“Rip Van Winkle would be the ideal stock market investor,” Richard Thaler, the Nobel prize-winning behavioral economist, points out. He adds: “Rip could invest in the market before his nap and when he woke up 20 years later, he’d be happy. He would have been asleep through all the ups and downs in between. But few investors resemble Mr. Van Winkle. The more often an investor counts his money – or looks at the value of his mutual funds in the newspaper – the lower his risk tolerance.”
Indeed, it is difficult to use sentiment in investing precisely because as humans, we become caught up in the waves. Too often we are unconsciously influenced by the stories and beliefs of the herd, adopting them as our own.
We can see a rough representation of the sentiment cycle using media sentiment averages. These seem to precede S&P 500
price movements since 1998. Below we can see that when the shorter-term moving average of sentiment (200 days) is greater than the longer term moving average (500 days), the market tends to rise. Those periods of more positive recent media are colored with green shading below. When the 200-day average is more negative, the market tends to follow to the downside (pink shading).
The chart shows that a long-term moving average of media sentiment has historically been broadly useful in predicting the ups and downs of public equity markets. In particular, it seems to do well in timing tops. On average, a long-short investor would outperform a buy-and-hold investor by 3x using the moving averages of sentiment to time long- and short bets. The media sentiment tone is perhaps reflecting or perhaps driving (or perhaps both) overall investment risk taking.
There are several potential explanations for this cycle. Confirmation bias may prevent investors from selling despite the warning signs at the top of a bull market. But that is not the whole story. Another bias — Thaler and Eric Johnson’s “house money effect” — also provides insight into the cognitive biases that may contribute to the cycle.
The house money effect can occur when people experience a sudden windfall — earning or receiving more wealth than they had expected. Thaler and Johnson found a subsequent increase in risky behavior. The researchers called this bias the house money effect because it is similar to gambler behavior at casinos. Having won money, most gamblers will take more risk. It is said that they feel they are playing with “the house’s money.”
Increasing risky behavior after a gain seems to contradict the “cutting winners short” bias, which is derived from Daniel Kahneman and Amos Tversky’s Prospect Theory. Investors cannot both take more risk and cut risk in the “domain of gains,” so one of these must be wrong.
The reason for this seeming contradiction is the mobility of the reference point in prospect theory. When people receive a windfall, especially if through their investments, they quickly learn to do more of the rewarded behavior. They feel compelled to take more risk in order to catch up with where they “should be” (the reference point). The reference point is moved higher as the media reports on “bitcoin
millionaires” and startup moguls who make it look easy. As long as investors are experiencing fear of missing out (FOMO), then they are in the realm of losses and will continue to take excessive risk via the house money effect.
However, after a long bull market, sometimes the market corrects, and investors lose a sizable portion of their highly-leveraged net worth. At these moment, their reference point shifts lower, and they begin to imagine they may lose their gains. They suddenly shift into the domain of gains (”at least I still have some profits”) and are more susceptible to cutting winners short — taking their profits. As they do so, prices are driven lower. A self-reinforcing cycle of losses may be triggered as self-preservation becomes the goal.
Note that the sentiment MACDs depicted here are better at timing market tops. As the media tone shifts to the negative, the gradual resetting of the reference point lower may be the cause of this superior performance.
Fundamental human behavior is consistent across cultures. Consistent with that observation, this pattern is in evidence globally.
The cycle that we see currently in the S&P 500 is also present — and profitable — in developed markets globally. Images of the remarkable predictive nature of this pattern are visible for the largest developed-markets benchmarks below through mid-November: the Dax
, the Nikkei
, the Hang Seng
, and the FTSE Europe All Cap.
FTSE Europe All Cap
Most developing markets respond to media sentiment more quickly than developed markets, and the 200- vs. 500-day MACD appears too slow in the Bovespa and the China Composite. Perhaps these markets are driven more by short-term speculators than long-term investors or slow-moving institutions, and so the volatility is more violent. Consistent with that observation, the 30-90 MACD of sentiment seems to work best in China and Brazil.
By the nature of their construction, MACDs force around 50% long- and short positions over any given time period. Note that our look back period is ideal for this type of study, with many stock markets making only modest gains since 1998. If we have a prolonged bull market, as we did from 1982-1999, then a buy-and-hold strategy would win out over a sentiment MACD.
In this currently bullish global market environment, investors should always keep in mind that when times are good, our reference point (expectations) rise, and we become susceptible to the house money effect. It is difficult to conceive of worsening conditions at such times. In truth, we ought to be most on alert precisely when everyone else is complacent.
Behavioral economist and psychiatrist Richard L. Peterson is CEO of MarketPsych, a training and sentiment analytics firm.