No indicator can predict with 100 percent certainty when the stock market is going to crash, but here is one indicator that identifies the possibility.
The Hindenburg Omen was adopted by Jim Miekka, editor and publisher of The Sudbury Bull and Bear Report (6735 14th Street, St. Petersburg, FL, (813) 866 - 8682), derived from a New High - New Low indicator developed by Gerald Appel many years ago. It worked for a time, but Jim then made certain changes in its makeup and calculation which gave it greater reliability. Because it signals the possibility of a stock market crash, I suggested that Jim name it after that famous ill-fated aircraft associated with the word "crash."
Jim warned us that a Hindenburg Omen is only occurring when the 10-week moving average of the NYSE Composite Index, based upon a weekly Friday close, is in a rising trend. If the NYSE Composite 10-week moving average Friday close is down, the signal is invalid. Second, if the McClellan Oscillator is in a positive configuration (i.e., with a + reading, above zero), you will be best served to postpone any selling until the McClellan Oscillator turns negative. However, when a Hindenburg Omen signal occurs with the McClellan Oscillator in negative territory, "Stand not upon the order of thy going..." - Sell at once!
Hindenburg Omens are relatively rare. In checking them from 1997, going back to 1970, Jim found that they usually occurred prior to major declines, with relatively few false signals. Sometimes two or three signals occurred in a row, within days of each other. I consider that a signal expires 30 days after its last occurrence, unless a decline has taken place within that time. There was only one signal before the 7% drop in October of 1989, but there were several days of signals in both 1987, and again in June- July before the 20% drop into October, 1990. There were several before the 10% drop in March 1994.
So what are the New High - New Low numbers necessary for a Hindenburg Omen? We get a CAUTION if, on any given day, BOTH New Highs AND New Lows number more than 79. At this development, we must check for a signal as follows:
First, find the daily number of New Highs, and the daily number of New Lows, on the NYSE. Then take the lower of those two numbers.
Next, divide the lower of those two daily NYSE New High - New Low numbers by the total number of issued traded that day on the NYSE (these days that number will probably be somewhere between 3,300 and 3,500 total NYSE issues, or thereabouts. Over the years, the number will vary).
Third, multiply the resultant number by 100. If the result is LARGER than 2.2, you have a potential Hindenburg Omen.
Fourth, check that the NYSE 10 week moving average is rising. If it is then, Fifth, check that the McClellan Oscillator is negative.
If it is, then we have a Hindenburg Omen, a high risk stock market crash signal. SELL.
EXAMPLE: On Thursday September 1, 2005, NYSE New Highs totaled 282, while New Lows for that day were only 26. Taking the lower of those two numbers, 26, and dividing it by the number of issues traded that day, 3,420, we get .0076. Multiplying that by 100, we get 0.76 - far less than the 2.2 necessary to render a Hindenburg Omen SELL signal.
However, suppose that instead of 26 New Lows, we had 79 New Lows. Well, dividing 79 New Lows by the number of NYSE issues traded that day, 3,420, we get 0.0230994. Multiplied by 100, that equals 2.30994, MORE than 2.2, so that hypothetical lower number of New Highs versus New Lows would have warned of a potential signal. If the 10 week NYSE moving average is also rising, and the McClellan Oscillator is also negative, then a Hindenburg Omen SELL signal has been generated.
One final note: Do not rely on YAHOO! Or other electronic New High - New Low figures, which are frequently off. Use the later newspaper figures to be accurate - and double-check those, as well!
[Editor's note: Bob McHugh has provided a detailed analysis of the Hindenberg Omen from 1985 - 2005: http://www.safehaven.com/showarticle.cfm?id=3880]
Best of luck!