Fear & Greed Index: does fear predict Bitcoin’s next move?

The Fear & Greed Index captures sentiment around Bitcoin, but it hasn’t proven to be a reliable standalone predictor of its next move. On November 7, 2026, Alternative.me recorded 25, “extreme fear”, versus 28 the day before and 50 on November 5. The change describes the state of the market, not where the price will be tomorrow.

The promise is tempting, though: if we could spot when panic is overdone, couldn’t we also spot when the market overreacts? We think two things that often get lumped together need to be separated here: measuring fear and predicting a reversal.

What does the Fear & Greed Index measure?

The daily index from Alternative.me scores sentiment from 0 to 100, with low values corresponding to fear and high values to greed. Despite the word “Crypto” in its name, the published methodology currently covers only Bitcoin.

It doesn’t ask all investors how they feel. It combines market data and online activity:

The first two components are compared with 30-day and 90-day averages. Social media covers activity on Twitter/X, not live analysis of Reddit. There is a transparency gap, though: it isn’t clearly explained how the weights are redistributed while the surveys remain inactive.

Because price and volatility feed into the calculation, fear may partly reflect losses that have already happened. Think of a thermometer: it shows the fever, but its reading doesn’t tell you exactly when it will break.

Why does “when everyone is afraid” sound so convincing?

The contrarian argument rests on the likelihood of an overreaction, not on some law of the market. Panic can push prices down sharply, while euphoria can come with excessive speculation.

In his 1986 letter to Berkshire Hathaway shareholders, Warren Buffett wrote that he tried to be fearful when others are greedy and greedy only when others are fearful. He was talking about stocks and business value, not Bitcoin. In the same passage, he described the timing, duration and degree of these excesses as unpredictable.

That second part often gets lost. A market can be fearful and keep falling. It can also be euphoric and keep rising.

What did the Bitcoin forecasting tests show?

The more rigorous academic tests of 2026 found no reliable improvement in return forecasts from using the index. That carries more weight than one striking example of a rebound after a panic.

Louis Gessie, in a March paper in Finance Research Open, examined daily data from 2018–2025. Changes in the index did not improve forecasts on data outside the estimation sample. Instead, Bitcoin’s returns helped predict subsequent changes in sentiment. Similar results emerged at a weekly frequency and with additional uncertainty measures. This predictive relationship does not prove actual causation.

The study by András Szeberényi and Mátyás Imre Kovács, published on September 1 in Digital Finance, examined 2,945 observations from February 1, 2018 to February 27, 2026. Higher readings were associated with higher subsequent one-day and seven-day returns, but the seven-day relationship became uncertain under a more conservative statistical test.

Above all, the models didn’t beat a forecast based on the historical average for horizons of one, seven or 30 days. What’s more, extreme fear was followed by extreme fear in 82% of cases, and greed by greed in 80%.

What do historical returns after fear show?

Historical comparisons give different answers depending on the time horizon, the thresholds and the episodes they include. A positive return after fear doesn’t automatically mean a better return because of fear.

The independent analysis by WhyCoinMoves, covering February 2018–May 2026, found a median one-year gain of 36.5% after extreme fear, versus 48.6% for all days. Over 90 days, however, the median return after readings of 0–24 was −4.5%, versus +3.5% overall. A positive outcome occurred in 44% and 54% of periods, respectively.

The 637 days of extreme fear weren’t independent experiments. Counting a new episode only after at least 30 days without extreme fear left 13 episodes: seven had a lower price 90 days later, with a median return of −8.6%. A small sample, overlapping periods and no costs limit the conclusions.

K33 Research, led by Vetle Lunde, reported on February 17, 2026 an average 90-day return of 95% after readings above 90, versus 2.4% after readings below 25. It also calculated a correlation of 0.71 with the previous 30-day return, versus 0.14 with future returns. The future-return data shown had a reference date of November 19, 2025.

How fragile are the averages? In an independent backtest by CycleSolved through September 22, 2026, the average 90-day gain after readings of at least 80 was 63.1%. Removing one extended period in 2020–2021 cut it to 0.1%.

When did fear and greed persist?

The index’s history includes prolonged sentiment regimes, not just turning points. CoinGecko’s data and Alternative.me’s records show why timing remains hard.

1. February 2018 The academic study’s daily dataset begins. It doesn’t cover all of Bitcoin’s history. In 2018, 288 of 334 days were fearful, or 86.2%.
2. July 14–15, 2019 The index dropped from 61 to 16 in a single day.
3. November 2020–February 2021 An almost unprecedented stretch without fear coincided with Bitcoin first breaking above $40,000. In total, 2021 had 199 days of greed.
4. 2022 There were 343 days of fear, 207 of them extreme fear. Fear covered the entire fourth quarter, including the collapse of Sam Bankman-Fried’s FTX.
5. November 2023–February 2024 Four straight months passed without a single day in the fear category.
6. February 12, 2026 The Alternative.me index hit 5, the lowest reading on record up to that point, as reported by The Block.

How do we tell a forecast from a nice story?

A reliable test needs a predefined rule, a time horizon and a benchmark. “It went up afterward” isn’t enough without a comparison with the historical average, simple buy-and-hold or price momentum.

It also needs realistic execution timing after the reading is published, fees, the bid–ask spread, slippage and taxes where they apply. For readers in Greece, a theoretical result without these is not the same as a net return.

Our position would change with repeated success on new data, after costs and without depending on a single bull run. Until then, we see the index as a descriptive tool. A February 2026 preprint by Murad Farzulla linked extreme sentiment to greater uncertainty, but the result was sensitive to the statistical method and didn’t cleanly separate it from the volatility built into the index.

Fear can be measured. A reliable market reversal doesn’t automatically follow from measuring it.