Hulbert points out the obvious flaw: The Super Bowl indicator is based on a historical correlation, but there is no compelling reason to think that it will continue to be accurate. He goes on to point to studies showing that the statistic that shows the highest degree of correlation with the performance of the S&P 500 is, you guessed it, butter production in Bangladesh.

But this got me to thinking: Is the Super Bowl indicator really any more ridiculous than technical analysis in general, or is it just an extreme example that shows how silly the whole field is. Before angry chartists send me hate-mail, let me be very clear: I am not a chartist but I do know several who have made money with the technique, and I am happy for them. But to paraphrase Burton Malkiel, I see no reason that technical analysis should work and, in most cases, it doesn't.

When examining investing strategies (along with strategies for life in general), we should look not for evidence of past success, but a compelling reason that a strategy works. For instance, Warren Buffett's strategy of buying well-run companies with strong competitive advantages at cheap prices relative to their cash flow makes intuitive sense. And it has yielded strong results in the past and seems likely to continue to do so. But what about homing pigeons and bullish haramis? Is there any reason that Fibonacci sequences would be the key to picking stocks that will beat the market? I

For an excellent book of interesting examples of statistical analysis and the supremacy of causal relationships over correlations, pick up a copy of Steven D. Levitt's book Freakonomics. While it's not an investing book per se, I think investors can draw numerous lessons from it.