Fear & Greed Index

The Fear & Greed Index plots one daily scalar between 0 and 100 against the same regime cuts the index has used since launch. Below 25 is Extreme fear; between 25 and 50 is Fear; between 50 and 75 is Greed; above 75 is Extreme greed. The four background bands on the chart correspond to those regimes. The muted line is Bitcoin’s USD price on the log left axis for cycle context.

Today’s reading is 25 — Fear. The trailing 7-day average is 32; the 30-day average is 23. The series carries 2,990 daily prints from 01 Feb 2018 through to the most recent close on 01 May 2026.

Six components combine with fixed weights into the daily 0–100 scalar. Each sub-score is normalised to 0–100 in its own right, then weighted into the composite:

Five of the six components derive from market data: volatility, volume, dominance, and the price-sensitive parts of the social / trends signals all move with the tape. Only the survey component is independent — and that component is currently paused. The structural consequence is that the index will move with price more often than it moves before price; a regime confirmation, not a regime forecast.

The four regime cuts are descriptive, not actionable on their own. Below 25 is Extreme Fear; from 25 to 50 is Fear; 50 to 75 is Greed; above 75 is Extreme Greed. Each band corresponds to a band of historical forward returns — documented in the cluster table further down — and the regime label is most useful as a context check on other indicators rather than as a standalone trigger.

Reading every cycle anchor since the index launched surfaces the regime-shift pattern. The 2021 cycle-top window printed sustained Extreme Greed for weeks around both peaks; the 2022 stress window dragged the reading into the single digits twice; the 2024 pre-halving high cleared 90 in early March. Prices are our own daily closes; the regime label is the canonical 0–25–50–75 split.

The cluster table is the page’s distinctive read. For every daily print of Extreme Fear (≤25) and Extreme Greed (≥75) since the series began, the median 30, 90, and 180-day forward BTC return is computed against our own daily-close series. The sample size is what most competitor pages omit — without it, “buy fear, sell greed” is a slogan; with it, the return distribution is a fact you can read off.

Three observations carry through the cluster table. First, the “buy fear” slogan has held in the data — with caveats. The median forward 90-day return from an Extreme Fear print since 2018 is positive, but the distribution is wide: individual prints during the November 2022 post-FTX week were followed by months of further drawdown before the recovery began in earnest. The median is real; the tail is real too.

Second, “sell greed” carries less force than the slogan implies. Extreme Greed regimes can persist for weeks before a topping inflection — the spring 2021 run held above 75 for nearly two months, the November 2021 top held for several weeks, and the March 2024 pre-halving high spent over a fortnight above 80. A practitioner analysis of the index documented this directly: the strategy “was only profitable with proper trade management by slowly scaling in and out over macrocycles.” Conditions can stay irrationally fearful or greedy for months at a time.

Third, the index is most useful in compressed form. The single-day reading whips on event noise; the 30-day average is the slow regime read. The cluster table treats every daily Extreme print as one observation, which over-weights long stretches of unbroken Extreme Fear or Greed; an alternative cut grouped by regime episode rather than by day would yield a smaller sample size with similar median direction. The single-day formulation is the simplest reproducible cut.

For a dollar-cost-averaging investor. The cluster table earns Extreme Fear prints a small tactical accelerator role — the median forward return is positive on a 90-day horizon. Treat them as a confirmation that the regime is fearful, not a confirmation that the bottom is in. The November 2022 window had four months of single-digit prints before the recovery; the 2018 bottoming had a similar arc. Add to a regular cadence on Extreme Fear; do not switch from cadence to lump-sum on a single print.

For a cycle-timing trader. The 30-day average crossing back through 50 from below has historically marked the start of multi-month recoveries; the same average crossing 75 from above has historically marked late-cycle distribution windows. Pair with MVRV-Z for the cohort-cap cross-check and realised volatility for the underlying input that drives the largest single component. The index alone is a slow signal; the 7-day average leads it by less than its own 30-day lag.

For a researcher. The series, the regime cuts, and the forward-return cluster are reproducible from the daily 0–100 scalar and our own daily-close BTC series. The sample size and median in the cluster table refresh every night; a researcher wanting the per-print distribution can compute it from the same two series.

The index is coincident, not predictive. Five of six components are price-derivative; the survey component that would have provided an independent input is currently paused. Sustained Extreme regimes can persist for months, and the index will follow rather than lead the next inflection. The November 2022 post-FTX stress window is the canonical case — sustained single-digit prints for weeks before the cycle low actually printed.

The single-day print whips on event noise. A liquidity flush, a regulatory headline, or a macro print can produce one-day swings of 20 points that retrace the next session. The 30-day average is the slower regime read; the one-day reading without context is the headline only on event days.

The composite hides component disagreement. Many useful regimes — rising volatility with falling volume, or rising dominance with falling momentum — collapse into a single index value that obscures the disagreement. Reading the index without checking the components is reading the summary statistic without checking the distribution.

The methodology is opaque in its weights but firm in its 0–100 output. The index publishes daily values without daily component breakouts; the weights are public, but the per-component daily contributions are not. A reader auditing why a particular day printed 47 rather than 53 cannot reproduce the calculation from the public methodology alone.

Canonical questions from Google’s “People also ask” block for bitcoin fear and greed index and crypto sentiment indicator, answered against the data on this page.