I've seen a lot of worry since Monday about the idea that we suffered a 90% down day, the second such occurrence in the past couple of weeks.

Different folks define a "90% day" in their own way, but the concept is usually credited to Lowry Research and defined as a day when 90% of total volume goes into issues that were negative on the day, and 90% of all points gained or lost were lost.

Personally, I find that just using volume is sufficient.  In that case, according to my data we did get a 90% down day on Monday, but we barely missed it on June 15th (that was an 89% down volume day).  But let's just pretend and say that we've now undergone the indignity of  two 90% down days within two weeks of each other.

Even worse (supposedly), this is occurring as prices are coming off of a multi-month high.  So let's go back to 1940 and look for any other time that the S&P hit at least a 3-month high within the past couple of weeks, and also recorded at least two 90% down volume days.

The last signal, in July 2007, preceded the current bear market.  But not before the market rallied for a few months first.  Same with the signal before that.  And before that.  And before that...

If you look at the Average and % Positive fields after the instances, they're not too bad.  In fact, take a peak at the "3 Months Later" column...100% winning trades, with an average return more than three times greater than random.  Returns across the board were better than should be expected under normal circumstances.

Ironically, even though this has in fact been pretty bullish for the market going forward, it's getting nothing but negative press.  But traders are tripping all over themselves pointing out the bullish signal given by the "Golden Cross".

The Golden Cross is defined as the 50-day moving average moving above the 200-day moving average, in this case for the S&P 500.  I've never been a fan of moving average crossover systems, but let's just take a look at the facts.

The instances above are filtered by those that occurred when the 200-day average is declining, like it is now.  Overall, the returns going forward, up to six months later, were little better than random and not statistically significant.  In fact, in the shorter-term they were a little worse than random.  Only when we look out a year do we see some out-performance.

I suppose we could argue that most of the poor performers were concentrated prior to 1942.  Since then, the performance in the S&P three months, six months and one year later was pretty good after these signals, and well above average.  The S&P was positive 11 out of 13 times since then, with average returns of +5.9%, +9.0% and +17.8% respectively.

Overall, I would not be using the Golden Cross, or any other moving average crossover, as a signal in the broad equity indexes.  There is just very little evidence that they provide any additional value.

And I would not be too concerned about the multiple 90% down days.  I agree that heavy selling pressure *can* be a sign of a top, but obviously from the chart above that's not always the case.  I'll be more concerned if we cannot rally well from short-term oversold conditions (there has been some initial signs of that, but not definitive), and we cannot bounce from technical support levels (880ish is probably the biggest test).

 Disclaimer  |  Privacy Policy