I saw and interesting piece of quant research by a very well respected investment bank which looks at the performance before and after a rate hike. It draws so conclusions and admits some mixed results in its analysis. The conclusion and headline is: “The Trade Weighted US Dollar Historically Underperforms After a Rate Hike”. Some clever maths is used to come to this conclusion. “World Class Asset Manager’s Innovative Solutions” research HERE.
The analysts who produce this are good at maths. They will be very well educated and qualified. But only deep in their research do they mention they are drawing their conclusions and using statements like “History Indicates” from just eight samples of the cycle since the 1980′s. The sample is eight and yet averages, trends and Standard Deviations are extrapolated and history indicates conclusions drawn. They do know that you can extract data from any series but for the answer to be robust you need a sample which is meaningful. Eight samples is not enough to conclude anything.
I am no mathematician, but I do know that even if eight out of eight samples in a series are the same it means nothing for the likelihood of the future. Eight Heads in a row still leave us with a 50/50 probability fo a Head on the next coin toss. This is because the events are independent. Given that there are no eight out of eight events in their test. Surely that should be their conclusion? Most likely there is no connection between the direction of the Dollar and rate hikes. No story here though.
We Demand Forecasts
The interesting thing here is, they certainly know this but do not tell us that their conclusions and forecasts for the Dollar have no meaningful statistical relevance. How come? They have to do the best they can with their eight samples and lack of validity is not going to stop them making forecasts. How do they get away with it? We demand forecasts. Even when a cursory look at them tells us they are not likely to be meaningful. We want forecasts. We need confirmation of our views and other behavioural biases. They make us blind to the bleedin’ obvious. This means nothing but it agrees with what I think. This is the service some quants provide: an academic looking view which agrees with me.