Close

    Investor Sentiment Indicators: Quite Contrary?

    by Mark Hulbert

    Every ideology runs the risk of becoming the refuge for scoundrels, and contrarian analysis is no exception. It is the rare adviser today who doesn’t claim to be a contrarian, regardless of whether he is bullish or bearish.

    It would appear that, at least in many advisers’ hands, contrarian analysis has become little more than an excuse for sloppy thinking. And it would be easy therefore to conclude that we should ignore the approach altogether. But just as we don’t throw out an ideology because it has been used disingenuously, we shouldn’t be too quick to dismiss contrarian analysis.

    For this article I studied the relationship between the stock market and several different measures of investor sentiment. The results show that there is something genuinely worthwhile in contrarian analysis.

    WHERE IT STARTED

    As far as I can tell, contrarian analysis as an investment tool traces back to a Vermonter named Humphrey Neill, who in the 1940s began publishing a newsletter called the Neill Letters of Contrary Opinion.

    Neill was suspicious of widely held beliefs not because he felt the majority is always wrong, but because it is less likely that the belief will be subject to critical scrutiny when everyone around subscribes to it. He recommended that we constantly be on our guard against groupthink.

    Nevertheless, Neill also said “when everybody thinks alike, everyone is likely to be wrong.” This appears to call for assessing the majority’s opinion and, when it is at an extreme, doing the opposite. And this is exactly what generations of his followers have tried to do.

    In order for this type of analysis to have a fighting chance, however, contrarians need an objective measure of the majority’s market views. Absent that anchor, it’s all too easy for contrarian-oriented advisers to “detect” a majority opinion that is conveniently contrary to their predetermined position.

    MEASURING SENTIMENT

    One of the earliest efforts to produce an objective assessment of the majority’s view was initiated in 1963, when Abraham Cohen of Investors Intelligence began measuring bullish and bearish sentiment among investment advisory newsletters. Cohen each week categorized a large group of newsletters into three camps: those that were outright bullish, those that were outright bearish, and those that were long-term bullish but expecting a near-term correction. That measurement continues to this day, under the current oversight of Michael Burke. Each weekly reading is widely quoted in the financial press, including Barron’s.

    Another widely quoted measure is AAII’s own survey of stock market sentiment among its members, which dates back to mid-1987. Members who participate are asked to indicate whether they are bullish on the stock market, neutral, or bearish. Weekly summary statistics are published on AAII’s Web site and in Barron’s.

    Many contrarians also pay attention to the Chicago Board Option Exchange’s Volatility Index, or VIX for short. The VIX is based on the implied market volatility among a basket of widely traded options on the S&P 500 index. Low readings are taken to mean that option traders have become complacent, while high readings are evidence of widespread fear. Yet another objective measurement of the majority’s views comes from the Hulbert Financial Digest. The Hulbert Stock Newsletter Sentiment Index (HSNSI) measures the average stock market exposure among a subset of several dozen newsletters that engage in short-term market timing. This subset was constructed so that the HSNSI would be extremely sensitive to changes in advisory mood.

    SENTIMENT INDICATOR PATTERNS

    Table 1 presents summary statistics on weekly readings of these sentiment measures over the last 60 months (August 1999 through August 2004).

    Several patterns emerge. You can see that sentiment tends to be skewed toward the bullish end of each series’ range. For example, the HSNSI’s range over the last five years has been from minus 81.8% on the low side to 79.5% on the high end. Yet the median reading, 30.6%, is much closer to the high end. The same is true for the other indicators as well.

    Table 1. Sentiment Indicator Summary Statistics on Weekly Readings: 8/99 to 8/04
      Investors Intelligence AAII Sentiment Index VIX
    (%)
    HSNSI
    (%)
    Bulls
    (%)
    Bears
    (%)
    Correct
    (%)
    Bull-Bear
    Spread
    (%)
    Bullish
    (%)
    Neutral
    (%)
    Bearish
    (%)
    Bull-Bear
    Spread
    (%)
    Average 49.5 28.7 21.8 20.9 45.2 27.7 27.1 18.1 23.3 26.5
    Median 50.0 29.2 22.4 21.6 44.5 26.8 25 19.2 22.1 30.6
    Standard Dev 6 6 5 7 12 8.1 10.1 20.7 5.7 25.1
    Maximum 62 43 31 44.1 75 51.4 57.9 62.9 43.7 79.5
    Minimum 28 16 8 -14.8 21.1 7.7 6.7 -36.8 14.4 -81.8

    This bullish skew is exactly what contrarians would have expected, since the stock market on balance has declined over the last five years.


    How They Correlate
    I performed two analyses to see whether these “indicators” were truly representative of market sentiment. In the first analysis, I examined the correlation of the readings of the four sentiment indicators. Because the HSNSI and VIX both report a net number, I needed to use a comparable reading for Investors Intelligence and AAII, so I used the percentage spread between bulls and bears.

    What I found was that there is significant correlation between the various measures—precisely what one would expect if these indicators are at least loosely measuring the same phenomenon.

    The second analysis examined the strength of the correlations that exist between the four sentiment indicators and the subsequent returns of the overall stock market, as measured by the Wilshire 5000 Total Return Index. This analysis did not use sophisticated models, but rather was simply designed to test two things: first, to see whether the relationship between the indicators and the market’s return is non-random; and secondly, to see whether the relationship between the indicator and the market is as hypothesized, with lower returns associated with greater bullishness/complacency and higher returns associated with greater fear.

    What I found was that all four indicators passed these two simple tests at a high level of statistical significance:

    • All four sentiment indicators are correlated with the stock market’s subsequent one-year return, and
    • All four sentiment indicators have the hypothesized relationship with the Wilshire 5000 returns.

    Passing these two tests is just the prerequisite, of course. It merely suggests that investors can profitably use these indicators as part of their overall strategy. Actually developing that strategy was beyond my scope.

    CURRENT READINGS

    From the point of view of the bulls, current readings from these indicators have improved significantly this summer. Earlier this year, all four were reporting excessively high levels of bullishness, which contrarians would see as a bearish sign.

    However, as of mid-August, three of the four indicators (all except the VIX) are reporting more bearishness than has existed in at least 80% of the weeks since August 1999.

    This doesn’t guarantee that the market is out of the woods. But my results do suggest that the data bear watching.


    Mark Hulbert is editor of the Hulbert Financial Digest, a newsletter that ranks the performance of investment advisory newsletters. It is published monthly and is located at 5051B Backlick Rd., Annandale, Va. 22003; 703/750-9060; www.hulbertdigest.com. This column appears quarterly and is copyrighted by HFD and AAII.

→ Mark Hulbert