S&P 500 Versus Its 6 Month Moving Average

Moving Average Ratio

One simple way of determining if the market is "overbought" or "oversold" is to compare it to its 6 month trailing simple arithmetic average. In other words, in a rising market that is getting ahead of itself, this month's close would no doubt be high (in percentage terms) above the average of the prior 6 months of closes (including this month). This analysis would be confirmed by an inverse relationship between the ratio of the close to the moving average (actually the percentage difference between the close and the 6 month moving average) and subsequent 12 month S&P500 performance. An opposite analysis would be that during times of high market momentum, a high ratio of close to moving average would yield poor subsequent performance.

Perhaps both are true, and that at times momentum begets momentum and at others extremes in the ratio mark turning points and periods of regression to the mean. This may explain why the overall pattern is so murky:

S&P 500 % above or below its 6 month arithmetic average versus next 12 months change, June, 1945-May, 2007, n = 725:
		Earnings yield			Next 12 month change in the S&P 500:
Percentiles:	and	From:		To:
Min-25%	181	-21.33%		-1.32%	10.11%
25-50%	182	-1.32%		1.91%	5.16%
50-75%	181	1.91%		4.88%	8.35%
75-Max	181	4.88%		14.24%	11.19%
		Correlation:			1.8%
All:	725		Average:		8.70%

Note that how far or below the S&P 500 is from its moving average tells us little about how the market will perform over the next 12 months. At the extremes the S&P500 moving average ratio seems marginally useful at identifying periods of excessive profitability (11.2% whenever the S&P500 moving average ratio is between 4.9 to 14.24% above its moving average, and 10.1% whenever it is 1.32% or more below it). Both the momentum camp (who would predict values high above the moving average preceding strong S&P 500 gains) and the contrarians (who would predict closes far below the moving average would lead to a reversion to the mean, and therefore strong performance) are partially vindicated by the historical record. However the correlation is a paltry .018.

Perhaps the time period is too short, and that 12 months is too long of a follow-up period. Yet shortening the follow-up period to 1 month - that is, measuring the subsequent 1 month change in the S&P500 - leads to even more splotchy results:

S&P 500 % above or below its 6 month arithmetic average versus next 1 month change, June, 1945-May, 2007, n = 724:
		Earnings yield		Next 1 month change in the S&P 500:
Percentiles:	n	From:	To:
Min-25%	181	-21.33%	-1.32%	0.75%
25-50%	181	-1.32%	1.92%	0.61%
50-75%	181	1.92%	4.88%	0.32%
75-Max	181	4.88%	14.24%	1.01%
		Correlation:		0.004
All:	724		Average:	0.67%

What this means is that at the extremes, the subsequent 1 month rate of return is not good, as is true when looking at the next 12 months S&P500 performance. In the short term, momentum begets momentum. The strongest average monthly gains follow months in which the S&P 500 closed 4.88% or more above its 6 month moving average. However, once again the correlation is very low (.004) so building a trading system around such an observation should be done with great caution.