I recently did a similar analysis for the stock market, but noted that these indicators (valuations, margin finance, and market capitalization as a proportion of the economy) have no time limit attached to them. Unlike the economic indicators, you can’t say that they suggest trouble within a given time frame.
Besides time frames, though, another way to assess success rates for signals is percentages. If a signal works four out of ten times, you can estimate that you have a 40-percent chance of some event occurring when the signal flashes. Again, multiple signals can be more reliable.
In light of our fast failure discussion the other day, let’s review the success of some signals at quantifying stock market risks.
Today, we’ll focus on one of my old standbys, moving averages. Rather than use the 200-day, as I usually do, we’ll use the 400-day. A longer-term trend indicator, it will give us fewer signals and be easier to analyze. We’ll call a signal valid if it results in a further market decline of 10 percent or more, and invalid if it does not.
First, let’s look at the S&P 500 Index. Using the 400-day moving average as an indicator, results were as follows over these periods:
Signals Valid Hit Rate
2005–2014 2 1 50%
1995–2004 1 1 100%
1985–1994 4 1 25%
1975–1984 4 2 50%
1965–1974 3 3 100%
1955–1964 5 2 40%
Total 19 10 52%
These numbers are subject to a bit of interpretation, as they were calculated on a monthly basis and in many cases involved bouncing around the average. Nonetheless, for our purposes, they are good. A break in the 400-day moving average typically has a 50-percent chance of leading to a further loss of at least 10 percent.
To get more confidence, though, we need to see whether this works in other areas as well. Let’s look at the Nasdaq Composite Index.
Signals Valid Hit Rate
2005–2014 2 1 50%
1995–2004 1 1 100%
1985–1994 4 2 50%
1975–1984 2 1 100%
Total 9 5 55%
The Nasdaq gives very similar results, although over a shorter time period.
Just for fun, let’s look at something completely different: the interest rate on 10-year U.S. Treasury bonds. I know these aren't stocks, but given the observed behavior of two different stock indices and the effects of interest rates on stock prices, it doesn’t seem out of the question that we might see a similar pattern.
Signals Valid Hit Rate
2005–2014 8 4 50%
1995–2004 7 4 56%
1985–1994 2 2 100%
1975–1984 5 2 40%
1965–1974 2 1 50%
1955–1964 6 2 33%
Total 30 15 50%
Based on two different stock indices and one bond interest rate, then, we see very similar results for the 400-day moving average indicator.
What does this mean?
By itself, probably not a lot. The 400-day moving average is useful—knowing there's a 50-percent chance of at least a further 10-percent decline is helpful for both portfolio management and trading—but not sufficient on its own to provide the kind of information we seek.
It is, however, a good start, and we’ll look at additional indicators in future posts.