Week 6: Investor Sentiment

1. What is investor sentiment?

This lecture asks how the collective psychology of investors affects markets. Three questions guide it: what does "sentiment" mean and how is it different from expectations, how can we measure it, and why do waves of optimism and pessimism create price anomalies?

The key word is non-fundamental. If a company reports higher earnings and its price rises, that is a reaction to fundamentals. If investors buy technology shares "because the future is digital" while nothing in the companies' profits or prospects has changed, that is sentiment. One investor holding such a belief would not move the market, but when millions share the same story, their trades push prices in the same direction.

Why does this matter in a course on behavioral finance? In the efficient market view, mistakes of individual investors cancel out, and arbitrageurs correct any remaining mispricing. Sentiment is the case where errors are correlated: many people are too optimistic at the same time. Correlated errors do not cancel out, and because arbitrage is risky and limited, they can move prices away from fundamental value for a long time.

2. From Shiller to Baker and Wurgler

Robert Shiller (1981, 2000) showed that stock markets display excess volatility: prices move much more than the later changes in dividends (the fundamentals) can justify. If prices were simply the present value of expected dividends, they should be smoother than dividends themselves, but in the data they are far more volatile. Something other than fundamentals, such as changing moods, must be moving prices. His book Irrational Exuberance (2000) appeared at the peak of the dot-com bubble.

Baker and Wurgler (2006) took the next step: they built an Investor Sentiment Index that combines several market measures into one number. The purpose is to quantify the psychological component of prices and test how it relates to future returns. The core idea: when sentiment is high, prices are overvalued, so future returns are low; when sentiment is low, prices are too cheap, so future returns are higher.

The logic of sentiment: the optimism, excess, correction cycle

The lecture summarises the mechanism as a loop:

1. Positive expectations (optimism): a story spreads that an asset or sector has a great future.
2. Mass buying (herding): investors imitate each other and buy.
3. Overpricing or bubble: prices rise above fundamental value.
4. The correction then feeds back to the start: when prices fall, expectations turn negative, and the cycle starts again in the other direction.

This cycle is the psychological basis of many market episodes, and it links this week to the next lecture on herding: herding is the behavioral channel through which sentiment becomes price pressure.

3. How is sentiment measured?

Sentiment cannot be observed directly, so researchers use proxies. There are two main approaches:

1. Survey-based measures: ask investors directly how they feel about the market. Examples are the AAII survey (American Association of Individual Investors) and Investors Intelligence (a survey of newsletter writers).
2. Market-based measures: use market variables that reflect mood, such as the number of new IPOs and their first-day returns, turnover (trading volume divided by market capitalisation), the dividend premium and the VIX volatility index, often called the "fear gauge".

Surveys measure what people say; market measures capture what people do with their money. Surveys are direct but can be noisy and depend on who answers. Market measures are available at high frequency but can also reflect things other than sentiment (for example, IPO volume also depends on interest rates).

The Baker and Wurgler index (2006)

Baker and Wurgler combine six market variables, each of which is thought to move with investor mood:

Some intuition for each component:

• Closed-end fund discount. A closed-end fund holds a portfolio of shares and itself trades on the exchange. Its NAV is the value of the shares it owns. If the fund holds shares worth 100 per unit but trades at 90, the discount is (100 − 90) / 100 = 10%. Since the underlying portfolio is known, a changing discount reflects the mood of the (mostly retail) investors who trade the fund. When they are optimistic, the discount shrinks.
• Turnover. If 2 billion dollars of shares trade in a month on a market worth 20 billion, turnover is 2 / 20 = 10% per month. High turnover suggests investors are confident and trade a lot on their own opinions.
• IPO volume and first-day returns. Firms go public when investors are willing to pay high prices. A first-day jump from an offer price of 20 to a closing price of 30 is a 50% first-day return: investors want the shares so badly that they pay far more than the issue price on day one.
• Equity share. Managers know their own firm. When they choose to raise money with shares rather than debt, they think shares are expensive. If firms raise 60 billion in equity and 40 billion in debt, the equity share is 60%. A high equity share is a sign that the market is overvalued.
• Dividend premium. Dividend payers are "safe", mature companies. When investors are euphoric they prefer exciting growth firms without dividends, so the valuation premium of dividend payers becomes small.

Each variable contains some sentiment and some unrelated noise. Baker and Wurgler extract the part they have in common using principal component analysis (PCA): the first principal component of the six variables is the sentiment index. PCA finds the weighted combination of the variables that explains as much of their joint movement as possible. If all six tend to move together, that common movement is interpreted as sentiment.

4. Sentiment and future returns

The main Baker and Wurgler finding has two parts:

• When sentiment is low, pessimism drives prices below value, and future returns are higher as a result.
• During high-sentiment periods, "speculative" assets (small, growth and loss-making firms) rise in the short run, yet over the long run they underperform.

So sentiment is inversely predictive of future returns. This is the signature of mispricing that corrects later: high sentiment today means prices are too high, and the correction shows up as low returns afterwards.

Which stocks are affected most?

Sentiment mainly affects stocks that are:

• small-cap,
• low-liquidity,
• unprofitable or "story stocks" (firms valued for a narrative rather than current earnings).

The reason is twofold. First, these stocks are hard to value: with no earnings and an uncertain future, there is a lot of room for opinions and emotions. A utility with stable dividends leaves less room for fantasy. Second, they are hard to arbitrage: they are expensive and risky to short, and there are fewer professionals trading them. Institutional investors absorb part of the sentiment effect, but not all of it. The behavioral pattern is: high sentiment creates speculative upward pressure now and lower returns later.

5. Textual and online sentiment

New technologies make it possible to measure sentiment from text and online behavior:

• financial articles and newspapers (Financial Times, Wall Street Journal),
• social media posts (Twitter/X, Reddit, StockTwits),
• Google Trends and searches for investment-related terms.

Text is turned into a number with natural language processing (NLP) and word lists called lexicons. Two well-known lexicons are the Harvard IV dictionary (a general psychology word list) and the Loughran and McDonald dictionary, built specifically for finance. The finance dictionary matters because many "negative" words in everyday language are neutral in finance: "liability", "tax" or "cost" are not bad news in an annual report.

The idea behind all these measures is that mass information becomes a psychological signal for the market.

Google Trends: Da, Engelberg and Gao (2015)

Da, Engelberg and Gao (2015) rely on the Google Search Volume Index (SVI) to gauge household fear. A rise in searches for words such as "recession" or "crash" is followed by lower stock prices in the next week. The SVI thus acts as a live gauge of fear or optimism. In the slide's stylised chart the scaled SVI goes from roughly 20 in week 1 up to a peak near 70 in week 4 and back down to roughly 25 by week 7: higher SVI points to growing anxiety and weak returns ahead.

Why does this work? People search before they act. Somebody who searches "recession" or "how to sell stocks" is worried, and worried households may soon sell. Search data reveal this mood before it shows up in trades or official surveys, and they cover millions of people rather than a few hundred survey respondents.

Google Trends as a data source

• Google Trends gives access to anonymous, aggregated search data.
• Searches usually happen before an economic or investment action ("recession", "buy Bitcoin"), so they can be a leading sentiment indicator.
• The researcher chooses keywords and downloads their search intensity over time (scaled from 0 to 100, where 100 is the peak in the chosen period).
• Choi and Varian (2012) showed that Google Trends can nowcast ("predict the present") many economic measures, such as unemployment claims or car sales, and improves short-term forecasts, because official statistics are published with a delay while search data are available immediately.

The "GT in action" slide compares searches for "inflation" and "baseball" in Greece over the last five years. Searches for "inflation" rise sharply during 2022, the period of high inflation, and then decline, while "baseball" stays flat and low. This is how a student can use Trends: pick a term that captures a worry, check that it moves with real events, and use it as a sentiment proxy.

Media sentiment: Tetlock (2007)

News headlines influence public mood. Tetlock (2007) measured the share of negative words in a popular daily Wall Street Journal column ("Abreast of the Market") and found that a negative tone in financial news predicts a market decline the next day (followed by a partial reversal later, consistent with a sentiment effect rather than new information). Journalists also reinforce dominant narratives (narrative reinforcement): once "the economy is heading for recession" is the story, new facts are framed to fit it. Investors with limited attention react to headlines rather than to full analysis, so they overreact. The lesson is that information is never neutral: it spreads through an emotional filter.

6. Behavioral interpretations of sentiment

Sentiment is the "macroscopic footprint" of individual biases. Four biases explain how individual psychology becomes market-wide mood:

• Affect heuristic: emotions dominate probability assessments. If a sector "feels" good, people judge its risk to be low and its potential high at the same time.
• Herding: investors imitate the mood and actions of the crowd.
• Confirmation bias: investors seek information that supports what they already feel and ignore contrary evidence.
• Availability bias: recent and vivid events shape current mood. After a crash, another crash seems very likely; after two years of rising prices, further rises seem normal.

7. Sentiment, volatility, liquidity and bubbles

Volatility

High sentiment goes together with low perceived volatility, the "calm before the storm". Investors underestimate risk in euphoric phases. When the mood reverses, volatility spikes sharply. This asymmetry is typical: markets tend to rise slowly and calmly and fall fast and violently.

Liquidity

When sentiment is high, liquidity increases because everyone wants to buy, so it is easy to trade. When sentiment drops, liquidity "freezes": sellers do not want to sell at low prices and buyers disappear. Brunnermeier and Pedersen (2009) describe liquidity spirals: falling prices cause losses for leveraged traders, who face higher margin requirements, are forced to sell, which pushes prices down further and makes funding even harder. The behavioral point of the slide: emotion drives capital flows, not the other way around.

Bubbles

Sentiment acts as the catalyst of bubbles, as in dot-com (2000), housing (2007) and crypto (2021). Shiller (2019) calls this narrative contagion in Narrative Economics: stories pass between people the way a virus does, and economic behavior shifts once a story goes viral. Once sentiment runs too hot, nobody pays attention to fundamentals anymore.

8. Macroeconomic and policy dimension

Sentiment does not only affect stock prices. It affects consumption, investment and monetary policy. Akerlof and Shiller (2009) revived Keynes's term "animal spirits" to describe how confidence, fear and stories drive economic cycles. Confidence indices (consumer confidence and business confidence) serve as sentiment proxies at the level of the whole economy. Behavioral economics in this way bridges micro psychology and macro dynamics.

Authorities monitor sentiment indicators to prevent complacency (too little fear, risk building up) or panic. Central bank communication, such as forward guidance (announcing the likely path of future interest rates), is a form of expectation management. "Behavioral macro-finance" studies how narratives influence bond markets and monetary stability. Example: the ECB or the Fed uses statements to calm investor sentiment, as in Mario Draghi's 2012 promise to do "whatever it takes" to preserve the euro, which calmed bond markets without any money being spent at that moment.

Recap: sentiment can be measured, it keeps returning and it has behavioral roots; it influences returns, volatility, liquidity and investor confidence; and it works as a "psychological variable" in capital markets.

9. The VIX: the "fear index"

An option is like insurance. When investors are scared, they pay more for put options (insurance against a fall), which raises option prices. Higher option prices imply higher expected volatility, so the VIX rises. A rising VIX signals fear and risk aversion and usually coincides with falling stock prices. A low VIX indicates complacency and a risk of overvaluation. The VIX is the market's "psychological thermometer".

The lecture's VIX charts (1990 to 2025) show that the index usually stays within bounds, mostly somewhere from about 10 to 30. Every so often it breaks out: extreme spikes mark moments of panic. On the 2007 to 2009 chart the VIX jumps to roughly 80 at the end of 2008, after Lehman went under. Plotting the VIX next to the S&P 500 over those years reveals the inverse relationship: more fear, lower prices. The S&P 500 dropped from over 1,500 to under 700 as the VIX soared, and prices came back once the VIX eased in 2009.

Behavioral interpretation of the VIX

• What the VIX captures is collective anxiety, which is not necessarily actual risk.
• By buying puts, investors pay a "fear premium": on average, implied volatility exceeds the volatility that is later realised.
• Very low VIX levels often come before a crash (complacency effect).

Shiller: "When markets show complete confidence, they are most vulnerable."

10. The AAII sentiment survey

The American Association of Individual Investors (AAII) runs a weekly survey asking members whether they are bullish (expect prices to rise over the next six months), bearish or neutral. The difference between the bullish and bearish percentages is the net sentiment (often called the bull-bear spread).

Historically, extreme optimism precedes low future returns and extreme pessimism precedes high future returns. Mass optimism is therefore often a contrarian indicator. Retail investors tend to react emotionally to news, and institutional investors use the AAII survey as a contrarian signal. Rules of thumb from the slides:

• Bullish > 50%: "crowded optimism", a correction is likely.
• Bullish < 20%: "fear zone", a rebound is possible.

Behavioral pattern: extreme sentiment ⇒ reversal of expectations.

The "Bull vs Bear" chart (1990 to 2025) shows that bullish and bearish readings swing widely, typically between about 20% and 60%. Two charts for May 2024 to May 2025 then illustrate the contrarian logic. In "Panic: everything is falling apart", bearish sentiment climbed to about 60% in spring 2025 (the tariff turmoil of April 2025), yet the S&P 500 measured six months forward was at its highest level, above 6,500. In the "#not" chart, bullish sentiment fell to around 20%, again exactly when forward returns turned out strong. The crowd was most scared just before the market rose.

11. The MOVE index

The ICE BofA MOVE Index tracks volatility in the U.S. bond market. It is constructed from a set of options on Treasuries and follows a weighted yield curve, using options with one month to expiry on Treasuries with 2-, 5-, 10- and 30-year maturities. Put simply, it is "the VIX of interest rates".

Remember that interest rates and bond prices move inversely. A bond paying a fixed coupon becomes less attractive when new bonds pay more, so its price falls when yields rise. When yields rise quickly, bond prices fall and the MOVE index rises.

The "MOVE vs Yields" chart (2021 to 2025) shows the MOVE rising from below 60 in 2021 to peaks near 200 in 2022 and early 2023, while the 10-year Treasury yield climbed from about 1.5% to around 4% to 5% as the Fed raised rates. By November 2025 the MOVE had fallen back to about 74, with the 10-year yield near 4.1%.

12. The Fear and Greed Index

CNN's Fear & Greed Index tries to gauge whether stock prices are fair by blending seven components:

1. market momentum (S&P 500 versus its moving average),
2. stock price strength (stocks at 52-week highs versus lows),
3. market breadth (volume in rising versus falling stocks),
4. put/call options (the ratio of put to call trading),
5. demand for junk bonds (the yield spread between junk and investment-grade bonds),
6. market volatility (VIX),
7. safe-haven demand (stocks versus Treasury bond returns).

Each component is measured by how far it deviates from its historical average, and the combined score ranges from 0 (extreme fear) to 100 (extreme greed). The "Danger zone" slide shows a reading of 21, "Extreme fear" (7 November 2025), compared with 23 the previous close, 46 (neutral) one week earlier, 50 (neutral) one month earlier and 59 (greed) one year earlier. For a contrarian, extreme fear is a potential buying signal and extreme greed a selling signal.

13. Sentiment indicators in practice

No single indicator is perfect. Together they capture the market's emotional pulse. For a research project this is a useful point: results are more convincing if they hold for more than one sentiment proxy.

14. Sentiment and the IPO market

IPO activity depends heavily on sentiment. In optimistic times the number and valuation of IPOs increase; new listings post a first-day jump and frequently underperform over the next few years. Behaviorally this comes from overconfidence and FOMO (fear of missing out), which push prices too high. Companies and their banks pick these windows of enthusiasm to go public. Because the IPO market gauges collective enthusiasm, two of the six Baker and Wurgler variables are taken from it.

15. Crypto, meme stocks and the attention economy

Crypto and meme stocks serve as "laboratories" for studying sentiment: prices have few fundamentals to anchor them, retail traders dominate and social media move fast. Reddit (WallStreetBets), Twitter and TikTok spread emotional contagion. The price of Bitcoin frequently moves with search and tweet volumes. Herding, sensation seeking and narrative contagion are the behavioral traits at work.

Today attention also moves markets. Stocks that trend on Google or Twitter often jump for a while. Attention-driven buying is sentiment trading in a modern form: retail investors cannot study thousands of stocks, so they pick the ones that catch their eye. Since a purchase means choosing from thousands of stocks but a sale only concerns the handful you hold, attention mostly generates buying pressure. Once the attention fades, an "attention crash" sets in. Behavioral volatility has gone digital.

16. Using sentiment in investment strategies

Contrarian strategies

Contrarian investors exploit extreme sentiment levels:

• High sentiment, so overpricing: go short or underweight.
• Low sentiment, so fear: go long or overweight.

These rules are often combined with the VIX, AAII or Baker and Wurgler indicators. The slide's chart shows the position as a straight downward line in standardised sentiment: fully long at sentiment −2, neutral at 0 and fully short at +2. Contrarian strategies work when the crowd overreacts.

Sentiment-based trading with text

In practice, market indicators are combined with sentiment extracted by NLP from news or social media. Common entry signals:

• positive sentiment plus a bullish trend: go long,
• negative sentiment plus bearish momentum: go short.

Note that this is the opposite of the contrarian rule: it follows sentiment in the short run, consistent with the Baker and Wurgler finding that speculative assets do well in the short term when sentiment is high. Machine learning models such as BERT (a language model that reads words in context) and LSTM networks (which handle sequences such as time series or sentences) are used to classify text sentiment. The slide's claim is that AI can measure the "market's emotion" better than humans, because it reads millions of texts consistently.

Correlation of sentiment with other factors

The value factor (cheap stocks minus expensive stocks) does badly when sentiment is high, because investors chase exciting growth stocks. Momentum is helped by sentiment because optimistic investors keep buying past winners. Sentiment permeates all return-driving factors.

17. Sentiment, society and consumer confidence

Investor sentiment ties in with the wider social mood, such as consumer confidence and political uncertainty. Optimistic periods coincide with more risk-taking in all markets, and sentiment travels via the media, political debate and shared narratives. Markets reflect how society as a whole feels.

The University of Michigan Consumer Sentiment Index (MCSI)

The survey includes questions on:

1. the household's financial situation now versus one year ago,
2. expectations for the next 12 months and the next 5 years,
3. intentions to buy durable goods (cars, appliances).

The MCSI captures the "economic mood" of consumers, who view the same macro data through an emotional filter. A rising index means growing confidence, with more consumption and investment. A falling index means fear takes over: people save more and spend less. The MCSI tends to lead GDP and the stock market, and drops in the index frequently come before recessions and bear markets. Behaviorally, consumer sentiment is a kind of animal spirits (Akerlof and Shiller, 2009), linking the emotions of individual consumers to sentiment at the market level.

Michigan versus the S&P 500

The first chart (2018 to 2025) shows that the MCSI and the S&P 500 do not move in lockstep. The MCSI dropped sharply in 2020 (COVID) and again in 2022 (inflation), and by November 2025 it stood at about 50.3, close to historic lows, while the S&P 500 was near a record of about 6,700. Consumers felt gloomy, but stocks kept rising.

The second chart (source: TheMarketStats) asks what the S&P 500 did after the MCSI fell below 56. The table of past signals:

Interpretation: after consumer sentiment dropped below 56, the S&P 500 was higher one year later in 100% of the cases, with an average one-year return of 18.41%. In the short run the picture is mixed (only 63% positive after three months, with 2008 showing −17.98%). This is the contrarian logic again: very low consumer mood has historically been a buying opportunity for patient investors. Keep in mind that the sample is small (six complete one-year observations), so this is suggestive evidence, not proof.

18. Conclusion

• Investor sentiment is more than a mood: it is a collective narrative.
• It affects returns, bubbles, liquidity and even economic policy.
• Understanding it requires a human-centred approach to finance.