MVRV Z-Score

MVRV stands for Market Value to Realized Value. Market value is just Bitcoin's market cap (price × circulating supply). Realized value — or "realized cap" — values each coin not at today's price, but at the price it last moved on-chain, then sums the whole supply. That makes realized cap a rough proxy for the aggregate cost basis of every holder combined.

The Z-Score converts the gap between those two numbers into standard deviations from the historical spread, which is what makes readings comparable across market cycles of very different sizes:

where σ(Market Value)is the standard deviation of Bitcoin's market cap over its full history. Dividing by that standard deviation is the "Z-score" step — it rescales a raw dollar gap that grows with every cycle into a unitless number that has meant roughly the same thing in 2015 and 2025.

Suppose Bitcoin's market cap is $1.4 trillion, its realized cap is $0.8 trillion, and the standard deviation of its market cap over history is $0.1 trillion. The gap between market and realized value is $0.6 trillion. Divide that by the $0.1 trillion standard deviation:

A reading of 6 says the market is valuing coins six standard deviations above the aggregate cost basis — deep into the historically stretched zone, though not yet at the most extreme peaks prior tops reached. Now flip it: if market cap fell to $0.75T against a $0.8T realized cap, the numerator goes slightly negative, the Z-Score prints below zero, and the market as a whole is holding coins at an unrealized loss — the condition that has historically appeared only at the bottom of bear markets.

A high Z-Score means the market is valuing coins far above what most holders actually paid — historically a sign of large unrealized profit sitting on the table, which is exactly the condition under which profit-taking tends to accelerate. Readings in the upper extreme of a given cycle's range have coincided with each of Bitcoin's major cycle tops (2013, 2017, and 2021).

A low or negative Z-Score means the market is pricing coins near or below their aggregate cost basis — the market, in aggregate, is underwater. That condition has historically shown up during the deepest parts of prior bear markets, when capitulation selling is most exhausted. One caveat worth internalizing: the peak reading has fallen each cycle (2017 topped far higher than 2021 did), so the "extreme" band is best read relative to the current cycle's range rather than as a fixed line.

Treat MVRV Z-Score as a position-sizing gauge, not a trigger. Its value is in the tails: when it is deep in either extreme, it is telling you the risk/reward of adding versus trimming has skewed a long way from neutral. In the wide middle band — which is where it sits most of the time — it carries little edge, and reading too much into small moves there is the most common mistake.

It also should not be used alone. Because it is slow-moving and valuation-based, it pairs well with a faster momentum or sentiment read (see the Fear & Greed Index and the Bull Market Support Band) that can confirm whether an extreme valuation is actually starting to unwind.

MVRV-family valuation is part of the on-chain & valuation signal family — the most heavily weighted of the four families the Risk Model combines into a single 0–10 score. Rather than acting on any one indicator's threshold, the model blends valuation with cycle, momentum, and sentiment signals so that no single metric's false signal can dominate the read. See the published category weights on the Methodology page.

Realized cap is a proxy, not a survey of actual cost basis — coins that haven't moved in years (lost keys, long-term cold storage) still anchor to whatever price they last transacted at, which can understate the true distribution of holder cost basis. And like any historically-derived threshold, "extreme" is relative to the market's own past range, not a fixed number that necessarily holds in a structurally different future market.

It is also a lagging, coarse signal: it confirms that valuation has reached an extreme, but not when the extreme will resolve. Every cycle so far has seen the top-zone reading arrive lower than the last, so mechanically applying a threshold from a previous cycle would have signaled "top" too early in the following one.