Book Review: The Signal and the Noise
Nate Silver's book, The Signal and the Noise, is among my fave non-fiction reads of the year. Highly recommended, especially for anyone that does any kind of modelling/forecasting as part of their work (as is the case for me).
Silver made a name for himself doing baseball forecasting, then moved on do playing professional poker during the 'poker bubble', then came onto the mainstream radar by pretty accurately predicting the 2008 election.
In his book, he looks at the poor track record many branches of science & business have in making predictions. He also points to some areas of success. In running through his examples (baseball, weather, politics, finance and earthquake prediction, to name a few), he looks at a variance of available historical data, signal to noise ratios, etc. He gives a nice overview of Bayesian principles of taking bias into account.
I don't want to give it five stars only because much of his work has been covered by others (e.g. Taleb's Black Swans for example), but it's a great book nonetheless. It made me realize some of the flaws in much of the modelling that we do in my work, and gave me some tools to (hopefully) improve.
The Signal and the Noise: Why So Many Predictions Fail but Some Don't