The Median Day on the Fear and Greed Index Reads Fear

Today the Crypto Fear & Greed Index reads 78 — Extreme Greed. That sounds like an outlier, and it is: only 8.9% of its published days have ever scored that high. What is less obvious is the other end of the scale. Across all 3,152 readings the index has published since February 2018, the median day scores 44 — which the index itself labels Fear.

Today: 78 · Extreme Greed · Median reading: 44 · Fear · Days published: 3,152 · Neutral band: 12.7% of days · Fear or worse: 52.2% of days

A gauge that is not centred

Most people read the index the way they read a thermostat: 0 is cold, 100 is hot, 50 is the middle. The published record does not behave that way. The mean reading is 45.6, the median is 44, and the band the index calls Neutral runs from 47 to 54 — eight points out of a hundred, covering 12.7% of all days.

That means the arithmetic centre of the scale sits above the typical day. A reading of 50 is not the average day; it is the 57th percentile. Bitcoin sentiment, as this index measures it, spends more of its life anxious than calm: 52.2% of readings land in Fear or Extreme Fear, against 34.9% in Greed or Extreme Greed.

None of that is a flaw in the index. It is a property of what it measures, and it is only misleading if you assume the midpoint is the mood of a normal day. It is not.

A reading of 50 is not the average day. It is the 57th percentile.

Where the bands actually are

The band edges are worth pinning down, because sites that display this index do not always agree on them. We read ours off the publisher's own labels rather than assuming them: every reading arrives with a classification attached, so the boundaries can be recovered exactly.

Extreme Fear runs to 25 and covers 23.4% of days. Fear runs 26 to 46 and is the single largest band at 28.8%. Neutral is 47 to 54, the narrow strip already mentioned. Greed is 55 to 75 at 26.0%, and Extreme Greed is 76 and above at 8.9%. The full range ever printed is 5 to 95 — the index has never touched either end of its own scale.

If a chart shows you a Fear/Neutral boundary somewhere other than 46/47, it is using a different set of edges from the one the data publisher uses, and the same number will carry two different labels depending on which site you are reading.

What it is made of

The publisher documents six factors with fixed weights: volatility 25%, market momentum and volume 25%, social media 15%, surveys 15%, Bitcoin dominance 10%, and search trends 10%.

The surveys component is marked “currently paused” on the publisher’s own methodology page, and the reddit half of the social-media component is described as still not in the live index. So a documented weighting that adds to 100% is not the weighting actually being applied to today’s number, and the published description is ahead of the published index.

The larger point is what the live components have in common. Volatility, momentum and volume are all computed from price and trading activity. When price falls hard, volatility rises and momentum turns negative, and the index drops — partly because sentiment soured, and partly by construction.

That matters for how much independent information the reading carries. A sentiment gauge built substantially from price is closer to a restatement of recent price action than to an outside opinion about it. It is still useful as a summary, but it is not a second witness.

The sample problem nobody mentions

The index is most often used as a contrarian trigger: buy Extreme Fear, sell Extreme Greed. That is a testable claim, so we tested it against our own daily close series.

The trap is that daily readings overlap. If you measure the forward 90-day return from every single day, you get three thousand results that are almost entirely the same few price moves counted over and over, and any pattern in them looks far more solid than it is. Using only non-overlapping 90-day windows — the honest version — 8.6 years of history yields 35 independent observations in total, spread across five bands. Two of them fall in Extreme Greed. Three fall in Neutral.

With samples that small, the medians are not worth printing as guidance, and we are not going to print them as guidance. What we will say is that they do not line up neatly with the folklore: the median non-overlapping window starting in Extreme Fear was mildly negative over the following quarter, and the widest single outcome in the whole set — a gain of 177% — started from Greed, not from fear.

How to actually use it

Read it as a description, not an instruction. The index tells you what the mood was, in one number, on a scale whose typical day is 44. That is genuinely useful context to have beside a price chart, and it is a reasonable way to notice that a week felt unusual.

Read the band, not the point. The difference between 44 and 46 is not information; the difference between 22 and 78 is. The bands are wide because the underlying signal is noisy.

And hold the contrarian reading loosely. Not because it is wrong — this data cannot show that either way — but because eight years of a single asset's history is not enough evidence to run a strategy on, however many decimal places the index prints.

Frequently asked questions

Is a high reading a sell signal? The index itself makes no such claim, and the record here cannot settle it. Only 35 independent 90-day windows exist in the whole published history, two of which began in Extreme Greed. That is not a sample you can build a rule on, in either direction.

Why does our reading sometimes differ from another site's? Two reasons. The published value updates once a day, so two sites can be showing different days. And the band edges vary between sites: ours are read off the publisher's own labels, where Fear ends at 46 and Neutral begins at 47. A site that draws that line at 45 will call a reading of 46 Neutral where we call it Fear.

Does it cover more than Bitcoin? The composite includes Bitcoin’s share of total crypto market capitalisation, so the wider market affects it, but it is published as a Bitcoin sentiment index and that is how it should be read. It is not a gauge of any individual altcoin.

What to watch instead

If what you want is a measure of whether the market is expensive rather than how it feels, the metrics that measure a quantity are more useful. The Mayer Multiple compares price to its own 200-day average, which is arithmetic rather than mood. MVRV compares market value to the aggregate on-chain cost basis, which is what holders actually paid.

Neither is a forecast either. But both are computed from something you can point at, and when they disagree with sentiment that disagreement is itself informative — a market that feels euphoric while trading close to its own cost basis is in a different position from one that feels euphoric at three times it.

The Fear & Greed Index is a good summary and a poor trigger. It compresses several noisy inputs into one number a reader can absorb in a second, which is a real service. What it cannot do is carry the weight people put on it: a scale whose median day reads Fear, built partly from price itself, with fewer than forty independent quarters behind it, is a thermometer rather than a forecast. Check it, note the band, and go look at something that measures a quantity.

Related reading: Are We There Yet? on valuation instead of mood, and The Strategy That Wins Less Often on why small samples mislead.

This essay is part of BTCDash Research. Figures were computed on 2026-09-22 from BTCDash’s own data. Nothing here is financial advice — it is analysis and background for your own research. Bitcoin is volatile; do your own diligence.