Fooled by Randomness, part 1
I don’t read too much into Mark Hulbert’s commentary over at Marketwatch. Although a contrarian he chooses to place a large emphasis on correlations that simply do not stack up imo.
Take for example the Dow Theory, a dated system that makes judgements based indicators one of which is divergence of Dow Transports Index from DJIA – begging the question why would Transports nowadays have advance knowledge of the market demand any more so than the producers they service? Producers have whole departments dedicated to forecasting demand for their goods. By the time a profits warning comes in at Transports you can be sure producers have already given appropriate guidance on their expectations. It’s part of market listing regulations.
Obviously when the Dow Theory was being developed markets were not quite as developed. So why use it today? Smacks of column filler rather than serious financial insight.
However he’s completely correct when he calls out the current chorus of praise by analysts for a leading indicator known as the ECRI WLI. It’s not the value itself that’s being lauded as ‘prescient’ by the investment community but the growth rate – which currently stands at -4.7% with a value of -10 a near certain sign of recession (and in this market’s parlance, the advent of the 2nd downer in the double dip).
Never mind the fact only 3 out of the last 7 -10 readings has the WLI actually succeeded in being a leading indicator of government recognised recession, Hulbert makes a more pertinent point. That relying on a metric that only has been backtested 7 times is completely useless in a statistical analysis. It’s one of the fundamental problems with market analysis today. The other – seeing correlation in numbers regardless of relationship – being Hulbert’s main weakness, but more about that another day.