I'm always amused when financial gurus attempt to predict the future by looking at the past. I recall looking at the monthly returns of the S&P500, over a 10 year period (from the 1950s to the 1990s). From these 600 monthly returns I select 120 of 'em. (That makes 10 years, right?) Then I plot the evolution of a portfolio using this random selection of 120 returns. Then I pick another random set of 120 returns, Then I get this (where the white curve is the result of a random selection): Now comes the big question What will the next ten years look like? In spite of this distain for predictions, I enjoy doing it meself. For example, I look at the gain made by some stock over the past month. Then I look at the historical monthly gains for that stock. Then I identify the historical "best fit". Then I see what happens in the next month (historically speaking). Then I use that to predict what'll happen in the next month, starting today. My brother-in-law owns CBQ. It invests in Brazil, Russia, India and China. It closed yesterday at $27.75. With optimal confidence, I predicted it'd hit $31 in the next 2 weeks: I forgot to mention this one, where you're given some historical (weekly) data and asked to guess, like this: Then you get the actual (historical) continuation, like this: That ain't easy !! |
Tuesday, October 6, 2009
Predictions
Labels:
Predictions
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment