Monday, April 28, 2014

Of Financial Advisors and Algorithms

If you've ever perused the promotional material of a stock market advisor/prognosticator you may be impressed at how often their charts and models seem indeed to fit the history of market activity (…of course when you give them YOUR money, suddenly their predictions aren't quite as stellar). This article (h/t to Jennifer Ouellette) recounts the ubiquitous idea of "backfitting" or "backtesting" (a "subtle mathematical con" as the author calls it)… let's just say that statistics, hedge funds, and stock modeling aren't always the most forthright mix:

The actual paper on which the article is based is here:

And finally, a related blog, "The Mathematical Investor," is here:

It's no wonder that "past performance is no guarantee of future results..."

No comments: