Get the direction right not the precise forecast

 

"I'd prefer to be approximately right rather than precisely wrong."  - Keynes 

Of course, I would like to be right with precision who wouldn't?  However, precision is a false goal. The first job of a good forecast is to get the hit rate right. Is the market going up when you say the market is going up. Nevertheless, an important part of modeling is getting the probability right with respect to direction. I cannot give you a point estimate, but I would like to know what the probability of the direction. 

Precision allows for variable sizing, but this only works if you get the direction right. So, I will follow the preference of Keynes.