Bitcoin Realized Price: What Moves It and Where It Fails

Bitcoin's realized price is the average of what each coin was worth, in dollars, on the day it last moved on the blockchain: realized cap divided by coins in existence. On October 21, 2026 that figure was $53,944 per bitcoin, against a closing price of $84,176 the same day, which put the market 56 percent above it. The number describes an on-chain average. It is not what holders paid, and it is not a signal.

One coin, then 100,000: how the average actually shifts

Take a single bitcoin that last moved on March 16, 2020, when the Coin Metrics daily close was $5,012.93. It has sat still since. Today it is spent, even just to another address its owner controls. At the October 21, 2026 close of $84,176.08 the metric now carries that coin at $84,176.08 instead of $5,012.93, so realized cap rises by $79,163. Supply is 20,101,595 BTC, so realized price rises by $79,163 divided by 20,101,595, or $0.0039. Less than half a cent. The average hardly notices one coin.

The general rule is one line, and it is the whole mechanism:

Change in realized price = coins moved x (today's price minus the price when they last moved) / total supply

Everything else on this page is that line applied at larger sizes, to different coins, on different dates. Coin sizes and prior prices below are hypothetical unless noted; the dates and the October 21, 2026 inputs (price $84,176.08, supply 20,101,594.87 BTC, realized price $53,944.08, MVRV 1.5604) are from the Coin Metrics series.

Three things follow from the table.

First, direction depends only on whether a coin is last-moved below or above today's price. Coins that sat through the 2020 low and move now push the average up. Coins that last moved near the October 2025 high and move now pull it down, because the metric re-values them at a lower price. Glassnode's guide for the metric says the same thing in words: a coin moved at the same price at a later date "will have no impact on realized price."

Second, scale matters more than anything else. A move of 100,000 coins is about half a percent of supply and shifts realized price by a few hundred dollars. To move the figure by $1,000 in one go from a $20,000 base, it would take roughly 313,000 coins to be re-priced at once, nearly 1.6 percent of everything issued.

Third, nothing in the table involves a sale. The metric sees a coin leave one output and appear in another. Whether that was a trade, a transfer to a new wallet, or a fund moving coins into custody is invisible to it. That is why the rest of this page looks at who moves coins, and why.

What the free series gives you, and what it does not

Realized cap is the sum, over every coin, of its quantity times the price on the day it last moved. Coin Metrics' metric catalog defines its realized cap series, CapRealUSD, as "the sum USD value based on the USD closing price on the day that a native unit last moved (i.e., last transacted) for all native units." Realized price is that sum divided by supply, in dollars per bitcoin.

The catch for anyone rebuilding this from free data: the Coin Metrics community API does not serve CapRealUSD for bitcoin. A request returns a "forbidden" error, and the community catalog of 31 daily bitcoin metrics does not list it. Two metrics that are free carry the same information. CapMVRVCur is "the ratio of the sum USD value of the current supply to the sum realized USD value of the current supply," and PriceUSD is a fixed daily close as of 00:00 UTC. Dividing price by that ratio gives realized price exactly, because market cap over realized cap equals price over realized price. Every realized-price figure on this page is PriceUSD divided by CapMVRVCur, in dollars, from daily data downloaded October 23, 2026. The last complete observation is October 21, 2026.

The ratio in the second-to-last row is a gap measured against realized price. The last row is the same pair measured against price. They are different percentages of the same two numbers, which is why a "56 percent above" and a "36 percent below" can both be correct on the same day. This site's MVRV Z-score page covers the ratio form and its standardized version. This page stays with the dollar figure.

One more definition matters for what follows. Glassnode's guide credits the realized cap metric to Antoine Le Calvez in September 2018 and states that it "reduces the impact of lost and long dormant coins." We treat both statements as the vendor's description, since Coin Metrics' own 2018 write-up could not be retrieved for this page.

How slowly it moves compared with price

Realized price is an average over roughly 20 million coins, and only the coins that move on a given day are re-valued. So it is much smoother than price. Over the 3,581 daily observations from January 1, 2017 to October 21, 2026, the median absolute one-day change in realized price was 0.072 percent, against 1.47 percent for the price itself. The 99th percentile of one-day realized-price changes was 1.79 percent.

The smoothness has limits, and the exceptions are informative. These were the largest one-day moves in the series since 2017:

We have not matched these days to news events and do not claim any cause. The size of the numbers is the point. On March 30, 2023 the price moved 1.3 percent, an ordinary day, while realized cap fell $5.90 billion. On November 18, 2022 price was flat and realized cap fell $4.30 billion. Those are days when a large pile of coins last moved at higher prices changed hands or addresses and was re-valued lower. The metric cannot say whether those were sales, exchange reshuffles or one holder tidying wallets.

Over the longer run the smoothness shows up as a different drawdown. Since 2012, the deepest peak-to-trough fall in realized price was 23.25 percent, from $360 on August 12, 2014 to $277 on October 14, 2015. Price itself fell 84.5 percent from its close on December 4, 2013 to its close on January 14, 2015. Those windows are different on purpose, because each series peaks and bottoms on its own dates, so the two percentages are not a like-for-like pair. The comparison is only that the average cost proxy never fell by even a quarter.

It can also move against price. Between December 31, 2024 and December 31, 2025, price closed 6.3 percent lower ($93,390 to $87,517) while realized price rose 37.0 percent, from about $40,946 to $56,103, consistent with coins that last moved at lower prices being spent and re-valued at the year's higher prices. In 2026 through October 21 realized price is down 3.8 percent, from $56,103 to $53,944, and its high of $56,463 came on November 26, 2025, which leaves it 4.5 percent below that peak today. For how large price drawdowns have looked by comparison, see our bitcoin volatility analysis.

Every run of closes below realized price, computed

The comparison most commonly made with realized price is whether the daily close sits under it. The table lists every such run in the Coin Metrics series, with the rule written out so anyone can reproduce it.

Rule: a day counts as below when CapMVRVCur is under 1 on the daily close, which is the same as PriceUSD being under realized price. Runs separated by 30 days or fewer of closes above are merged into one run. That merge choice changes how many rows there are (with no merging there are 27 runs, with a 14-day gap there are nine), so the rule matters and is stated. The series starts July 18, 2010. Data before 2012 are thin, with bitcoin priced in single dollars, so the 2011 and early 2012 rows should be read with that in mind.

The seven runs hold 775 closes below realized price in total. The last was January 12, 2023. As of October 21, 2026 that is 1,378 days ago, more than three years and nine months.

Four observations, all about the past, none about what comes next.

Realized price itself fell during every run. The last column is negative in all seven rows, from 1.7 percent to 23.3 percent. When price falls below the average cost estimate, the fall does not wait for the estimate to catch up; coins are being re-valued at the lower prices as they move, which drags the average down. A reader who treats realized price as a fixed floor is looking at a line that did not stay put in any of these episodes.

Runs vary enormously in length and depth. Seven days in March 2020, 214 days in 2022 and 2023, 310 days across 2014 and 2015. The deepest close ranged from 6.8 percent under realized price to 61.3 percent under. A single "below realized price" label hides all of it.

How a run ended differed too. The close immediately after each run's last below-day had a ratio of 1.131 (January 4, 2012), 1.004, 1.006, 1.004, 1.129 (April 2, 2019), 1.116 (March 19, 2020) and 1.007 (January 13, 2023). Four runs ended with price creeping across realized price, three with a single-day jump of more than 11 percent. We draw no timing claim from this and no pattern: seven events, two eras, very different markets.

The 2026 drawdown did not produce a run. Price fell 53.1 percent from $124,824 on October 6, 2025 to $58,525 on June 30, 2026. The lowest MVRV in 2026 was 1.103 on that date, which put price 10.3 percent above realized price of $53,070. A common claim, repeated in a NewsBTC piece on TradingView that cites CryptoQuant, is that "past bear markets generally saw Bitcoin spend time at or below" realized price. The table agrees for four of the major declines (2011, 2014 to 2015, 2018 to 2019, 2022 to 2023). It also shows a drawdown of more than half in 2025 to 2026 with no close below. Whether that says anything about what happens next is not something a seven-row table can settle. Our halving-cycle history puts these episodes next to supply-issuance dates.

When a fund moves coins: the iShares trust's own cost basis

An exchange-traded fund that holds bitcoin is a large holder whose coins move in blocks. iShares Bitcoin Trust ETF reports its holdings, their cost and their fair value every quarter, which gives a rare dated look at one big holder's cost basis. These are accounting figures, so the bases need stating. The trust's "cost" is the fair value of each bitcoin when it came into the trust, carried at average cost under U.S. GAAP. Realized price is a different thing: each coin's on-chain last-moved price, averaged across all supply. The two are comparable in spirit, not in definition.

Source: iShares Bitcoin Trust ETF quarterly report on Form 10-Q for the quarter ended June 30, 2026, which we read through a web archive copy because the SEC's site rejected automated requests; per-coin figures and shares of supply are our division.

On both dates the fund's average cost per coin sat well above the network's average realized price: 49.6 percent above at the end of 2025 and 56.5 percent above on June 30, 2026. That does not mean the fund is "more underwater" than the average coin in any precise sense, because the two measures treat dates and price series differently, and the fund's fair value per coin ($59,101.49) is itself marked with the CME CF Bitcoin Reference Rate, New York Variant, while the Coin Metrics close was $58,525.08 the same day, 0.98 percent lower. It does show the scale of the effect. One fund holds close to 4 percent of all bitcoin ever issued, and its cost per coin is far above the network figure.

The quarter ended June 30, 2026 shows how such a holder's activity could flow into the average. The trust's reported activity for those three months:

The 10-Q says baskets of shares can be redeemed "in exchange for an amount of bitcoin" or for cash proceeds from selling it, and that the trust's activity includes $3.29 billion of bitcoin purchased in-kind and $3.85 billion of fair value paid out in-kind in the quarter. In-kind means coins actually changing addresses on the blockchain, which is the event a realized-cap series observes. The trust's own books record a $1.60 billion realized loss on the quarter's redemptions, the gap between $8.83 billion of cost and $7.24 billion of value.

Now a labelled hypothetical, not a measurement. Suppose Coin Metrics carried the redeemed coins at the trust's cost, and re-valued them at the redemption-day price when they left. The effect on network realized cap would be the $1.60 billion loss, and on realized price, $1.60 billion divided by 20.05 million coins, about -$79.60. Over the same quarter the series fell $1,091, from $54,161.90 on March 31 to $53,070.46 on June 30. That puts the hypothetical at roughly seven percent of the move, with the rest coming from every other coin that moved. The real figure is unknown, because Coin Metrics does not say which addresses belong to which holder here, and it could be larger or smaller if the custodian's internal wallet moves re-valued these coins at other dates. The point is the mechanism: a fund's cost basis and an on-chain realized price are connected, and a single holder's quarter can move the average by tens of dollars.

For how those baskets work and what a share does and does not hold, see our guide to how spot bitcoin ETFs work, and for fund-by-fund details, spot bitcoin ETFs compared.

Exchange wallets: the other big mover of coins

Coin Metrics also publishes how many bitcoin sit in wallets it attributes to exchanges. Its catalog describes SplyExNtv as "the sum of all native units held in hot or cold exchange wallets that interval," and adds that values "reflect activity of addresses currently known to belong to the entity." The second clause is the caveat: the series is built from the addresses identified so far, and as labelling improves, history is restated. The latest values carry a "flash" status, meaning preliminary.

Because those labels are retrospective, the 17.2 percent peak and the 13.3 percent latest reading should be read as "of the addresses known today," not as a fixed historical fact.

What matters for realized price is the traffic through those wallets, since every deposit and withdrawal is a coin changing address. From January 1 to October 21, 2026, 294 days, Coin Metrics' daily flows show about 6.30 million BTC sent into exchange wallets and about 6.64 million BTC withdrawn, excluding exchange-to-exchange transfers. Those totals are 2.35 and 2.48 times the 2.675 million BTC sitting in exchange wallets on October 21. The same coin can be counted more than once as it moves in and out, and the values are preliminary, so this is a measure of churn rather than of unique coins. It still shows that the exchange stock turns over several times a year, and each turn can re-value coins.

Custody practice changes what these transfers mean. A trade between two customers of the same exchange is usually a database entry; no coin changes address. When the exchange later consolidates or rotates its wallets, a large block of old coins can be re-valued in one go without anyone having bought or sold, which is the kind of day the daily-change table above shows. On-chain measures also cannot see inside a custodian: they register the wallet-level move, not the customers behind it. Our exchange reserve risk checklist covers what reserve disclosures do and do not show, and the block explorer guide shows how to see a single move yourself.

Lost and dormant coins: how much the headline number depends on them

The headline realized price divides realized cap by every coin issued, including coins that may never move again. Glassnode's guide makes the point with a case: bitcoin last spent in 2009, before BTC had a price, and now lost carries an economic value of zero in realized cap. Such coins sit in the denominator, at or near zero in the numerator, and pull realized price down. They would raise it sharply if they were removed.

No source we could fetch gives a reliable count of lost coins, so the table below is a sensitivity grid with hypothetical inputs, not an estimate. It asks: if some number of coins carried at some average last-moved value were permanently unspendable, what would realized price of the remaining coins be? Inputs are the October 21, 2026 realized cap of $1,084.36 billion and supply of 20,101,595 BTC; the baseline is $53,944.

Two points. The count of coins matters; their assumed value barely does, since low-priced coins contribute little to realized cap whatever the exact figure. And because the supply figure includes every coin ever issued, any chart that excludes dormant coins will print a higher realized price than the one here. Neither version is wrong. They answer different questions: the all-coins number is a cost-basis average over the whole ledger, the active-supply number is an estimate over coins likely to move.

Short-term and long-term holders: what the cohort versions change

Several dashboards publish realized price for subsets of coins. Glassnode's definitions are the common reference. Its short-term holder (STH) metrics use a 155-day threshold, and its short-term holder MVRV "is MVRV that takes into account only UTXOs younger than 155 days." The guide on holder supply calls 155 days the point after which dormant coins become "increasingly unlikely to be spent," describes the split as based on statistical patterns in past spending, and applies a smoothing curve with a midpoint at 155 days and a transition width of 10 days. It also says its holder metrics are entity-adjusted, meaning on-chain entities are identified by clustering addresses, and that supply held on exchanges is excluded.

Applied to realized price, the cohort construction has a plain consequence. The short-term figure is built from coins that moved within about the last five months, so it is made of recent closes and tracks price. The long-term figure is made of older closes and moves slowly. Each is a different average from the all-coins number above, and none of the three is a "true" cost basis. The 155-day line is a modelling choice with a statistical rationale, not a property of the blockchain.

We do not quote cohort values here. The free Coin Metrics community feed does not include them, and we did not retrieve them from a primary source. The point for a reader is narrower: when a chart says "realized price" without a cohort label, it means all coins; when it says short-term or long-term, check the age rule, whether coins are grouped by entity, and whether exchange balances are included.

Why two sites print different realized prices on the same day

Put the pieces together and the reasons for disagreement are concrete.

The practical rule is to cite a realized price with its source, date and definition together, and to compare numbers only from the same source and date. The standardized version of the ratio is covered on the MVRV Z-score page linked earlier, and the on-chain and cycles hub lists the other pages in this series.

For a different kind of record built from the same daily price data, every Bitcoin bear market measured lists the peak, trough and recovery time of each drawdown deeper than 50 percent, which is the depth side of what this page looked at from the cost-basis side.

Frequently Asked Questions

What is bitcoin's realized price today?

On the latest Coin Metrics daily observation, October 21, 2026, it was $53,944, computed as price divided by the MVRV ratio, against a closing price of $84,176. Other providers print different figures for the same day because of price source, entity grouping and which coins count, so compare only within one source and date.

Is realized price the average price investors paid for bitcoin?

No. It values each coin at the price on the day it last moved on-chain, which may be a purchase, a self-transfer, an exchange reshuffle or a custody move. A coin moved between your own wallets is re-valued at the day's price even though you did not trade. It estimates an aggregate on-chain cost basis, and it can differ widely from what any individual paid.

What does it mean when bitcoin trades below realized price?

It means the daily close was under the average last-moved price of all coins, equivalent to an MVRV ratio under 1. In the Coin Metrics series that happened in seven runs since 2011, the last one ending January 12, 2023. The deepest was a close 61.3 percent under in 2011 and 43.6 percent under in January 2015. This page describes those runs and draws no forecast from them.

How is realized price different from MVRV?

MVRV is price divided by realized price, so it is the same information expressed as a ratio instead of a dollar level. On October 21, 2026 the dollar level was $53,944 and the ratio was 1.56. The ratio is easier to compare across years, while the dollar figure is easier to put beside a price chart.

Where can I get realized price for free?

Coin Metrics' community API serves daily bitcoin PriceUSD and CapMVRVCur without a key, and dividing the first by the second gives realized price. Its CapRealUSD series returned a "forbidden" error on the free tier when we tested on October 23, 2026. Cohort versions such as the short-term holder realized price are not in the free feed.

Do ETF purchases and redemptions change realized price?

They can, but only through coins that actually change address on-chain. In-kind creations and redemptions move coins and re-value them at that day's price, while cash transactions in which the fund buys or sells through a counterparty move coins too. The effect depends on the price at which the coins last moved. In the iShares trust's second quarter of 2026, the redemptions' effect on realized price would have been roughly -$80 if on-chain values matched the trust's cost basis, which is a hypothetical rather than a measurement.

Sources

CryptosEyes publishes general educational research, not investment, legal, or tax advice. Realized price is a descriptive on-chain statistic computed from third-party data that can be revised, and it does not predict future prices or indicate when to buy or sell any asset.