In market analysis, price and volume are the most widely used data. Cycles may vary, but open-high-low-close charts with synchronous volume are ubiquitous. There are variations, such as point-and-figure and candlestick charts, and then there are data that include a myriad of advance/decline and options statistics, group averages, sentiment figures, and so on. Still, the data is finite, while the indicators that can massage the data are endless.
Thus, the analyst has a choice. He can shotgun a data stream with multiple indicators each supposedly yielding slightly different information, hoping to weigh the evidence to glean clues to market opportunities, or like a surgeon precisely applying his scalpel, he can use select indicators on different data streams that are unique at providing the market information he needs. However, even to the eyes of an experienced analyst, a plethora of market indicators on a given data stream is rarely the best solution.
Consider an ice fisherman. He can cut a single hole in a lake of ice and then drop his fishing line with 20 baited hooks on that single line. Or, he can cut 20 different holes in the ice in different places around the lake, each with a line and hook of its own. Similarly, the market analyst can apply 20 indicators to the same data stream, or he can select 20 different data streams upon which to apply a single indicator.
The second approach gives the analyst a better chance of arriving at conclusions about valid market anomalies. The conclusions drawn could suggest confirmation of a market theme, or premise, rather than simply a repetition of results from the same data.
DIFFERENT DATA, INDICATORS
In “Options redux: 25 years into the revolution” (November 2008) and in “As smart money goes, so goes the market” (June 2009), we introduced two indicators. The first indicator, a measurement of investor sentiment, relied solely on “weighted” call/put dollar-volume data. The second indicator reflected the tendencies of so-called “big money” by measuring the bias of the most active stocks in an advance/decline series. Both indicators had unique calculations and relied on different data streams to identify key turning points in the stock market. That they produced separate, but coincident, results added to their validity.
Keeping in mind the plausibility of using different indicators on different data streams to produce better results, let’s take a look at some new data on both intermediate (weekly) and major (monthly) cycles.
Created in 1932, the Barron’s Confidence Index (CI) is one of the oldest market sentiment indicators. It has operated on a relatively simple premise for nearly eight decades: a rising CI index is bullish for stocks; a falling CI is bearish. According to Barron’s, the indicator reflects a “high-grade index divided by an intermediate-grade index; decline in latter vs. former generally indicates rising confidence pointing to higher stocks.” More precisely, the index is calculated by dividing the average yield on high-grade bonds by the average yield on intermediate-grade bonds. Investor confidence is supposed to vary relative to the discrepancy in the yields.
Unfortunately, as a pure buy and sell indicator — buying stocks when the index is rising and selling stocks when it is falling — the CI has not worked well. In his seminal work, “The Encyclopedia of Technical Market Indicators,” Robert W. Colby sums up his research on the Confidence Index: “While the Confidence Index produced profits for a long-only strategy in the distant past, it no longer offers worthwhile possibilities for timing the market.”
While we defer to Colby’s broad conclusion about the timing capabilities of the CI, the last 30 years of data show that the CI has some uses beyond just a straight buy and sell indicator. It is possible, especially on the major cycle, to observe inflection points, with levels of support and resistance, trendlines and especially divergent statistical action, that can be used as an adjunct to basic price analysis much as an analyst might use a classic advance/decline series (see “Major applications,” right).
The charts accompanying this article depict weekly dates for the S&P 500 index from November 1980 to date, as well as one monthly chart covering the entire period of data collection. The CI is plotted directly below the S&P data. A slow stochastic (input: 10) is applied to both the stock index and the CI. In the major cycle chart, the points where the slow stochastic coincides with the stock index and the CI tend to mark major turning points. For example, notice the lows in mid-1982, late 1987, late 1994, early 2003 and recently in March 2009. Major tops in 2000 and 2007 were coincident with overbought readings and divergent statistical readings in both data series. That underscores the suggestion that an indicator applied to different data streams can produce confirmatory signals.
1981 THROUGH 1990
There have been several distinct periods reflected in the CI. The first period encompassed the early years of the primary bull market from the early 1980s to late 1990 (see “Upward bound”). The CI relative to the S&P 500 had a generally upward bias to suggest that investors were relatively willing to take on risk and were generally optimistic. In addition, lesser-cycle intermediate-term peaks and valleys can also be identified using the slow stochastic indicator for both the S&P 500 and the CI.
The second general period in CI followed the October 1990 price lows in the stock market (see “Leading the recovery,” left). The CI not only led prices into the late 1990 correction, but the index remained in an erosive phase until late 1994 and as the market continued to work higher during that five-year period. And while the slow stochastic indicator on the monthly S&P chart remained in relatively overbought territory, the slow stochastic indicator applied to the CI fluctuated between modestly overbought and oversold readings. That tendency lasted until late 1994 when deeply oversold readings in both the S&P and the CI were followed by an upward spike in CI as market prices rallied to new all-time highs.
While investor confidence as reflected in the CI continued to do well into mid-1996, there was marked deterioration in CI as the market rallied into the 2000 price highs. Thereafter, there was steady deterioration coincident with the ensuing bear market that lasted into the late 2002, early 2003 lows.
In its third phase, the CI uptrend that lasted for nearly five years and until October 2007 peaked in early 2004 (see “Forecasting the fall,” left). But that peak, which was about halfway through the eventual stock price rally, was never surpassed. In fact, a massive divergence in the CI that preceded the 2007 highs suggested that investors were increasingly skeptical of the uptrend, just as was the case at the top of the tech boom back at the 2000 market highs. The eventual bear market low in CI (45.2) the week ending Dec. 19, 2008, was 54% below the statistical high of 98.7 reached the week ending Aug. 16, 1996.
CURRENT MARKET REVIEW
While the CI reached its nadir into the November 2008 market price lows with coincident oversold readings on both the intermediate and major cycles, in subsequent strength CI has failed to significantly overcome major resistance created by the 2002, early 2003 statistical lows in CI. That failure is in spite of S&P 500 index prices that are currently well above similar price levels (1091 vs. 768).
Clearly, the CI is reflecting less optimism about future market prospects than market prices, as some participants would have us believe. In addition, recent slow stochastics readings in both the S&P and the CI have moved back into overbought territory to suggest the possibility of market vulnerability as both data series have lost upside momentum in the vicinity of defined 200-day moving averages. A shorter-term concern reflects the fact that the CI peaked the week ending Aug. 7, 2009, at 71.1 and has yet to revisit that level despite strength to new intermediate-term highs by the market.
With finite amounts of data and an infinite number of potential indicators available, market analysts may be tempted to think that a more-indicators-is-better approach to analysis might be the best market strategy. But if two indicators can provide information, then will 20 indicators, each with its own performance variables, do an even better job?
Such an approach can create analytical nightmares. When an increasing number of indicators must be coincident before action can be taken, the end result can be slower and more cumbersome.
A better approach might be to use a variety of time-tested indicators on different data. The results can provide a better result than throwing multiple indicators at the same data for results that may yield little more than repetitive signals that can prove to be not only redundant, but even conflicting and confusing.
Experience shows that the Barron’s CI, when combined with a reputable indicator such as slow stochastics, can give the analyst market insights that can influence buy and sell decisions. In other words, if the Barron’s CI has begun to diverge from market price action, there is a possibility that the current trend may be nearing a turning point.
Toward the end of the fourth quarter of 2009, not only was the CI facing defined statistical resistance, but the slow stochastic indicator that monitors movement in the CI moved back into overbought territory and a zone of vulnerability. Coincident, upside, statistical extremes generated by the slow stochastic indicator using S&P 500 index monthly closing prices suggest investors could become less optimistic about the stock market into the first quarter of 2010, or at least until the stochastics on both data streams appear less vulnerable.
Robert McCurtain is a technical analyst, market timer, and private investor based in New York City. He can be reached at email@example.com.