Fear and Greed in History: Follow the Crowd or Go Against It?

Level: advanced, written so that a beginner can follow it. "Be fearful when others are greedy" is one of the best known rules in investing. But does it hold up against the data? This article looks at more than eight years of daily sentiment readings in Bitcoin and three and a half years of the EllTec Analysis Fear & Greed Index for stocks, compares procyclical behaviour (following the crowd) with contrarian behaviour (going against it), and ends with why and where EllTec Analysis uses sentiment in its analyses. All figures are calculated by EllTec Analysis from public data. Nothing here is investment advice. The six most important points Sentiment is a thermometer, not a timer. It tells you how hot or cold a market is, not when it turns. Extreme fear has often been the better moment to buy, but only with patience: in Bitcoin the typical result three months after an extreme fear reading was still a loss (median −4 %), twelve months later a clear gain (median +36 %). Extreme greed alone has been a poor reason to sell. In Bitcoin the strongest three-month returns came after extreme greed readings (average +40 %). Euphoria can last for months. Mechanical contrarian rules failed in Bitcoin: buying at extreme fear and selling at extreme greed turned 100 USD into 104 USD from 2018 to 2026, while buy and hold reached 874 USD. In broad equity markets the contrarian view worked better in 2023 to 2026: after fear readings the S&P 500 ETF rose 9.2 % over the next three months on average, after greed readings 3.4 %. The best use of sentiment is in combination: with the trend, with the wave structure and with a clear risk plan. That is exactly how EllTec Analysis uses it. 1. What sentiment is and why it moves prices Prices are set by people, and people are not machines. When prices rise, buyers feel confirmed, newcomers fear missing out and lenders become more generous. When prices fall, the same mechanism runs backwards: losses hurt about twice as much as equal gains please (loss aversion), investors sell to stop the pain, margin calls force further selling. This collective mood is called market sentiment. Sentiment matters for two reasons. First, it feeds on itself for a while: rising prices produce optimism, optimism produces buying, buying produces rising prices. That is why trends exist. Second, it eventually exhausts itself: when almost everyone who wants to buy has bought, there is nobody left to push prices higher, and the smallest disappointment can start a reversal. The same is true for selling at a panic low. Both effects are real, and the whole question of this article is which of the two dominates, and when. Warren Buffett put the contrarian side into one sentence in his 1986 letter to Berkshire Hathaway shareholders: the aim is "to be fearful when others are greedy and to be greedy only when others are fearful". Trend followers answer with the old saying "the trend is your friend". Both camps can point to great successes. The data below shows that both are partly right. 2. How a Fear & Greed Index measures the mood Mood cannot be measured directly, so Fear & Greed indices combine several market indicators that reflect it. Each indicator is converted into a score from 0 to 100 and the scores are averaged. The usual scale is: ReadingZoneWhat it usually means 0 to 25 Extreme fear Panic selling, high volatility, investors flee to safe havens 25 to 45 Fear Caution dominates, rallies are sold 45 to 55 Neutral No clear emotional pressure in either direction 55 to 75 Greed Optimism, dips are bought 75 to 100 Extreme greed Euphoria, crowded positioning, little fear of losses 2.1 The Crypto Fear & Greed Index The best known index for crypto is published daily by alternative.me since February 2018. It combines the volatility of Bitcoin, momentum and trading volume, activity on social media, the dominance of Bitcoin in the crypto market and search trends. Its long daily history makes it the best data set for testing sentiment strategies, so it is the basis of most studies in this article. 2.2 The EllTec Analysis Fear & Greed Index For stock markets EllTec Analysis calculates its own index for eight regions (global, USA, Germany, Europe, Asia, the Americas, Africa and the Middle East, Oceania). For every region up to nine indicators are measured every trading day: momentum (price against its 125-day average), strength (position within the 52-week range), volatility (inverted), demand for safe havens (stocks against US Treasuries), demand for junk bonds, market breadth (equal-weight against cap-weight S&P 500), the local currency, gold against stocks and the St. Louis Fed Financial Stress Index. Each indicator is ranked against its own last three years, so every reading is judged against its normal range. The full method and the current values are on the macro dashboard under Market Sentiment. One important property: a Fear & Greed Index is built from price behaviour. A sharp fall almost automatically produces fear, a strong rally greed. The index is therefore partly a summary of recent price action, which explains a lot of the results below. 3. Two ways to react: procyclical and contrarian Procyclical (also called cyclical or trend-following) behaviour means moving with the crowd: buying when the mood improves and prices rise, selling when fear spreads and prices fall. It is the natural behaviour of most investors, and in its professional form it is called momentum or trend following. Contrarian (anti-cyclical) behaviour means doing the opposite: buying when others panic and selling when others celebrate. The idea is that extremes in mood mark the point where the move has run out of fuel. ProcyclicalContrarian Buys when sentiment and prices rise when sentiment is depressed and prices have fallen Sells when sentiment turns to fear when sentiment is euphoric Strength catches long trends, avoids the deepest bear markets buys cheaply, sells into strength Weakness buys late and sells late, suffers in sideways markets with false signals catches "falling knives", sells too early in strong trends Psychology easy to do, feels comfortable hard to do, feels wrong at the moment of action 4. A short history of fear and greed in Bitcoin From February 2018 to October 2026 the crypto index spent 21 % of all days in extreme fear, 29 % in fear, 16 % in neutral territory, 25 % in greed and 9 % in extreme greed. Fear is more common than greed, which fits the long bear markets of 2018 and 2022. A few episodes show how differently extremes have played out: DateIndexBitcoin3 months later12 months laterWhat happened 26 Jun 2019 95 13,016 USD 8,621 USD 9,265 USD Euphoria marked the top of the 2019 rally; six months later Bitcoin was at 6,641 USD. 28 Mar 2020 8 6,242 USD 9,163 USD 55,951 USD Covid crash: one of the best buying opportunities in Bitcoin's history. 17 Feb 2021 95 52,149 USD 42,909 USD 40,538 USD Bitcoin still rose to 63,503 USD within two months before the crash of May 2021. 19 Jun 2022 6 20,553 USD 20,128 USD 26,851 USD Not yet the low: Bitcoin fell further to 15,787 USD in November 2022 (FTX collapse). 22 Nov 2024 94 98,998 USD 98,334 USD 84,648 USD Euphoria after the US election; the rally continued to 109,678 USD in May 2025. 16 Nov 2025 10 94,177 USD 69,768 USD open Fear came early: Bitcoin lost another third before the low in February 2026. 23 Feb 2026 5 64,617 USD 76,981 USD open Close to the low of the 2026 correction; 80,243 USD on 26 October 2026. The lesson from these examples: extremes often occur near turning points, but "near" can mean weeks or months, and the price can move a long way in between. Extreme greed in February 2021 was followed by another 22 % gain; extreme fear in June 2022 was followed by another 23 % loss. 5. Study 1: what happened after each mood in Bitcoin To get beyond single examples, every day from February 2018 was sorted by its index reading, and the Bitcoin return over the following 30, 90 and 365 days was measured. This gives several hundred observations per zone. The table shows the average return, the median return (the middle value, which is not distorted by a few huge outliers) and the share of positive outcomes. Zone on the starting dayNext 3 months: averagemedianpositiveNext 12 months: averagemedianpositive Extreme fear +2.6 % −3.7 % 46 % +99.4 % +36.0 % 61 % Fear +9.3 % +2.2 % 51 % +82.6 % +42.9 % 68 % Neutral +19.5 % +7.4 % 60 % +99.1 % +95.1 % 71 % Greed +16.4 % +4.8 % 58 % +54.7 % +36.8 % 61 % Extreme greed +39.9 % +10.4 % 60 % +63.1 % +16.1 % 63 % All days +13.9 % +3.3 % 54 % +79.1 % +41.7 % 65 % Three findings stand out: Short term, greed was followed by more gains, not fewer. Over three months the best results came after extreme greed (+39.9 % on average, +10.4 % median), the weakest after extreme fear (median −3.7 %, fewer than half of the cases positive). In the short run sentiment behaved like a trend signal: strong markets stayed strong, weak markets stayed weak. Long term, the picture turns. Over twelve months extreme fear produced an average gain of 99 %, extreme greed only 63 %, and the median after extreme greed was the weakest of all zones (+16 %). Euphoria tends to end, panic tends to end too, but it takes time. Averages hide a lot. After extreme fear the average twelve-month return (+99 %) is almost three times the median (+36 %). A few spectacular rebounds such as March 2020 pull the average up. A typical buyer in extreme fear did well, but not as well as the average suggests, and almost four in ten such purchases were still under water a year later. 6. Study 2: five ways to trade Bitcoin with sentiment The next step is to turn these ideas into simple rules and test them day by day from February 2018 to October 2026. To keep the test honest, the signal of one day is only applied to the return of the following day, every switch costs 0.1 %, and money outside the market earns nothing. Rule100 USD becamePer yearWorst fall from a highTime investedSwitches Buy and hold 874 USD +28.2 % −77 % 100 % 1 Contrarian: buy at extreme fear (below 25), sell at extreme greed (above 75) 104 USD +0.5 % −78 % 70 % 16 Contrarian: invested only while the index is below 50 60 USD −5.6 % −81 % 56 % 210 Procyclical: invested only while the index is above 50 988 USD +30.0 % −50 % 41 % 229 Procyclical: buy at extreme greed, sell at extreme fear 846 USD +27.7 % −41 % 30 % 15 The result surprises many readers. The famous contrarian rule, applied mechanically, was a disaster in Bitcoin: it sold early in every big rally (extreme greed appeared long before the tops of 2020/21 and 2024/25) and bought early in every bear market (extreme fear appeared months before the lows of 2018 and 2022). The procyclical rules, by contrast, earned about as much as buy and hold while being invested only 30 to 41 % of the time, and they roughly halved the worst loss. Why? Because Bitcoin has been a strongly trending market, and because the index itself is built partly from momentum. Fear clusters in falling markets and greed in rising ones. A rule that follows the index is, in effect, a trend-following rule. Two caveats: the procyclical rule with 229 switches would cost considerably more in practice through spreads and taxes, and the period from 2018 contains an exceptional rise of Bitcoin from about 9,000 to about 80,000 USD. A backtest describes the past; it is not a forecast. 7. Study 3: fear in a falling market and fear in a rising market If sentiment works partly like a trend signal, it makes sense to look at both together. Here every day was additionally marked by whether Bitcoin traded above or below its 200-day average, the most common simple definition of an uptrend. SituationCasesNext 3 months: medianNext 12 months: averagemedianpositive Extreme fear while below the 200-day average (falling market) 644 −3.2 % +107.7 % +38.9 % 63 % Extreme fear while above the 200-day average (sudden panic in a rising market) 35 −15.4 % −13.5 % −15.2 % 40 % Extreme greed (all cases were in a rising market) 281 +10.4 % +63.1 % +16.1 % 63 % Extreme fear after a long decline was, over twelve months, a good starting point; the first three months were still typically negative. The rare cases of extreme fear in the middle of an uptrend were often the start of a larger decline, not a buying opportunity. And a test of "follow the 200-day trend, but leave the market at extreme greed" earned only 12.6 % a year instead of 25.9 % for the trend rule alone: selling because of greed cost half of the return. 8. Study 4: stocks and the EllTec Analysis Fear & Greed Index Stock markets behave differently from Bitcoin: they are broader, less speculative and pay dividends. The EllTec Analysis Fear & Greed Index has a history back to March 2023 (weekly readings until October 2025, daily since then). For the USA and for the world, every reading was compared with the return of the S&P 500 ETF (SPY) and of the MSCI ACWI ETF over the following three months (63 trading days), dividends included. ZoneUSA: next 3 monthspositiveWorld: next 3 monthspositiveUSA: next 6 months Extreme fear +20.3 % (3 readings) 100 % +13.7 % (7) 100 % +28.3 % Fear +9.2 % (62) 97 % +8.4 % (63) 100 % +14.5 % Neutral +4.4 % (72) 88 % +3.9 % (67) 82 % +10.7 % Greed +3.4 % (179) 83 % +3.4 % (157) 89 % +8.8 % Extreme greed −0.6 % (9) 44 % +3.3 % (31) 84 % +8.8 % In equities the contrarian view looks clearly better: the more fearful the mood, the higher the following returns. The episodes of extreme fear were the yen carry unwind in August 2024 (global index 21), the tariff shock in April 2025 (global index 4, US index 17, the lowest readings in the history) and the correction in March 2026. All of them were followed by strong rallies. Extreme greed in the USA, mainly in the summer of 2023, was followed by a flat quarter. The limits of this study must be said clearly: three and a half years are short, the period was mostly a bull market, and there are only a handful of extreme readings. In a long bear market such as 2000 to 2002 or 2008, fear lasted for many months and early buyers lost a lot. Treat these figures as an illustration of a pattern, not as proof. 9. What research says The data above matches two well documented findings of financial research that seem to contradict each other: Momentum (procyclical): assets that have risen over the last months tend to keep rising for a while. Jegadeesh and Titman (1993) showed this for individual stocks, Moskowitz, Ooi and Pedersen (2012, "Time Series Momentum") for futures across equities, bonds, currencies and commodities: the return of the past twelve months helped to predict the return of the next months. Overreaction and reversal (contrarian): De Bondt and Thaler (1985, "Does the Stock Market Overreact?") found that the biggest losers of the previous three to five years later outperformed the biggest winners. Baker and Wurgler (2006) showed that after …