How do you predict the stock market?

I recently had a discussion with a colleague regarding best predictors of the stock markets. Having followed this subject both academically and as an intellectual pursuit, it is very interesting to note how the human mind is conditioned to create relationships using short term data where none exists. Detailed beautifully in "Fooled by Randomness" by Nassim Taleb, there exists a multitude of such phony correlations. "A mathematician plays the stock market" also has similar stuff.

 

Taleb mentions in his book an example of this. According to the research of Cal Tech professor David Leinweber, the "single best predictor of the S&P 500's performance" over the period 1983 – 1993 that he found was … hold your breath … butter production in Bangladesh. The lagged correlation is almost unity, which means that a y% rise/fall in butter prices in any year is followed by exactly a 2y% rise/fall in S&P 500 in the subsequent year.

 

So, if you tracked butter production in Bangladesh during that time, you could have made a killing on the stock markets and everyone would've taken you to be a genius.

 

Tough luck you didn't know this before hand.

 

That's the problem with back testing. It's what they say, "hindsight is always 20-20".