Butter production in Bangladesh

In the 90's a US economics professor ran a study to try and find a leading indicator for the S&P, he used all the world macroeconomic data he could get his grubby little mitts on.

The results were that in the period of 1983 to 1993 the best leading indicator of the S&P 500 was butter production in Bangladesh! Yes that's right, if butter production went up the S&P was sure to follow slightly after, and if butter production fell the S&P better watch out!

Something GammaJammer said in this thread ( http://www.trade2win.com/boards/forex-discussion/66880-tick-volume-useless.html ) made me remember the above example. Do not get correlation mixed up with cause and effect.

So if you are backtesting and you find an amazing correlation there is a good chance that you are about to trade the equivalent of butter production.

The only way a back test can have the power to predict future movement is if you are backtesting a genuine market occurance, perhaps you happen to know that HSBC move a few billion USD into GBP at the exact same time and day each year in order to pay its dividend, you get the idea.

Remind me of an old joke. Can't remember it completely so gonna ad lib a bit...

Two students were talking to each other:

1: You know, i used to think that if Event A followed Event B then surely Event B must have caused Event A!
2: You don't believe that anymore?
1: Not since I took my Causality class, where we proved that this isn't necessarily the case...
2: Ah! So you could say that you took the Causality class and that changed your opinion on the Event A/Event B scenario?
1: Well...not necessarily...


I'm sure the original was 10x better but you get the point.