The 2018 FIFA World Cup is over, with France crowned victor. At the start of the tournament, Goldman Sachs predicted that Brazil would ultimately triumph, in a replay of the 2002 final when they beat Germany, who entered 2018 as the defending champions.
Brazil, France, Germany and Portugal would make it to the semi-finals, Goldman said, with Brazil beating France and Germany beating Portugal, before a score of 2-1 in the final would take the trophy to Latin America.
Nothing even remotely close to the prediction happened. The company tweaked its model and re-ran its simulations as the tournament progressed, and continued to miss the mark. After Germany crashed and burned in the group stages, Goldman proclaimed Brazil would beat England in the final. Its ultimate precidtion was a Belgium vs England final, but reality dealt us a France vs Croatia match.
The world’s most famous financial firm had no idea what it was talking about.
Contrast this with 2014, when Goldman predicted Brazil, Germany, Argentina and Spain in the semi-finals, and a Brazil vs Argentina final. Three of the four teams it forecast made it to the semis, and one of them made it into the final. So much better, yes?
So what happened this time?
Medianation LLP believes it was an overdose of data. Goldman used 1 million variables to run its predictions for 2018; it used far fewer for its 2014 predictions.
To be sure, Goldman did even portray the possibility of Croatia in the 2018 final, and even gave France a chance. But that came much later in the tournament, when the odds of getting something right always increase since the number of teams have decreased significantly.
The story has a moral for people who are forecasting. And even for us at Medianation LLP. Too much is too bad. The next time you — or we — catch yourself (or ourselves) overthinking something, it’s time to recall Goldman and its predictions. The point isn’t that Goldman was wrong. The point is the end result of overthinking. You to grapple with unnecessary complications and ultimately end up with the wrong conclusions.
Keep it simple. Get it right. Make progress.