. In our example about the probability of stock investment crashing, the 15% to 30% probability is our external view, so even if our internal view assessment is quite bearish, we are unlikely to see a change. of opinion driven by emotions. The probability, therefore, could be 30% or 40%, but it is unlikely to be as high as 80% or 90%, as the original news headline suggested.
Step 4: Update our forecasts frequently♻️
Tetlock found that good analysts update their forecasts much more frequently than average analysts. In other words, when facts change, so should our forecasts. For example, the original probability regarding the probability of a recession across Europe should be adjusted now that we know that Russia could shut off European gas and trigger a recession in the region.
Step 5: Learn from mistakes
If we had to guess the quality that, according to Tetlock, makes a good analyst, what would it be? Yes, at first we would all think of intelligence, but Tetlock found a quality approximately three times more powerful. A relentless commitment to updating his perspective and focusing on self-improvement. That means spending time analyzing both successes and failures, focusing on what was done right (or wrong) and what could have been done differently.
