Total value locked is the dollar value of crypto assets users have deposited into a protocol's smart contracts, counted by multiplying each token balance by its current price. It is the headline adoption number in DeFi and it is also the least standardised one: two reputable providers reading the same protocol on the same day routinely publish different figures, because each one decides for itself what counts as locked, what counts twice, and what should be excluded.
TVL is a measurement, not a fact, and the measurer chooses the rules. Before you compare two TVL figures, check whether they came from the same provider, on the same day, with the same exclusions applied.
Scroll to see the full diagram
What a TVL figure is actually counting
DefiLlama, the largest aggregator of the metric, defines TVL as the sum of the value of crypto assets that users have deposited to a protocol in order to earn rewards or interest.1 CoinMarketCap's definition is close to the same shape: the total value of digital assets deposited or locked in a protocol, expressed in US dollars, calculated as the number of tokens locked multiplied by the current price per token.
Two consequences fall straight out of that formula. The first is that TVL moves when prices move, with no user doing anything at all. A protocol whose deposits are denominated in a volatile asset can lose a quarter of its TVL overnight while every depositor stays put. The second is that TVL is a stock of deposits, not a flow of income. It tells you what is sitting in the contracts. It tells you nothing about what the protocol earns for holding it, and those two numbers can move in opposite directions for a long time.
How the figure gets produced matters as much as how it is defined. DefiLlama's data comes from an open-source adapter written per project. Most adapters make onchain calls to return the balance, but some read subgraphs or APIs instead, and the adapters are public so anyone can check which approach a given protocol uses. Most TVL series update hourly.1
Same protocol, same day, two different numbers
On 3 August 2026 we read Ethena from both major providers within minutes of each other. CoinGecko's API returned a total value locked of $3,855,382,456.2 CoinMarketCap's coin page for the same protocol showed a TVL of $4.09 billion.3 That is roughly $235 million of difference, about 6%, on a number both sites present as a plain fact with no methodology note attached to it.
The supply figures on that same pair of reads agreed exactly, at 9.56 billion circulating out of 15 billion, and the market caps were within half a percent of each other.2 So this is not a stale-data problem or a price-feed problem. It is a definitional one. The two providers are counting a slightly different set of contracts, or applying different exclusions to the same set.
The design consequence is unglamorous and important. If your investor deck cites a TVL you pulled from one site and your competitor's deck cites theirs from another, the comparison is not a comparison. Pick one provider, name it, date it, and use the same one every quarter.
One deposit, counted four times
The structural problem with TVL is that DeFi composes. A user deposits ETH into a liquid staking protocol and receives a liquid staking token. That token gets restaked. The restaking receipt goes into a liquidity pool. At each step a protocol can legitimately claim the deposit as its own TVL, and the same original asset is now counted several times across the aggregate.
DefiLlama addresses this with an explicit exclusion rule: it does not accept assets that a protocol generates and that are then locked into other protocols, because that is the later protocol's TVL rather than the originating project's.4 It also declines to count native token staking, giving ATOM staking to secure the Cosmos Hub as its worked example, and it excludes assets that are not on the blockchain at all, naming dollars held in Tether's bank account as something it does not treat as TVL.4
Those rules are a considered position, not an obvious one, and a different aggregator can take a different one in good faith. That is how you get a 6% gap on Ethena and much wider gaps elsewhere.
Scroll to see the full diagram
What the aggregator deliberately filters out
DefiLlama splits TVL into filterable categories rather than publishing one blended number. Staking covers the platform's own token. Pool2 covers staked LP tokens where one side of the pair is the platform's own governance token. Borrows covers deposits that were borrowed from the platform. Offers covers funds approved for spending but never actually deposited into the contracts.4
The most instructive category is Vesting, and DefiLlama's own explanation of why it exists is worth reading directly: tokens that are not circulating or not issued yet, where a token with a $10 million market cap and a $1 billion FDV could have $500 million locked, and in those cases it makes no sense for the TVL from that token to be $500 million when its market cap is only $10 million.4 That is an aggregator explicitly refusing to let an FDV-priced locked allocation inflate an adoption metric.
DefiLlama has also been stripping out what it calls unproductive or artificial liquidity: assets deposited only to earn rewards or boost metrics without taking real market risk, positions that come from a small number of large wallets, and TVL inflated through circular or non-backed assets.4 Read that as an admission from the largest aggregator in the category that the headline number has been gameable and was being gamed.
The BIS finding: some of this cannot be checked at all
The Bank for International Settlements published a working paper on this in May 2025, and its framing is blunt. TVL is the main metric used to assess the economic significance of DeFi projects, its calculation is not standardised, in some cases it relies on self-reported off-chain data, and inconsistent methodologies across aggregators can lead to large discrepancies in reported figures.5
The paper analysed 939 DeFi protocols on Ethereum. It found that 10.5% rely on external off-chain data sources, which makes their TVL figures difficult to verify independently, while 78.6% use standard balance queries that anyone can reproduce.5 The authors propose a verifiable total value locked measure built only from onchain data and standardised balance queries, designed specifically to avoid double-counting.5
One in ten protocols publishing a headline adoption number that no outsider can reproduce is not a rounding error. When a central bank research group builds a replacement metric because the existing one is not reproducible, treat that as the ceiling on how much weight the original can carry.
What chasing TVL does to a protocol
TVL responds to incentives faster than it responds to product quality, which makes it the easiest metric in DeFi to buy. Emit enough of your own token and deposits arrive within hours. Stop emitting and a large share of them leave within days, because they were never there for the product.
This is the pattern we spend the most time unwinding. A protocol sets a TVL target, funds it out of the token allocation, hits the number, uses the number to raise, and then discovers that its cost of deposits is permanent while its revenue per deposited dollar is close to zero. The token paid for the metric and the business never got built underneath it.
The diagnostic we use is simple and it is not TVL. What does the protocol earn per dollar of deposits, and what share of deposits stays when emissions are cut by half? A protocol that can answer both with real figures has a business. One that can only quote a TVL headline has a marketing number and a liability. The token is infrastructure for the business, and TVL only means something when there is a business underneath it collecting fees.
Quoting a TVL figure without misleading anyone
Four habits, and they cost nothing. Name the provider. Attach a date, because TVL is a live series and not a constant. State whether the figure is gross or net of double-counting. And say which categories are included, because Pool2 and Borrows can be a large share of the total on some protocols.
As an illustration of why the date matters more than people expect: DefiLlama's own global DeFi TVL series read $139,582,088,952 for 3 August 2026.6 Independent research working from the same source, adjusted to exclude double-counting, put the aggregate at over $116 billion as of May 2025. Those are not contradictory figures, they are the same metric at two points on a moving series, and a figure quoted without a date implies a stability the number does not have.
For founders the takeaway is narrower than it looks. TVL is a useful operating metric for your own protocol tracked against your own consistent definition over time. It is a poor comparison metric across protocols, a poor valuation input on its own, and a bad target to manage toward.
Common questions
What does total value locked mean in crypto?
Total value locked is the dollar value of crypto assets users have deposited into a protocol's smart contracts, calculated as tokens locked multiplied by current price. DefiLlama defines it as the sum of assets deposited by users to earn rewards or interest.1 It measures deposits sitting in contracts, not revenue, users or profit, and it rises and falls with token prices even when no user deposits or withdraws anything.
Why do CoinGecko and CoinMarketCap show different TVL?
Because each provider decides for itself what counts. Reading Ethena on 3 August 2026, CoinGecko's API returned $3,855,382,456 while CoinMarketCap's page showed $4.09 billion, roughly 6% apart, with supply and market cap figures that agreed closely.2 The difference comes from which contracts are included and which exclusions are applied, not from stale data. Name your provider whenever you quote the number.
Is TVL double counted?
It can be, and aggregators handle it differently. When a deposit is wrapped into a liquid staking token, then restaked, then supplied to a pool, each protocol can claim it. DefiLlama excludes assets a protocol generates that are locked into other protocols, treating them as the later protocol's TVL.4 A BIS working paper proposed a verifiable TVL measure built specifically to avoid double-counting.5
Is high TVL a good sign for a protocol?
Only if it stays when the incentives stop. TVL responds to token emissions faster than to product quality, so a large figure can reflect rented deposits rather than demand. DefiLlama has been removing what it calls unproductive or artificial liquidity, including positions from a small number of large wallets that take no real market risk.4 Ask what the protocol earns per deposited dollar instead.
How is TVL calculated?
Multiply each token balance held in the protocol's contracts by its current market price, then sum them. DefiLlama runs an open-source adapter per project; most make onchain calls to read balances, though some use subgraphs or APIs, and most series update hourly.1 A BIS study of 939 Ethereum protocols found 10.5% rely on off-chain data sources that outsiders cannot independently verify.5
See Tokenomics Audit for how this applies in practice.
Sources
- DefiLlama Docs, Frequently Asked Questions
DefiLlama, 2026
TVL definition, open-source adapter methodology, onchain calls versus subgraphs and APIs, and hourly update frequency. - CoinGecko public API v3, coin record for Ethena
CoinGecko, 2026
TVL, supply and market cap fields read on 2026-08-03. Point-in-time values that change continuously. - Ethena (ENA) price and metrics page
CoinMarketCap, 2026
TVL, market cap to TVL ratio and supply figures read on 2026-08-03. Point-in-time values that change continuously. - DefiLlama Docs, What to include as TVL
DefiLlama, 2026
Filterable categories, the vesting-token example, exclusion of cross-protocol receipts and native token staking, and the unproductive-liquidity cleanup. - Towards verifiability of total value locked (TVL) in decentralized finance, BIS Working Paper No 1268
Bank for International Settlements, 2025
939 Ethereum protocols analysed; 10.5% rely on off-chain data sources; proposes a verifiable TVL measure designed to avoid double-counting. - DefiLlama public API, global TVL chart series
DefiLlama, 2026
Aggregate DeFi TVL of $139,582,088,952 for 2026-08-03. A live series, not a constant.
Last reviewed 2026-08
Know the terms but not sure how they apply to your project? That is what an engagement is for. We design, document, and stress-test the whole token economy inside the Tokenomics Data Room.
100+ projects advised. Complete tokenomics in 4 to 6 weeks.