Buy Bitcoin on extreme fear? We tested the fear and greed index on 3,157 days

What the index is

The Crypto Fear & Greed Index, often called the Bitcoin fear and greed index, is published once a day by alternative.me. It runs from 0, extreme fear, to 100, extreme greed. By the publisher's own description it blends Bitcoin's volatility (25% of the score), momentum and trading volume (25%), social media (15%), surveys (15%, currently paused), Bitcoin's share of the whole coin market (10%) and Google searches (10%).

The publisher also says how to read it: “Extreme fear can be a sign that investors are too worried. That could be a buying opportunity.” And when investors get too greedy, “the market is due for a correction.” That advice is the belief we tested.

How we tested it

We took every daily reading from 1 February 2018 to 27 September 2026, 3,157 days in all, and Bitcoin's change in US dollars over the 24 hours after each reading was published, and over the 7 days after. The reading is published at midnight UTC, so a reader could act on it before any of those hours. Then we compared the days of extreme fear with all other days, in two halves: February 2018 to May 2022, and June 2022 to September 2026.

To measure luck, we slid the whole series of readings against Bitcoin's prices by a random number of days, at least 30, and repeated that 2,000 times. Each slide keeps the index's runs of fear and greed but breaks any real link with what came next. The rules for a pass were written down before the index's history was downloaded: extreme fear had to be followed by better days than the rest in both halves, by a margin that fewer than 5% of the random calendars could match, and by a better week in both halves too.

What we found

The day after an extreme-fear reading, Bitcoin rose 0.084% on average; after all other days it rose 0.132%. Fear days were a little worse, not better, and they were worse in both halves: by 0.053% a day in the first and by 0.049% in the second. To beat 1 in 20 of the random calendars, fear days would have needed a lead of 0.258% a day. As it was, 66.8% of the random calendars fitted the belief at least as well as the real one.

A week did not change the picture. In the first half, the 7 days after extreme fear gained 1.58% on average against 0.91% after other days; in the second half they lost 0.04% against a gain of 0.89%. Across the whole period it was +0.85% against +0.90%.

Year by year

No steady pattern shows by year either. Extreme-fear days were followed by better days than the rest of the year in five years, by worse days in three, and in 2019 the two were level. In 2022, the year with the most fear (207 days), fear days were followed by +0.15% and the year's other days by −0.71%, the weakest of any year. 2023 and 2024 had only 3 and 8 days of extreme fear, too few to judge.

Buying the first day of fear

A trader might wait for the first day the index turns to extreme fear, rather than buy on every fearful day. There were 39 such days, counting a day as a first only if the last extreme-fear day was at least 7 days earlier. The 7 days after them lost 2.11% on average, while any 7 days gained 0.89%, and only 16 of the 39 ended higher. Over 30 days it was −0.42% against +3.88% for any 30 days, again with 16 of 39 higher.

These figures describe the data; they were not among the rules set before the test, and 39 cases are few. Still, they give no sign that the first day of fear is a buying day.

And extreme greed?

The other half of the advice is to step aside when greed is extreme. Over the whole period the day after extreme greed was better than the rest by 0.229% on average, the opposite of the belief, and 86.9% of the random calendars fitted the belief at least as well. The two halves disagree: +0.497% a day on 185 days from February 2018 to May 2022, and −0.271% on 97 days since June 2022. We tested only the publisher's reading, not the reverse.

Could you trade it?

Not on this evidence. The test counted no trading costs at all, and the edge was missing even before them. Bitcoin is also too jumpy for small effects to show: an average day gained 0.120%, with a typical swing (standard deviation) of 3.28%, and the rules said before the run that only a large effect could be seen. A miss means there is no large effect; a small one could still hide in the noise.

Half of the index is built from Bitcoin's own volatility and momentum, so it partly restates what the price has just done. That makes it a fair summary of the market's mood. As a signal for the next day or week, it did no better than chance.

Questions people ask