I interrupt this political commentary for some brief finance blogging.
Over the weekend, the media seized on something called the Hindenburg Omen: a syzygy of financial statistics that supposedly portends stock market collapse. The list of indicators that have to align just so to produce the Omen is so long I can’t even follow it myself. Here is how CNBC puts it:
The omen is triggered when more than 2.2 percent of the NYSE Composite Index’s stocks are finding new highs while another 2.2 percent or more of the issues are creating new lows. The lesser of the two numbers has to be larger than or equal to 69. The NYSE 10 also has to be rising and the McClellan Oscillator — a measure of market breadth based on advancing and declining stocks — has to be negative on that day.
If all of those criteria are met then the warning bell sounds.
Got that? Me neither. Nor do I plan to waste my time figuring it out, for I can already tell that the Hindenburg Omen — like other examples of “technical analysis” — is buncombe.
Take any random data set, and one is bound to find some striking coincidence or other. The larger the dataset, the more inevitable yet individually improbable those coincidences will be. The stock market provides endless examples. You’ve probably heard them before: “The stock market always goes up the day after NFC linebackers intercept more passes than AFC safeties,” or “A lunar eclipse on the West coast means that the market will go down.” The technical name for this is data snooping. Keep testing a set of random data, and eventually you will find a rule that fits.
Sometimes we are savvy enough to realize that these past correlations do not predict future moves in the market. But, for some reason, when the correlations involve purely financial stats like trading volume or stock price movements, we are more easily suckered. The academic literature has confirmed what theory would predict: take any “technical indicator” such as the Hindenburg Omen, and it turns out that, even though it may seem to account for stock market outcomes during a given backward-looking sample period, it does no better average in subsequent periods. (The Hindenburg Omen isn’t even a very good piece of data snooping — even with past data, it only “works” about 25% of the time.) Nevertheless, to this day, many people on Wall Street make their living as “technical analysts,” even though their methods have no more merit than homeopathy or astrology.
None of this means that the market won’t crash in September. Still, if your broker calls you to make sure you’ve read the reports about the Hindenburg Omen, you should fire him. He makes money by convincing you to make as many (unnecessary) trades as possible. “Technical analysis” is just another way for him to egg you on.