An Unexpected Response to "Moneyballing the Financial World"
David Gartner's recent artilce on "Moneyballing the Financial World" was inspiring. I work in the field of Analytics and it really is a game changer for companies to increase profits and revenues by simply knowing the stats and using them to optimize all aspects of their business.
But when it comes to prognostication, one has to be very careful - especially if what you are predicting is largely random. For purely random events, the best prediction is often wrong more often than it is right - in fact, that is the fundametal principal behind Casinos around the world.
Are markets completely random? Leonard Mlodinow's book "The Drunkard's Walk: How Randomness Rules Our Lives" shows that they probably are (there is a great section where he talks about how Mutual Fund past performance is a horrible indicator of future results).
The idea that some prognosticators have the "secret sauce" to make you money, while others are just blowing hot air -- is just plain false.
What will a scorecard really tell you? Past performance. Does that tell you how good they will be in the future? No. Let's revisit the Casinos -- people do win - does that mean they know something everyone else doesn't know? No it doesn't.
Many of you will totally disagree with me. Some will say "There are fundamentals that can be followed" or that "Some parts of the market aren't random - investors that find these non-random facets will do well" or that "I trade based on market sentiment and momentum, which isn't random but rather about the human condition". Even my own strategy "Bet on historical low PEs" has no real predictive power - even though it is wildly successful when you backcast it. No strategy is safe. Everyone will fail at some point or another.
Final point: Using historical information and scoring is all we have to measure analytsts, strategies and predictions, just remember that it isn't really that good of a predictor for the future.