A recent post by The Pragmatic Capitalist on the topic of investor sentiment has me curious regarding the correlation between sentiment and subsequent returns of the S&P 500 index. Last weeks’ survey has bullish sentiment at 45%.
This analysis will consist of data from the AAII Investor Sentiment Survey and be divided into four sections. The first section will look at the entire data series, which started in 1987. The second will look at instances of “extreme” bullishness where bullish sentiment was 50% or higher, one standard deviation above the average bullish sentiment of 39%. The third section will look at instance of “extreme” bearish sentiment where bullish sentiment was 28% or lower; one standard deviation below the average. Finally, the last section will look at periods where the eight week moving average of bullish sentiment was above 50%.
The entire data series since 1987 has 1,202 periods. I have elected to look at the returns over the following thirty, sixty and 360 days from the date of the survey along with returns leading up to the survey of thirty and 360 days. Summary statistics can be found in the following table.
Notice the subsequent returns of the S&P 500 are negatively correlated to the sentiment survey, meaning bullish sentiment and S&P 500 returns tend to move in opposite directions; low sentiment, higher returns and vice versa. This is to be expected as the general consensus is that sentiment surveys are contrary indicators. The correlations across all time periods are fairly low with the strongest correlation being between prior 30 day return and the survey. Notice the positive correlation here which means the bullish sentiment tends to increase with returns over the prior month.
Next we look at the 193 periods were investor sentiment was at or greater than 50% bullish.
Notice that, even when looking only at the most bullish periods, the low correlation across all five time frames. The strongest correlation is between subsequent 360 day returns and the survey but despite the negative correlation the average return under this condition is still positive at 1.9%. The highest average return is 10.5% from the prior 360 days which suggests that “extreme” bullish sentiment is influence by the prior year’s returns. See the following histogram for the distribution of subsequent 360 day returns under this condition.
In terms of “extreme” bearishness, where bullish sentiment is below 28%, we have 213 periods to analyze.
Again, notice the low correlations. The average subsequent return when investors are most bearish is positive (13.6% average, subsequent 360 day return) as well as negative or flat prior returns seemingly pushing investors toward this bearish sentiment. See the following histogram for the distribution of subsequent 360 day returns under this condition.
Finally looking at an eight week rolling average there are 100 periods where the bullish sentiment is above 50%.
Again, the correlation across all time periods is fairly low but again we see subsequent returns when investors are most bullish are, on average, negative. The highest correlation and average return comes from the prior 360 day return which again suggests investor sentiment is strongly driven by what has happened in the past and not the outlook for the future.
In summary, while there is a fairly low correlation between the AAII sentiment survey and either subsequent or past return on the S&P 500, a contrarian approach does appear to be advantageous during period of either “extreme” bullish or bearish sentiment. Average 360 day returns during periods of extreme bearish (bullish) sentiment are 13.6% (1.9%) versus the complete period average of 7.8%. This indicator is best used in conjunction with other data and not as a trading tool in and of itself but in periods of “extreme” bearish sentiment investment conditions appear more favorable.