On Lies and Statistics

 |  Includes: DIA, QQQ, SPY
by: Matthew Claassen

Lies, Damn Lies and Statistics

I often enjoy using statistics based on historical behavior to support or contradict a position. In recent days I have heard from multiple sources a statistic I find of interest. The statistic, as relayed to me, suggests that when the S&P 500 was down more than 5% during the month of May, the follow up performance was positive through August in seven out of nine occurrences.

I am not sure what time period was used for the above argument. At first glance, 5% seems a rather small percentage to use and, considering normal volatility of the markets, can be subject to too much statistical noise to determine a trend. In looking at the S&P 500 from 1961 to the present, I have found the index produced 528 five percent swings. That averages to just over once per month, with the highs and lows fairly evenly distributed from over each of the twelve months.

When we consider that the market decline from April’s high to this week’s low is 13.64%, perhaps we could examine the current decline from another perspective. We find 108 ten percent swings in market price over at least a five day period since 1961. Of note; only twice has a market decline of 10% resulted in a 10% rally that started during the month of May. The first was in May 26, 1970 where after a 31% decline the index rallied 16.54% over two weeks, peaking on June 09, 1970. The second was this year, May 06, 2010 where the market decline of 12% was followed by a 10% rally into May 13th.

Market lows of 10% or more are also unusual during the month of June, with only two occurrences since 1961 (1962 and 1965). When looking for market lows that generate at least a 10% rally, fewer market lows have occurred in the months of April, May or June than any other three month period of the year. If we upped the percentage changes to 15% or even 20%, we can easily state the market changed trend fewer times in the April through June period than other three month window. Only one in eleven rallies or corrections of greater than 15% have ended during this time period.

So, what does all this mean? Well, prior to October, 2007 we could have argued that the US markets had never made a bull market high during the month of October. Now, we say that October, 2007 was one of the most important bull market peaks in U.S. history. Statistics can be used to support or refute almost any market trend, and regardless of how valid a statistic may be, there are always exceptions.

For now, with a decline of more than 13% and the February lows violated around the globe, the market trend is negative. While a relief rally is possible, it would be unusual, but not unprecedented, for that trend to change during the months of May or June.

Disclosure: No positions