The Fear & Greed Index for Long-Term Holders

The Fear & Greed Index is the most quoted number in crypto that nobody reads carefully. It appears in headlines as “the market is in extreme fear” and disappears again the next morning when the print has moved fifteen points. For someone holding across a cycle, that churn is noise wrapped around a genuinely useful signal.

This article takes it apart: what goes into it, why it is a bitcoin index rather than a crypto index, why we use a seven-day average, and how often the extremes everyone talks about have actually occurred. Every frequency below is computed from our own copy of the series — 3,094 daily prints from 1 February 2018 to 26 July 2026.

What the index is made of

The index is published by alternative.me, once a day, as a single integer from 0 to 100. Zero is maximum fear, one hundred is maximum greed. Six inputs feed it, with published weights:

Input	Weight	What it is meant to capture
Volatility	25%	Current volatility and drawdown against 30- and 90-⁠day averages. Unusual volatility is read as fear.
Market momentum and volume	25%	Buying volume against its own 30- and 90-⁠day averages.
Social media	15%	Interaction rates on crypto hashtags.
Surveys	15%	A public poll. Currently paused, so this share is redistributed in practice.
Bitcoin dominance	10%	Bitcoin’s share of total crypto market capitalisation, read as a flight to relative safety.
Google Trends	10%	Search volume for bitcoin-related queries, weighted by how the query itself reads.

Two things follow. Half the index is volatility and volume — it is substantially a price-derived measure wearing a sentiment label, and a sharp drawdown produces a low print whether or not anyone feels afraid. And the survey component is currently paused, so the published number comes from five inputs rather than six, with the rest carrying more than their nominal share.

We use the index anyway, at a 25% weight inside the Hodlometer Index, because it is the only component of ours that is not derived from our own price maths. An imperfect independent axis beats a fourth restatement of price.

It is a bitcoin index, not a crypto index

This is the single most common misreading. alternative.me publishes one print, and that print is for bitcoin. There is no per-asset version — no ether index, no “altcoin” index. Bitcoin dominance appears inside the calculation as an input, which is a different thing from covering the rest of the market.

Our own index applies the bitcoin print to ether as well, and we would rather say that plainly than bury it. Bitcoin sets the cycle for both, and sentiment is the least asset-specific of our four components. The cost shows when ether diverges — as it has through 2026, sitting about a quarter below its four-year baseline while bitcoin sits roughly on top of its own. This component cannot see that difference; the other three, carrying three quarters of the weight, can.

Why a seven-day average, not the daily print

The daily print is a jumpy series. Across the whole history the average day-to-day move is 4.3 points, and the largest single-day move on record is 45 points — from 61 to 16 between 14 and 15 July 2019. On a seven-day average the average day-to-day move is 1.3 points. That is the whole argument in two numbers.

The practical consequence shows up at the thresholds. The daily print has crossed the 20 line 134 times in eight and a half years; the seven-day average crossed it 46 times. Above 85 the ratio is similar: 26 crossings against 10. Every one of those extra crossings is a headline that says the market has entered or left extreme territory, and most of them reverse within days.

Measure	Daily print	7-⁠day average
Average day-to-day move	4.3 points	1.3 points
Times it crossed the 20 line	134	46
Times it crossed the 85 line	26	10
Separate episodes below 20	67	23
Separate episodes above 85	13	5
Lowest reading	5	7.4
Highest reading	95	94.0

Note what smoothing does not cost you: the lowest seven-day average, 7.4 on 20 June 2022, sits only marginally above the lowest daily print of 5. The averaged series still reaches the extremes, it just refuses to visit them for an afternoon.

There is also a bookkeeping benefit. A quarter of the days when the daily print came in below 20 — 91 of 346 — had a seven-day average that was not below 20 at all. A holder tracking the daily number logs a quarter more “extreme fear” events than happened by any sustained measure.

How often the extremes actually happen

The contrarian framing that circulates — buy fear, sell greed — usually arrives without any statement of how often those conditions occur. Here they are, on the seven-day average, across 3,076 windows:

Condition	Days	Share of history	Separate episodes	Longest episode
7-⁠day average below 20	335	10.9%	23	62 days (10 May – 10 Jul 2022)
7-⁠day average above 85	90	2.9%	5	67 days (11 Nov 2020⁠–⁠16 Jan 2021)
7-⁠day average above 75	272	8.8%	—	—

The asymmetry is the interesting part. Sustained extreme fear is roughly 3.7 times more common than sustained extreme greed. Some of that is the construction of the index — a volatility-heavy measure will print low during any violent move, and violent moves down are more violent than moves up. Some of it is that bear phases simply last longer than melt-ups.

The greed side is rarer than most people assume. In eight and a half years the seven-day average has spent 90 days above 85, and 83 of those 90 fall inside the winter of 2020–21. Six more sit in a single week of November 2024, and the last one is an isolated day in June 2019. If your mental model is that extreme greed shows up once or twice a year, the data disagrees — it shows up roughly once a cycle, and then not at all for years.

Fear, by contrast, is a regular visitor. 2026 alone has produced four separate episodes below 20 totalling 100 days, through 26 July. Full-year averages of the daily print tell the same story: 30.8 in 2018, 52.9 in 2020, 25.3 in 2022, 63.3 in 2024 and 22.0 so far in 2026.

What we deliberately do not publish

We do not publish a table of average forward returns by fear level. It would be easy to produce: take every day the seven-day average was below 20, measure what bitcoin did over the following year, print the mean. The number would be real and the table would be meaningless.

Those 335 days sit in 23 episodes, and days inside an episode are not independent observations — the effective sample is closer to twenty than to three hundred. Those episodes are concentrated in a handful of drawdowns across barely more than one halving cycle, so any average would be dominated by what happened next in 2018, 2020, 2022 and 2026: the shape of four events, dressed as a probability.

What the frequencies above can tell you is how unusual today is. That is a smaller claim, and one the data supports.

Where it sits today

On 26 July 2026 the Fear & Greed Index printed 26, and the seven-day average was 28.4. That is fear, but not extreme fear, and it is a recovery from the run below 20 that ended on 6 July.

Inside the Hodlometer Index the figure that matters is the seven-day average for the previous closed day: 28.7 on 25 July 2026, which ranks as a percentile score of 25.8 out of 100 against the trailing four-year window. That ranked third of the index’s four components: warmer than the 200-week distance at 17.2, colder than Mayer at 28.0 and funding at 35.2. The two price components average 22.6 between them, so sentiment is currently a shade less depressed than price.

For context on where 28 sits in the full distribution: the all-history median daily print is 44, the 25th percentile is 26 and the 10th percentile is 18.

Limitations worth knowing

• The method is not reproducible. The provider publishes weights, not inputs or code. We cannot recompute the series, only record it.
• It is partly a price indicator. Volatility and volume are half the weight. When people call it an independent read on sentiment, that is only three quarters true at best.
• One print, applied to two assets. Ours is a site for holders of bitcoin and ether; this component only sees bitcoin.
• Eight and a half years is a short history. It covers one complete halving cycle plus parts of two others. Percentile statements from a sample this size describe; they do not predict.
• Smoothing costs a few days. A seven-day average confirms a regime change roughly three to four days after the daily print does. That is a deliberate trade, and it is a bad trade if your horizon is a week.

How a holder can use it

As a mood check on yourself, mostly. It is at its most useful when it disagrees with you: if the market feels like it is falling apart and the seven-day average reads 28, the panic is probably local to your own timeline. Sustained readings below 20 happen on 11% of days — if you cannot name a plan for those days in advance, you will improvise one under pressure.

It is a poor timing tool and an even worse standalone one. Pair it with something price-based on a different horizon — the Mayer Multiple for the seven-month view or the 200-week moving average for the full-cycle view — and note the reading whenever you make a decision. Two years later that log is the only way to tell whether your process worked or the market simply went your way.

Frequently asked questions