"In his excellent book "
The Signal and the Noise," there's a section where Nate Silver covers Bayesian reasoning - essentially, a formula for estimating the probability of [X] based on some new piece of information, given the likelihood of [X] *prior* to receiving that new information. I know, I know; your eyes are glazing over at the mere
thought of mathematics, but bear with me for a second.
Silver outlines a storybook scenario wherein a woman comes home to find, in her boyfriend's bedroom, some unmentionables of unknown origin. What, then, is the likelihood that said boyfriend is being unfaithful?
The answer is much lower than you would
intuitively think, and that is a problem if you are an investor, because you're likely putting far too much weight on incremental data points and far too little weight on "priors" - i.e. the underlying "base rates" of how likely things are.
A less lascivious but perhaps more useful example: if you believe (based on good evidence!) that it's extremely unlikely that someone will be able to shout a word and make it start raining, and then someone does exactly that 10 times in a row, you
still shouldn't believe that they're actually a rain god - they're just really, really lucky.
As it relates to investing, historical data largely suggests that
investors overreact to data points... which is where the opportunity for value investing arises: if you can develop a strong knowledge of the fundamental nature of things and deeply understand base rates, you'll have the ability to swing hard when you see a compelling opportunity."
