Friday, December 27, 2019

What financial forecasters can learn from weathermen



Weathermen are considered the best forecasters for any field of study. We know because there is a lot of data for testing and weathermen are good at using probabilities. They often provide a probability forecasts so that phrases like "there is a 75% chance of rain" can actually be tested.

Studies have shown that a seven-day forecast can accurately predict the weather about 80 percent of the time and a five-day forecast can accurately predict the weather approximately 90 percent of the time. However, a 10-day—or longer—forecast is only right about half the time. 

Their forecast accuracy should be kept in mind when we hear end of year forecasts from many market strategists. The likelihood that their prognostications will prove correct is slim.  Financial forecasts are notoriously poor. 

Models that try and explain financial data have R-squares that are often in the single digits, and while models actually improve over the longer-run, the amount explained is still low. Out-of-sample predictions show further declines as measured y mean squared error. Likely, there is evidence that even a model with low explanatory poor has some value for asset allocation; nevertheless, we should place limited weight on these year-end strategy pieces. 


What should these end of year outlooks tell us? I would highlight three things that should be useful.

1. Provide analysis of where we have been over the last twelve months. A good end of year strategy piece should recap the year and explain why any forecasts were in error or were correct. What surprised investors and why should this not occur again in 2020. 



2. Highlight different issues that can cause change in the coming year. The year-end outlook should focus on the potential issues that will change the current state of markets, the market catalysts. 

3. The outlook should consider different scenarios and outline what will be the impact if those scenarios are realized. The scenarios can be in the form of a base, optimistic, and pessimistic case.  Each should be given some probability weight. 

Most important is that the outlook has precision and some measure of likelihood. Forecast should also be given a path and a horizon. It is not enough to just say that stocks will do better than bonds, or credit spreads may widen. Precision requires thought.  Be like the weatherman who will give a specific forecast, a likelihood, some rationale for what may go wrong, and a narrative for why the forecast will work.

No comments: