Intermediate-term Clues from SPX R-Squared & Linear Regression Slope

Linear Regression utilizes what’s known as the ‘least squares’ method to plot a statistically straight line through price. Since the indicator is “fitting” a line to the data points rather than averaging them, the Linear Regression line is more responsive to price changes than a moving average. This chart (snapshot | interactive chart) highlights two technical indicators based on the linear regression technique – Slope & R-Squared. Slope tells you which direction the linear regression line is pointing (above zero is bullish, below zero is bearish), while R-Squared quantifies the strength of the trend. Specifically, r-squared values show the percentage of movement that can be explained by linear regression. Over the last four days, the 14-bar R-squared indicator has been running over 90%, which means 90%+ of the movement of the S&P can be explained by linear regression. The other 10% is considered random noise. As you might imagine, it’s unusual to see R-squared hold at such a high level for an extended period of time. I thought it would be interesting to look back at previous instances in which the SPX R-Squared held above 90% four consecutive days along with a positive Slope. I found a total of 29 separate occurrences since 1970 (that didn’t occur within a month of a previous signal)…

Four Consecutive Sessions w/ SPX 14d R-Squared >90% & Positive Slope
04/07/11… S&P500 ??? one month later
03/16/10… S&P500 +4.4% one month later
11/12/04… S&P500 +1.2% one month later
01/02/04… S&P500 +2.4% one month later
07/08/98… S&P500 -7.3% one month later (*)
03/26/98… S&P500 +0.6% one month later
01/23/97… S&P500 +3.1% one month later
11/27/96… S&P500 +0.2% one month later
05/24/96… S&P500 -1.4% one month later
02/01/96… S&P500 +0.9% one month later
09/15/95… S&P500 +0.2% one month later
05/09/95… S&P500 +1.8% one month later
09/01/93… S&P500 -0.9% one month later
02/12/91… S&P500 +2.5% one month later
05/18/90… S&P500 +0.6% one month later
12/09/88… S&P500 +1.2% one month later
06/18/87… S&P500 +2.9% one month later
01/20/87… S&P500 +6.1% one month later
02/13/86… S&P500 +8.8% one month later
11/18/85… S&P500 +6.0% one month later
05/22/85… S&P500 -1.0% one month later
01/25/85… S&P500 +1.1% one month later
08/31/82… S&P500 +1.8% one month later
08/09/79… S&P500 +2.1% one month later
08/09/78… S&P500 +0.9% one month later
06/21/77… S&P500 +0.8% one month later
01/14/76… S&P500 +3.8% one month later
04/16/71… S&P500 -1.2% one month later
01/29/71… S&P500 +1.2% one month later
12/09/70… S&P500 +3.0% one month later

The string of elevated R-Squared values led to a higher S&P one month later in 24 out of 29 occurrences since 1970, or 83% of the time, significantly above the 59% random chance for a higher S&P twenty sessions later in the same time  frame. Only one instance led to a loss of more than 1.5% while thirteen led to a 1.5%+ gain.

Categories : Archived EOD Commentary, End-of-Day Commentary

 Print This Post