Many in the community of stock traders that rely on technical analysis are clamoring about the Hindenburg Omen signal that occurred on August 12, 2010. The Hindenburg Omen refers to a series of criteria that supposedly occur prior to all significant stock market declines and crashes. A signal is said to occur if the following happen on a single trading day on the New York Stock Exchange:
1. The number of New York Stock Exchange trading companies whose shares close at 52 week highs, as well as those that closed at 52 week lows, must each exceed 2.2%of the total number of companies that traded on that day;
2. The New York Stock Exchange ten week moving average must be rising;
3. The McClellan Oscillator must be negative. It is calculated using an exponential moving average of the daily differences between the number of companies whose stock prices rose less the number of companies whose stock prices declined. The exponential moving average gives recent data greater weight. If the oscillator provides a negative number, it is considered to be bearish in outlook since it reflects money leaving the market; and
4. The number of New York Stock Exchange trading companies whose shares close at 52 week highs cannot be more than twice the number of companies whose shares close at 52 week lows.
Others have added to or revised these criteria in order to increase the accuracy of the Hindenburg Omen or have required that it occur more than once in quick succession in order for the signal to be a “confirmed” one.
A Hindenburg Omen is said to have preceded every stock market drop of 15% or more in the past 25 years. Does this mean that investors should sell all of their stock holdings and head for the hills? Probably not. While Hindenburg Omen signals have occurred before market plunges, they most often result in Type I errors (false positives where the signal occurs, yet the stock market does not decline). These false signals occur 75% of the time, according to the Wall Street Journal.
Another concern that critics of the Hindenburg Omen have overlooked is that this indicator is relatively new. It is attributed to mathematician Jim Miekka who formulated it in 1995, although initial formulations of similar signals date back farther. Since the Hindenburg Omen is only about 15 years old, there is very little data upon which to analyze its accuracy. There have only been 27 confirmed signals in the past 25 years and many of these predate Miekka’s indicator.
Signals that occurred prior to Miekka’s formulation of the indicator are troublesome. The odd assortment of the signal criteria (e.g., why 2.2% of new highs and lows, why a rising ten week moving average?) suggest that Miekka resorted to backtesting. Backtesting involves evaluating signal criteria on the basis of prior trading data. Criteria are altered until the best fit with prior data is obtained. In this instance, Miekka likely adjusted his Hindenburg signal criteria until the maximum number of possible signals occurred just prior to major stock market declines. However, such a theory relies on the assumption that what occurred in the past will happen in the future. It also often results in what is known as data-mining bias, the practice of determining signal criteria by extensive searching through a dataset until one finds a combination that appears to be significant.
The accuracy of the Hindenburg Omen in the recent past has been intriguing. However, the predictive value of the Hindenburg Omen signal can only be evaluated after more time has passed and a statistically significant number of signals have occurred. In the meantime, do not bet the farm that a double-dip recession is imminent.
