MVRV Z-Score Explained: Formula, Data and Limits

MVRV is bitcoin's market capitalization divided by its realized capitalization, which values every coin at the price it last moved on-chain instead of today's price. On October 21, 2026 Coin Metrics' series put the ratio at 1.56. The MVRV Z-score subtracts realized cap from market cap and divides by the standard deviation of market cap, which on the same day computes to about 0.98. Both describe how far the market's price sits above the network's estimated average cost basis. Neither predicts what comes next.

The arithmetic first: realized cap on three coins

Realized capitalization is a variation of market cap that values each coin by the price when it last moved. Glassnode's documentation describes it as valuing "each UTXO based on the price when it was last moved, as opposed to its current value," and notes it is often used as a proxy for the network's aggregate cost basis. Bitcoin's ledger is built from unspent transaction outputs (UTXOs): each one is a chunk of coin with a record of when it was created. The toy example uses three.

Suppose bitcoin trades at $50,000 today. Market cap is 6 times $50,000, or $300,000. Realized cap is $192,000. MVRV is 300,000 divided by 192,000, or 1.5625. Read plainly: at this price, the average coin is worth about 56 percent more than the price at which it last changed hands.

Now move one coin. UTXO A, the 2 BTC last moved in 2020, is sent to a new address today. Its realized value resets from $6,000 times 2 to $50,000 times 2, or $100,000. Realized cap rises from $192,000 to $280,000, and MVRV falls from 1.5625 to 300,000 divided by 280,000, or 1.0714. No one bought or sold anything, the market cap did not change, and the ratio dropped by almost a third. The only thing that happened is that old coins were spent, and the metric treats a spent coin as having been "bought" at the price of the spend. Everything that follows has to be read with that in mind.

From ratio to Z-score

The MVRV ratio is dimensionless: 1.0 means price equals the average last-moved price, above 1 means the average holder is in profit on paper, below 1 means the average coin was last moved at a higher price than today's. Coin Metrics defines its series, CapMVRVCur, as "the ratio of the sum USD value of the current supply to the sum realized USD value of the current supply."

The Z-score rescales the gap between the two caps by how volatile market cap has been:

MVRV Z-score = (market cap minus realized cap) divided by the standard deviation of market cap

The idea, as secondary explainers describe it, is that a difference of $500 billion means something different when market cap historically wanders by $50 billion than when it wanders by $600 billion. Glassnode's metric guide dates the Z-score to October 2018; who created it is covered below. The original article could not be retrieved for this page, so the formula here is as the vendor guide and explainers state it and as reproduced in the computation below.

Using the toy numbers makes the point. Market cap minus realized cap is $300,000 minus $192,000, or $108,000. If the standard deviation of market cap over the history were $60,000, the Z-score would be 1.8. Double that standard deviation and the Z-score halves, even though the gap is the same. The denominator does a lot of work, and as a result, the choice of denominator is where chart providers differ.

Reading the real series: how the numbers below were computed

Coin Metrics publishes daily bitcoin market cap (CapMrktCurUSD) and MVRV (CapMVRVCur) through its community API. Realized cap is recovered as market cap divided by MVRV. The standard deviation here is the population standard deviation of daily market cap over the whole history from July 18, 2010 up to each day (an "expanding" window), which is the plain reading of the formula. The data were downloaded on October 23, 2026; the last observation is October 21, 2026. Daily prices come from the same source (PriceUSD).

The following table shows the readings on dated extremes. These are observations about the past, not a pattern to extrapolate.

Four things stand out when the numbers are laid next to each other.

Why the 2026 readings look low

Between October 6, 2025 and October 21, 2026, bitcoin's market cap fell from about $2.49 trillion to about $1.69 trillion, while realized cap moved from about $1.09 trillion to about $1.08 trillion. Realized cap barely changed because the metric only reprices coins that move. The ratio fell from 2.29 to 1.56 because the numerator dropped and the denominator hardly did. On June 30, 2026 the ratio touched 1.10, the lowest reading since March 2023, and the expanding-window Z was 0.18. That is a description of the past year's path, and the table says nothing about what follows.

Splitting the Z-score into its two halves

A Z-score is a numerator over a denominator, and most commentary talks only about the numerator. Taking the same Coin Metrics series apart shows that the denominator, the standard deviation of market cap, explains most of why successive peaks read lower.

Compare the two 2021 rows. The gap between market cap and realized cap is almost the same on both dates, $800.8 billion in February and $820.0 billion in October. The Z-score fell from 7.15 to 3.67 because the denominator doubled, from $112.0 billion to $223.6 billion, in eight months. Nothing about how stretched price was relative to cost basis changed in dollar terms; the yardstick grew. By October 21, 2026 the standard deviation is $618.5 billion, larger than the whole gap, which is why a ratio of 1.56 prints a Z-score under 1.

Two consequences follow. A Z-score threshold from one era is not comparable to a reading from another era on an expanding window, even before any question of what the threshold means. And the metric's apparent "shrinking peaks" are partly produced by its own construction, not only by a changing market. Both are reasons to read the ratio, which has no such drift, alongside the Z-score.

Year by year, from the same series

The table collapses 5,400 daily readings to one row per year. Realized price here is the day's price divided by the day's MVRV, which is realized cap per coin in circulation, so a price below it means the ratio is below 1. The last four columns count days in each MVRV band. Because the Z-score is negative exactly when market cap is below realized cap, the first band's count is also the count of days with a negative Z.

Source: Coin Metrics community API, CapMVRVCur, CapMrktCurUSD and PriceUSD, downloaded October 23, 2026; Z-score on the expanding window described above; realized price is our division.

Reading it as description, not pattern: the price closed a year below realized price only in 2018 and 2022, with 2014 closing at parity. Three years, 2013, 2017 and 2021, spent between 89 and 97 days with MVRV at 3 or higher. 2024 and 2025 spent none, and so far 2026 has not spent a day above 2. A reader could take that as evidence the metric has changed, as evidence the market is different, or as a sample of four cycles that cannot distinguish the two. The same series in dollars, with every run of closes below realized price, is in bitcoin's realized price, measured. The table does not choose among them.

Variants: which chart uses which standard deviation

"The MVRV Z-score" on two different sites can differ by a point or more, because the formula leaves the choice of denominator open. The table shows how much the choice matters on the same data.

Two lessons. First, a threshold quoted for one variant does not transfer to another: a reading of 7 on the expanding window corresponds to about 5.9 on the rolling four-year version on the same day in February 2021. Second, the gap is largest in earlier years and narrow now, because the recent four years of market cap make up most of the full history. When you see a chart, find out which window it uses before comparing it to a number from a guide.

Other variants exist, such as using a moving average of the MVRV ratio instead of the Z-score, or dividing by a different measure of dispersion. The principle is the same: whatever the denominator, the metric describes the gap between price and a cost-basis estimate and does so relative to a historical yardstick.

Where the metric misleads

The caveats are not footnotes. They are the main reason that this metric should be read as context.

Glassnode's documentation states the repricing point directly: spending a coin resets its valuation to the current price. It also cautions that a coin moved at the same price at a later date has no effect on realized price, and that dormant or lost coins are discounted as low in economic value. The last row is the most important. With four cycles on record, any claim of the form "when the metric crossed X, price then did Y" rests on a handful of events, and the 2021 example above shows the events do not all behave alike.

What the metric can and cannot tell you

A descriptive metric can answer descriptive questions.

It cannot tell you any of the following, and the data above do not support those readings.

The dates in these tables sit inside the issuance cycles laid out in our halving cycles history. For how headline supply-based numbers get distorted, see market cap versus fully diluted value, and the bitcoin volatility guide measures how large drawdowns have actually looked. The coins behind the metric are the outputs you can inspect by reading a transaction in a block explorer, and the rest of our pages on this theme are in the on-chain and cycles topic hub.

Who defined it, and how vendors describe it

Attribution varies by source. Glassnode's metric guide lists the MVRV Z-score's creator as Awe_andWonder, with a creation date of October 2018, and identifies that name as a Medium author. Some explainers instead credit Murad Mahmudov and David Puell. This page could not retrieve an original article that settles it, so it names the vendor guide's attribution and treats the question as unresolved.

The same Glassnode guide gives the formula the way this page does, the gap between market cap and realized cap over the standard deviation of market cap, without saying which window the standard deviation uses. It states that a market value significantly above realized value "has historically indicated a market top (red zone)" and the opposite "has indicated market bottoms (green zone)," and it links an outside article whose title claims the metric "predicts market tops with 90%+ accuracy." The guide gives no numeric thresholds in its text and no caveats. This page tests none of those claims and endorses none of them. What the data above do show is that a zone defined by a fixed number would have behaved differently in each year, and that the window choice, which the guide leaves open, moves the number by more than a point.

Reproducing the numbers, and the early-years trap

Anyone can rebuild the tables from public data without a key. Download daily CapMrktCurUSD and CapMVRVCur for bitcoin from the Coin Metrics community API starting July 18, 2010. Realized cap is market cap divided by MVRV. In a spreadsheet, the expanding standard deviation in row n is STDEV.P over market cap from the first row through row n, and the Z-score is the gap divided by that. Check any row in the tables above before trusting your own build; the October 21, 2026 value should be near 0.98.

Start your counts in 2012. In 2010 and 2011 bitcoin's market cap was a few hundred thousand to a few hundred million dollars, and an expanding standard deviation computed on the first few weeks of data is close to zero. The result is huge Z-scores that describe the arithmetic of a tiny denominator and say nothing about the market, such as 7.68 on June 8, 2011. The Z-scores quoted here are our computation from the free series, using the population standard deviation. A sample standard deviation, a log transform or a rolling window gives different numbers from the same formula, and the free feed does not publish realized cap directly, so ours is recovered from market cap and the published ratio.

For another headline number whose meaning shifts with how it is built, read how bitcoin dominance is calculated, where the choice of which coins to include changes the answer in the same way the choice of standard deviation does here.

Frequently Asked Questions

What is the MVRV Z-score?

It is bitcoin's market cap minus its realized cap, divided by the standard deviation of market cap. Realized cap values each coin at the price when it last moved. The score shows how far price sits above or below the network's estimated cost basis, relative to historical swings. It is descriptive and not a forecast.

What is a good MVRV Z-score?

There is no good or bad reading. It describes the gap between price and estimated average cost basis. Historically the highest readings came in 2013, 2017 and early 2021 and negative readings in 2015, 2018 and 2022, but those are observations about the past and not a rule for the future.

Why do different charts show different MVRV Z-scores?

They use different windows for the standard deviation, different price feeds and different handling of dormant coins. On the same day, an expanding-window and a rolling four-year version can differ by about one point or more, as the table above shows for 2017 and 2021.

Who created the MVRV Z-score?

Glassnode's metric guide lists Awe_andWonder, October 2018. Some explainers credit Murad Mahmudov and David Puell. This page could not retrieve an original article that settles the attribution.