AVIV Z-Score

Valuation z-score (MVRV-Z style), against price. Extremes have marked cycle tops and bottoms.

Data to 2026-09-11. On-chain metrics are fetched on a rotation, so this can trail the build by a few days.

How to read it

A z-score turns a valuation measure into standard deviations from its own long-run mean, which makes extremes comparable across eras. Readings far above zero say Bitcoin is expensive relative to its own history; far below zero says the opposite. It answers 'is this unusual?' rather than 'is this expensive?'

How it's calculated

The underlying valuation ratio is expressed as the number of standard deviations it sits from its historical mean.

What it did in past cycles

The all-time high reading is 3.09, set on 2013-11-18, a fortnight before that cycle's top. Cycle-peak readings since: 2.78 (2017-12-16), 1.87 (2021-11-08), 1.53 (2025-10-06) — the same downward drift visible in raw MVRV, which is a fair warning that normalising by standard deviations has not removed it.

Bottoms have come in near the floor: 0.37 (2015-01-14), 0.47 (2018-12-15), 0.51 (2022-11-09), against an all-time low of 0.20 in October 2011. The June 2026 low read 0.76, above all three prior bottoms.

What it doesn't tell you

Z-scores assume the past is a fair guide to what counts as normal. Bitcoin's history is short and its market structure has changed repeatedly — spot ETFs, institutional custody and derivatives did not exist for most of the sample — so the mean it is measured against is a moving target.

There is a subtler problem: every new day of data is added to the sample that defines 'normal'. A long stretch of high valuations raises the mean and widens the distribution, which mechanically pulls future z-scores toward zero. The indicator is quietly grading on a curve, and the curve moves.

Reading it with other metrics

This is the metric to reach for when MVRV's absolute thresholds feel unreliable — which, given how far its cycle peaks have fallen, is most of the time. The two are built from the same valuation idea, so they agree on direction and differ only on scale.

Pair it with something structurally different for confirmation: the 200-week moving average for a price-only view of the same question, or the power law, which measures distance from a fitted long-run trend rather than from a statistical mean.