I’ve been reading Fooled by Randomness and its followup, The Black Swan. Both are must reads for anyone that has money in the financial markets.
There are several themes in the first book, but the two that stood out the most for me are: 1) Stock prices basically behave as a random process; and 2) Humans are hard-wired to recognize patterns, even where there are none. A couple of the more humorous stock market patterns you may have heard about include the Hemline Theory (prices are correlated to the length of skirt hemlines), and the Vanity Fair ad pages (prices are inversely correlated to the # of VF ad pages). These two theories sound pretty absurd, but they illustrate that if you look hard enough it’s possible to correlate a random process with just about anything.
Some of the more “legitimate” stock price pattern recognition algorithms out there include poring over financial statements (a.k.a. fundamental analysis), drawing trendlines on historical charts (a.k.a. technical analysis), or maybe something more exotic like custom written AI algorithms. All of these have some degree of success, but anyone who has actually tried to use techniques such as these can attest to the fact that correlations may hold up for a subset of the random data, but the pattern undoubtedly fails.
The important point that gets almost zero focus in any kind of financial media is that success in the stock markets depends very little on the quality of your pattern recognition algorithm. The difference between having one that’s 70% accurate and one that’s 99% accurate is moot, because they will both fail eventually. The larger factor that determines long-term success in the markets is what’s done to manage the risk for the failure case.
One of the most common schemes that gets touted as risk management is to diversify and have a long term investment horizon. The argument is that historically, the stock market indexes have consistently returned something like 7% per annum for periods of time measured in decades. That sounds like a pretty compelling argument, but that leads to one of the themes in the second book, The Black Swan.
Consider for a moment, the hypothetical scenario of the allies losing World War II and the west never becoming the technological and economic superpower that it is today. Imagine what the 20-30 year returns would look like for somebody who bought the North American indexes in 1945. This is the kind of “highly improbable event with a massive impact” that the second book discusses. The point being that proper risk management should not rely solely on something that in theory has a very low probability of occurring, because it probably has a higher probability than you think.
Both the books are well-written and easy to read. They don’t give specific financial advice, and in fact there’s very little content that’s actually finance related. They’re more general interest than anything and can thus be applied to other areas. I just found them enlightening when applied in the context of the markets. Highly recommended, check them out if you have a chance.