Key Takeaways
- What it is: A sum over every unit of supply of that unit's quantity multiplied by the market price on the day it last moved on-chain.
- What it is not: A market capitalization. Market cap applies today's price to all supply; realized cap applies a different price to every unit.
- It is path dependent: Two networks with identical supply and identical price can carry very different realized caps, and the figure cannot be reconstructed from a ledger snapshot.
- Lost coins anchor it: Supply that can never move again keeps its last-move price permanently and quietly holds the total down for the life of the chain.
- The "money invested" description is loose: Every spend has a seller taking money out as a buyer puts money in, and a self-transfer reprices supply with no counterparty at all.
Who This Guide Is For
Read this page if you have seen realized capitalization described as the amount of money invested in a network and want to know how far that description can be pushed. The construction answers the question by itself, and the four decisions behind it are what make two providers publish different numbers from the same chain.
This is educational material about a measurement technique, not individualized financial advice. Every figure below is invented for illustration.
What Does Realized Capitalization Measure?
Realized capitalization measures the aggregate cost basis of a network's supply, with each unit's basis set to the market price at the moment it last changed position on the ledger. It is a stock figure built out of a history of flows.
Market capitalization asks one question of the ledger, how much supply exists, then multiplies by one price. Realized capitalization asks every single unit when it last moved and what the price was that day. A network gets as many prices as there are distinct last-move dates.
Plain-language definition
Take every coin in existence. Look up the day it last moved and the price that day. Multiply. Add the products together. That total is the realized capitalization.
Technical definition
Partition current supply into lots, a lot being a quantity q sharing one last-movement timestamp t. With P a reference price series, realized cap is the sum of q × P(t) across every lot. On a UTXO chain the lot is a native object, an unspent output whose creating block fixes t without ambiguity. On an account chain no such object exists, and the lot must be synthesized by an accounting rule the provider chooses.
Why the sum is path dependent
The set of last-move timestamps records everything that has ever happened on the chain. Two hypothetical networks could hold a million units each, trade at the same price today, and report realized caps an order of magnitude apart, purely because one saw its supply change hands repeatedly at high prices and the other did not. No snapshot of balances answers the question, which is why realized cap cannot be verified from a block explorer balance page.
How Is Realized Capitalization Constructed?
Construction is four decisions, each baked permanently into every value the series will ever publish, because the sum accumulates rather than resetting.
1. What counts as a lot
UTXO chains hand this over for free. Each unspent output records the block that created it, so its price date is a lookup, not an estimate. The Bitcoin Developer Guide: Transactions describes the output structure this depends on. Account chains store a balance as a single integer: an address holding 10 units that sends 3 has not identified which 3 left. Providers impose first-in-first-out, last-in-first-out, or a weighted-average rule, and the three produce different realized caps from identical chain data. The UTXO versus account model guide covers this asymmetry.
2. Which price counts as the last-move price
A block has a timestamp measured in seconds, but most price references are daily. Daily close, daily volume-weighted average, and block-time estimate give three different answers, and the gap widens on volatile days. It does not average out, because moves cluster on volatile days: the convention matters most exactly when the most coins are being repriced.
3. Whether a self-transfer reprices
The ledger cannot distinguish an owner moving coins between two of their own wallets from a genuine change of ownership. A naive implementation reprices both; a provider running entity clustering suppresses the self-transfer and leaves the old basis intact. This choice is the largest source of divergence between two published series, and it matters most during custodial reorganizations, when a large quantity of old supply is repriced by an operation that changed nothing about who owns what.
4. What price newly issued supply enters at
Coins created by block rewards were not purchased; they were received. Pricing them at the market price of their creating block is the usual convention, but that treats as basis something no one paid. Early supply from a period when no market existed has no price to look up at all, so the provider's starting convention sits permanently inside every later value. Reorganization handling and revision policy are covered in the on-chain data quality methodology guide.
Formula and Measurement Logic
Realized cap = sum over all lots of (lot quantity × market price on the lot's last-movement date)
Everything interesting about the metric follows from the shape of that sum. Three consequences:
It updates only on movement. A billion dollars of trading on an exchange order book moves no coins on-chain and changes realized cap by exactly zero. It responds to withdrawals and deposits, not to price discovery, so while trading is internal to custodians realized cap can sit almost still as price moves a long way.
It rises only when coins move up. A spend removes q × old price from the sum and adds q × today's price. The net change is q × (today's price - old price), which is positive only when the coin moves at a higher price than it last moved at. That is the net realized profit or loss on the spend, so the day-over-day change in realized cap and the realized profit and loss series are two views of one number, differing only by newly issued supply.
It has no normalization. The raw sum grows with the chain and with the price level, so it is not comparable across networks or eras undivided. Dividing by supply gives the realized price; dividing market cap by it gives the MVRV ratio.
| Decision | Options | What it changes |
|---|---|---|
| Lot definition | Native output, or synthesized by FIFO, LIFO, or average cost | Nothing on UTXO chains; everything on account chains |
| Price convention | Daily close, daily volume-weighted average, block-time estimate | Diverges most on volatile days, which are the days with the most repricing |
| Self-transfers | Repriced, or suppressed by entity clustering | The largest single source of divergence between providers |
| Issuance pricing | Market price at the creating block, or a fixed starting convention | Permanently embedded in every later value of the series |
| Lost or unspendable supply | Almost always included, because it cannot be identified reliably | Anchors part of the sum at a price that can never update |
| Supply universe | Total issued, circulating, or entity-adjusted | Changes realized price, and must match the denominator used with it |
How Should Realized Capitalization Be Interpreted?
Read a change in realized capitalization as a statement about repricing: supply moved, above or below where it sat before. Read the level as accumulated basis, and a flat series during a large price move as evidence that almost nothing is moving.
Why "money actually invested" is a loose description
The sum has the right units and roughly the right shape. It is loose for five reasons, and they do not point the same way.
- Every spend has two sides. When a coin is repriced upward, a buyer paid and a seller was paid. The sum records only the new basis on the coin, not the cash that left with the seller.
- Self-transfers reprice without a counterparty. A wallet consolidating its own holdings changes realized cap by the full old-to-new price difference, and no money was invested by anyone.
- Issued supply was received, not bought. Block rewards enter at their creation price, which nobody paid.
- Off-chain trading is invisible. Coins traded thousands of times on an exchange keep whatever basis they had at deposit until withdrawal.
- Lost coins never update. Unspendable supply holds its last-move price permanently, so part of the sum reflects a market that no longer exists.
The lost-coin anchor
The anchoring effect is silent. Unspendable supply contributes a fixed dollar amount to a sum that otherwise grows, so its share of realized cap shrinks over time even as its share of counted supply stays constant. That asymmetry drags realized price down more than realized cap, which matters because realized price is what gets compared against market price.
Step-by-Step Workflow
- Establish whether the chain is UTXO-based or account-based, and on an account chain which lot convention the provider applies.
- Record the price convention: close, volume-weighted average, or block-time estimate.
- Establish whether the provider suppresses self-transfers through entity clustering. A series that does not is a different metric, not a noisier one.
- Establish the issuance pricing convention and the treatment of the chain's earliest supply.
- Before reading a jump, check whether one large old lot moved.
- Compare the day-over-day change against the published realized profit and loss series. They should agree once issuance is accounted for.
- Normalize by supply or against market cap before comparing across time or chains.
- Write down what would have to be true for a repricing to reflect genuine transfer of ownership, then look for it.
Worked Hypothetical Scenario
Every figure here is invented for illustration. Take a hypothetical chain with exactly 1,000,000 units of supply, sitting in three tranches that last moved at three different prices.
| Tranche | Units | Price at last move | Contribution to realized cap |
|---|---|---|---|
| A | 600,000 | $2.00 | $1,200,000 |
| B | 300,000 | $18.00 | $5,400,000 |
| C | 100,000 | $40.00 | $4,000,000 |
| Total | 1,000,000 | $10,600,000 |
Realized cap is $10,600,000. Divided by supply, realized price is $10.60. If the market price today is $25.00, market cap is $25,000,000 and the ratio is about 2.36. No single unit on this chain has a basis of $10.60: the aggregate is a weighted average of $2.00, $18.00, and $40.00, sitting in a gap where no supply is.
Two identical transactions, opposite effects
Now suppose 50,000 units move at today's price of $25.00. The effect depends entirely on which tranche they came from, though size and execution price are identical.
| 50,000 units from tranche A | 50,000 units from tranche C | |
|---|---|---|
| Old basis removed from the sum | 50,000 × $2.00 = $100,000 | 50,000 × $40.00 = $2,000,000 |
| New basis added to the sum | 50,000 × $25.00 = $1,250,000 | 50,000 × $25.00 = $1,250,000 |
| Net change in realized cap | +$1,150,000 | -$750,000 |
| Realized cap after | $11,750,000 | $9,850,000 |
| Realized price after | $11.75 | $9.85 |
The same quantity, at the same price, on the same day, pushed the aggregate up by $1,150,000 in one case and down by $750,000 in the other. The metric is answering a question about the past, not about the transaction. It is the realized profit and loss identity in miniature: 50,000 × ($25.00 - $2.00) = $1,150,000, and 50,000 × ($25.00 - $40.00) is a loss of $750,000.
The lost-coin anchor, quantified
Suppose 120,000 of tranche A's 600,000 units are permanently unspendable. At the starting position they contribute 120,000 × $2.00 = $240,000, which is 2.26 percent of the $10,600,000 total while occupying 12 percent of counted supply. Strip them out on both sides and realized cap becomes $10,360,000 over 880,000 spendable units, a realized price of about $11.77 instead of $10.60. The dead supply understated the live supply's aggregate basis by roughly 10 percent, permanently.
What Can Make the Interpretation Wrong?
- Treating it as net capital inflow. The sum records a new basis on a coin. It does not net a buyer against a seller, and it counts issuance nobody paid for.
- Reading a custodial reorganization as market activity. One entity moving old supply between its own wallets shifts the aggregate visibly in a single day unless the provider clusters entities.
- Assuming an exchange withdrawal is a purchase. A withdrawal reprices coins at that day's price whether they were bought yesterday, bought years ago, or never bought at all.
- Comparing levels across chains. The sum is unnormalized and reflects each chain's entire trading history, issuance schedule, and fee level. Cheap chains reprice supply far more often.
- Ignoring the account-model convention. On an account chain the lot structure is an assumption, and FIFO versus average cost changes the answer without changing the data.
- Expecting it to react to price. A flat realized cap during a violent price move is not a data failure. The supply is not moving, and that is the observation.
- Reusing an old threshold. Because the sum accumulates and never resets, a level that was extreme early in a chain's life is unremarkable later, with no change in behavior.
Cross-Network and Provider Comparison
Realized capitalization transfers across networks badly, because the thing being summed is defined differently on each one.
On a UTXO chain the sum is close to mechanical. Each output carries its creation height, so the only judgment calls are price convention and self-transfer handling. On an account chain the lot has to be invented first, and invented lots interact badly with contracts: coins deposited into an automated market maker, lending pool, or bridge escrow are pooled into one balance shared by many depositors, and every interaction reprices supply that no individual holder touched.
Fee level is a second confounder. Where a transfer costs a fraction of a cent, supply moves constantly for operational reasons and realized cap tracks close behind market cap. Where transfers are expensive, supply sits still and realized cap lags far behind. The gap between the two capitalizations is therefore partly a measure of transaction cost, so comparing that gap across two chains with different fee markets compares fee markets.
Wrapped and bridged representations add a third problem: the same economic unit exists twice, and summing both the escrowed original and the wrapped copy double counts the network.
Advanced Analytical Methods
Splitting the sum by cohort
Because realized cap is a sum, it decomposes cleanly. Restrict it to lots whose last-move date falls inside an age band and you have that band's realized cap; restrict by entity balance band and you have a size decomposition. The parts add back to the whole exactly, which is not true of most on-chain decompositions.
Reconciling the change against realized profit and loss
The day-over-day change in realized cap should equal net realized profit and loss plus newly issued supply valued at its creation price. Computing both independently and checking that they agree is one of the few internal consistency tests available here, and a persistent discrepancy localizes the problem to filtering rather than arithmetic.
Price convention sensitivity
Recompute the series under a second price convention and plot the difference. If the two versions disagree by more than the effect you are describing, the effect is a pricing artifact.
Ex-lost-supply variants and repricing rate
Removing supply that has not moved since before a chosen date, from both the sum and the supply count, gives a live-supply realized price; the cutoff is arbitrary and must be published beside the number. Tracking what share of supply was repriced in a window separately distinguishes a change driven by a small quantity moving a long way from one driven by broad turnover.
Practical Checklist
- I know whether the lot is a native output or a synthesized accounting lot.
- I know which price convention fixes the last-move price.
- I know whether self-transfers are suppressed by entity clustering.
- I checked whether a single large old lot explains the change I am describing.
- I normalized before comparing across time or across chains.
- I described the movement as repricing rather than as capital inflow.
- I noted that unspendable supply is included and cannot be repriced.
Conclusion
Realized capitalization is a sum over supply of quantity multiplied by last-move price, so it only moves when coins move, only rises when they move upward, and holds unspendable supply at a price fixed forever. Every failure mode above restates that one sentence, which is why calling the figure money invested in the network gives the wrong intuition about all of them.
Frequently Asked Questions
Is realized capitalization the amount of money invested in a network?
Not accurately. Every repricing has a seller taking money out at the same instant a buyer puts money in, so the sum is not a net flow. It also counts block rewards at their creation price, which nobody paid, and it changes when a wallet moves coins to itself with no counterparty involved.
Why can realized capitalization fall?
Because a spend removes the coin's old basis from the sum and adds today's price. When coins move at a price below where they last moved, the replacement is smaller than the removal and the total drops. This is why sustained declines in realized cap are associated with supply changing hands at a loss rather than with supply leaving the network.
Does exchange trading change realized capitalization?
No. The metric only sees on-chain movement, so trades settled internally on an exchange ledger are invisible to it. Coins can be bought and sold thousands of times while sitting in the same custodial wallet and keep the basis they had when they were deposited. Realized cap responds to the deposit and the withdrawal, not to anything in between.
How do lost coins affect the number?
They hold their last-move price permanently, because repricing requires a spend and they can never be spent. Their contribution is a fixed dollar amount inside a sum that otherwise grows, so their share of realized cap shrinks over time even though their share of counted supply does not.
Can realized capitalization be computed on an account-based chain?
It can be approximated, but not read off directly. An account balance is a single integer with no record of which units arrived when, so the provider has to impose first-in-first-out, last-in-first-out, or average-cost accounting to create lots. Different conventions give different answers from identical chain data.
How does realized capitalization relate to realized profit and loss?
They are two views of the same arithmetic. A spend changes realized cap by quantity multiplied by the difference between today's price and the old basis, which is exactly the realized gain or loss on that spend. The day-over-day change therefore equals net realized profit and loss plus newly issued supply valued at its creation price.
How are newly issued coins valued in realized capitalization?
At the market price when they were created, since the creation is the first time those units appear on chain and it sets their initial cost basis. Issuance therefore adds to realized capitalization continuously at whatever the price happens to be, independent of any purchase. On a network with substantial ongoing issuance this contributes a persistent upward drift that is not driven by anyone buying, and separating it from the component driven by coins changing hands requires tracking issuance separately.
Why does realized capitalization behave like a smoothed market capitalization?
Because it only updates for coins that move. Market capitalization revalues every unit of supply at the current price, so it responds instantly and fully to any price change. Realized capitalization revalues only the portion that transacted, leaving the rest at its previous basis, so a price move changes it by the fraction of supply that moved during the move. That damping is the source of its usefulness as a slower reference series and also the reason it lags rather than leads.
Does realized capitalization double count coins that move repeatedly?
No, but it does overwrite their history. Each unit of supply contributes exactly once, at the price of its most recent movement, so a coin that has moved ten times contributes only the tenth price. Nothing is added twice. What is lost is every earlier basis, which means the measure records where supply was last repriced rather than the total value ever transacted. A coin moved daily contributes the same as one moved once, provided both last moved at the same price.
References
These sources should be reviewed during editorial verification. They support data structures and methods, not the hypothetical conclusion. Provider formulas, chain rules, and APIs can change. Confirm current documentation before publication.
- Glassnode: Realized Capitalization Guide: the construction of the sum over last-move prices.
- Coin Metrics: Valuation Metrics: an independent realized capitalization definition to compare conventions against.
- Bitcoin Developer Guide: Transactions: the output structure that makes the last-move date a native property on a UTXO chain.
- Ethereum.org: Technical Introduction: the account and state model that forces a synthesized lot convention.
- Coin Metrics: Network Data Glossary: ledger and supply definitions used in the denominators discussed above.
- Coin Metrics API v4: frequency and revision behavior for a cumulative series that can be restated.