The potential to use investor sentiment as a gauge for future market trajectory is an appealing proposition. Sentiment is often cited as a contrarian indicator; excessive bullishness signals market exuberance poised for a reversal, whilst extreme bearishness may be the precursor to a market recovery. Such indicators are most useful when they are at extremes, and less so when the readings are neutral, which tends to be most of the time.
The reason for assessing sentiment is that, when investors are extremely bullish, they tend to be fully invested, leaving little available cash to drive asset prices higher. On the other hand, when extreme bearishness prevails, the abundance of cash sitting in portfolios can be deployed to buy cheap assets, creating the foundation for a bull market.
There are broadly two approaches to quantifying investor sentiment: attitude and activity. The former are typically surveys, gathered through proxies, of near-term expectations of active investors or market commentators. The second method maps risk appetite through trading activity, measured by asset flows, positioning and market volatility. In practice, the usefulness of these sentiment indicators for predicting stock market returns is somewhat uncertain. In this note we will examine a few of the most quoted metrics.
Starting with surveys, we look at both institutional investment advisor sentiment as well as individual investor sentiment, represented by US Investor Intelligence (II) and the American Association of Individual Investors (AAII), respectively. The US Investor Intelligence series is one of the oldest sentiment indicators, dating back to 1963, collating the opinions of over 100 newsletter writers through weekly surveys. Currently, the series is displaying extreme pessimism. The bull/bear ratio dipped below 95th percentile for the first time since 2008, or 2011, if you include the very close encounter of that year. We use percentiles instead of the usual standard deviation measure due to the skewed nature of the series (positive skew, in this case), defining extreme observations as the outermost 5% on either end of the bullish and bearish spectrum.
On this measure, some simple analysis shows that it does in fact have predictability, albeit rather one-sided and, somewhat surprisingly, more reliable over the longer term of 2-3 years and less useful for near-term returns of one year or under. The caveat here is the limitation of a small sample size, as this level of pessimism has only occurred a handful of times in history.
The table below shows the average forward returns of the S&P 500 index from the point which sentiments become extreme. Given this is a contrarian indicator, we would expect to find negative future returns if one invest at the point of bullish extremes, and vice versa. And indeed, we do observe visible predictability at bearish levels, whereby future returns are generally positive, and becomes more pronounced as the horizons extend. The quality of the return pattern also improves towards the longer end of the time horizon as the number of negative observations decreases drastically for 2-3 years (from 45% at 1M to 5% at 3Y), meaning you will make money most of the time, whilst in the short-term (<6M), it can be an almost 50:50 chance of getting positive or negative future returns.
By contrast, on the extreme bullish levels, there is no emergence of a contrarian indication as the future returns tend to still be positive on average, and the quality of the result remains mixed even as time progresses. One possible explanation for this may be that the bears can be too early to call a market top, as an irrational bull market can extend for several years before it eventually crashes, as it did in the Global Financial Crisis, thereby making the reading less predictable.
The main difference between professional investors and individual investors seem to be a matter of emphasis. For example, the bullish sentiment leading up to the TMT bubble is much more pronounced on the AAII but barely registered on the II. The peak bullishness in 2015 displayed in II, on the other hand, is much more muted on AAII. On this occasion, both surveys appears to be giving a similarly extreme signal where the bears outnumber the bulls by a hefty margin, as illustrated by the AAII Bulls-Bears spread chart below.
In forecasting returns, it is also the case for AAII that extreme negative sentiment provides a better indication of future performance than positive sentiment, on both standard deviation and 5th/95th percentile basis, as we see in the table below. Although compared to Investor Intelligence, the level of mixed result remains higher even at the longer horizon of 3Y, whereas the chance of getting a negative return after 3Y is all but eliminated for II (% of negative return in 3 years: 17% vs. 5%), which means that the chance of being too bearish too early is still relatively high. The bullish side does give consistently low or negative future returns up to 2 years on average, but again, the quality of the result is questionable given the high percentage of positive readings.
Other commonly used investor activity indicators are also trending towards extremes, including the CBOE equity put/call ratio, which records simply the total volume of equity puts traded divided by the calls traded in the options market, also a contrarian metric.
Flow data which measures the pace at which investors are rotating out of equity is gaining momentum, signalling that they are becoming more risk-averse, as shown in the ICI mutual fund flows data for equity funds:
The bearish sentiment reflected in the shift out of equities is also mirrored by a rise in the price of gold, where gold is typically regarded as a safe haven asset and a good gauge for market uncertainty, having the tendency to perform well in times of stress. This trend is supportive of an increasingly uneasy investment community, with its 3 month price change the quickest gain in value just short of two standard deviations above the long-term mean.
Finally the VIX, also known as the ‘fear indicator’, is a measure of the 30-day implied volatility of S&P 500 index options. Whilst not extreme, it has recently surpassed values last reached in 2011.
Summary & Conclusion
Such indicators are most useful when they are at extremes, and less so when the readings are neutral, which tends to be most of the time.
Each sentiment indicator measures the market from a slightly different perspective, therefore should never be used on a standalone basis, and it is advisable to monitor in conjunction with other market indicators (especially fundamental).
This is by no means a comprehensive list of all the possible sentiment metrics. Not surprisingly, those which we have surveyed are consistent in implying a growing bearishness.
It is too early to call a bottom as not all of the indicators have reached extreme levels. The danger with sentiment watching is that theoretically it can always become more optimistic or pessimistic, driven primarily by price momentum.
As portfolio managers, who have a keen interest in odds (and tipping them in our favour) we monitor sentiment indicators to look for extremes. Should such an extreme be reached, our understanding of these readings will influence asset allocation.
Extracts from this article first appeared in The Daily Telegraph on 5th March 2016