TVL is the industry's default size metric and one of its least standardised. The definition changes with the incentive of whoever is publishing it.
Dario FennDeFi & Markets Lead · DeFi protocols, yield, market structure and on-chain data
Total value locked is the dollar value of assets deposited in a protocol. It differs between sources because there is no shared definition: some count the protocol's own staked governance token, some count borrowed funds redeposited elsewhere, and some count the same collateral in each protocol it passes through. It is also a price-times-quantity figure, so it falls when the market falls without any deposit leaving. Use deposit counts, revenue and net flows alongside it.
Total value locked is the number quoted whenever someone wants to describe how big a protocol is. It is a real measurement of something, and it is also the easiest headline metric in crypto to inflate without doing anything untrue.
TVL is the sum of the market value of assets deposited in a protocol's contracts. For a lending market, deposits supplied. For an exchange, liquidity in pools. For a staking protocol, assets staked.
That definition already contains the first problem: value means price times quantity. A protocol holding exactly the same tokens on two dates shows a TVL that has halved if the market has halved. Falling TVL in a downturn usually reflects prices, not withdrawals — and rising TVL in a rally is not adoption.
The fix is to look at TVL denominated in the underlying asset where it makes sense. Total ETH staked, total USDC supplied. Those numbers move only when people move them.
Composability means one deposit can legitimately appear several times.
Deposit ETH with a liquid staking protocol and receive an LST — that is TVL at the staking protocol. Supply the LST to a lending market — TVL there too. Borrow stablecoins against it and deposit them in a yield vault — TVL again. One user, one original deposit, three protocols reporting it.
None of those protocols is lying. Each genuinely holds assets it must honour. But summing them produces a total that overstates the capital in the system, and the same effect at the chain level makes a small amount of real capital look like a large ecosystem when it is being recycled through several applications.
Aggregators handle this differently. DefiLlama publishes both raw and double-counting-adjusted figures and lets you exclude liquid staking and borrowed funds explicitly, which is the correct approach and the reason its numbers frequently differ from a protocol's own dashboard.
The second distortion. A protocol whose TVL is largely its own staked governance token is measuring the market value of something it issued.
This is circular in a specific way: the token's price is partly a function of the protocol looking successful, and the protocol looks successful partly because the token's price supports its TVL. When the token falls, TVL falls faster than any real capital leaves, and the appearance of collapse accelerates the actual one.
Strip it out. TVL excluding the protocol's own token is a far better measure of whether outside capital is present.
**Borrowed funds.** In lending markets, counting both supplied and borrowed amounts counts the same capital twice within one protocol.
**Incentivised deposits.** Capital present because of a live emissions programme is rented, not resident. Check whether the programme has an end date, and what happened to TVL the last time incentives were reduced.
**Bridged and wrapped assets.** A wrapped token counted on the destination chain while the original sits in a bridge on the source chain can be counted on both.
Four metrics that are harder to inflate.
**Fees and revenue.** What users actually paid to use the protocol, and what the protocol kept. This is real economic activity and it does not double count, because the fee is paid once.
**Unique depositors and their distribution.** A hundred million dollars from twelve addresses and from twelve thousand are different businesses. Distribution also tells you how quickly the number can leave.
**Net flows.** Deposits minus withdrawals in native units, which separates capital movement from price movement entirely.
**Utilisation and depth**, for lending markets and exchanges respectively. These measure whether the deposited capital is doing anything.
When two sources disagree, the question is not which is right but what each is counting. Check whether liquid staking is included, whether borrowed funds are netted, whether the protocol's own token is in the total, and which chains are covered.
That is the same discipline we apply in the ratings methodology: a size figure is only comparable when the definition is stated, and where we use scale as a scoring input we say which measure and from where. On the market pages, price and liquidity are sourced and labelled for the same reason — an unattributed number is a claim, not data.
Price tells you what the market thinks. On-chain metrics tell you what's actually happening underneath it. Here's a starter toolkit for reading network health directly from the ledger.
A token's price chart tells you nothing about the supply mechanics working against it. Here's a framework for reading tokenomics properly before you decide what a token is actually worth holding.
Nothing is hacked. You approve a transaction whose real meaning is hidden behind a wallet prompt that says almost nothing.