Total value locked, and the double-counting problem
TVL is a dollar valuation of deposited assets, which means it moves with prices, and the same dollar can legitimately appear in several places at once.
Total value locked is the dollar value of the assets currently held in a protocol's contracts. It is a stock, not a flow: a photograph of balances at an instant, not a record of activity over a period. Two properties make it far less straightforward than the phrase suggests — it is a price multiplied by a quantity, and the same underlying dollar can be counted by more than one protocol at the same time without anyone lying.
Price does most of the moving
Total value locked is computed by taking each asset balance in the contracts and multiplying by that asset's price. If deposits do not change at all but the deposited tokens rise thirty percent, TVL rises thirty percent. Headlines describing capital "flowing in" or "fleeing" frequently describe nothing but the price of the assets already there.
The correction is to look at balances in native units next to the dollar figure, and to compare TVL change 30d with the price change of the dominant deposited asset over the same window. Where a protocol's deposits are mostly stablecoins, the dollar figure is closer to a real measure of deposits; where they are mostly the chain's own volatile token, it is closer to a price chart wearing different clothes.
The same dollar, counted several times
Double counting in this sector is usually structural rather than fraudulent. Composability means one deposit legitimately exists inside several protocols at once, and each of them can honestly report it.
| Structure | What happens | How the count inflates |
|---|---|---|
| Liquid staking | A token is staked and a receipt token is issued against it | The staking protocol counts the deposit; any protocol accepting the receipt counts it again |
| Lending loops | A user deposits, borrows against it, redeposits the proceeds | Each pass adds to TVL although only the first deposit is outside capital |
| Bridged assets | An asset is locked on one chain, a claim is minted on another | The lock counts on the origin chain and the claim counts on the destination |
| Restaking | Already-staked capital is committed to secure additional services | The same capital secures several systems and may be counted by each |
Each row is a real economic arrangement. Liquid staking, restaking, wrapped tokens and bridges all exist to let capital do more than one job. The problem is only that summing protocol-level figures into an ecosystem total then counts the same money repeatedly, and the sum grows fastest exactly when leverage and layering are increasing — which is to say when the aggregate is least representative of outside capital committed.
What TVL does not measure
TVL is not revenue, and a protocol holding large balances that nobody trades against or borrows from earns very little. It is not a measure of user funds available for withdrawal either; deposits may be lent out, committed as collateral or subject to a queue. It is emphatically not a safety rating, and the 2022 failures of lending and trading businesses made the point at scale: large balances were reported right up to the moment they could not be returned.
Nor is TVL independent of incentives. A protocol paying liquidity mining rewards can raise deposits substantially and see them leave when the program ends. Deposits chasing an emission schedule are a real balance and a poor indicator of durable usage, and separating the two requires looking at what the deposits produced rather than at their size.
Comparability across categories is the last trap. A lending market must take in collateral worth more than it lends, so over-collateralization mechanically produces a large deposit figure; a derivatives venue holding margin can support far greater notional exposure with far smaller balances; an exchange using concentrated liquidity can offer the same depth as an older design with a fraction of the deposits. Ranking those three against each other by deposit size ranks their capital requirements rather than their usefulness, and the design needing less capital to do the same work appears smaller precisely because it is more efficient.
The ratios that make it useful
TVL earns its place as a denominator. Volume to TVL asks how hard the deposited capital is working; a pool turning over many times its balance each day is doing something a dormant pool of the same size is not. DEX volume to TVL narrows that to trading, and the same ratio flags implausible readings when it is absurdly high. Market cap to TVL and FDV to TVL compare the market's valuation of the token to the capital sitting in the system — useful as a comparison across similar protocols, and misleading across different ones, since a design requiring little capital will always look different from one requiring a lot.
None of these ratios is a valuation method. They are consistency checks: a way of asking whether the flows, the balances and the market's figure describe the same business. Where fully diluted valuation is one of the inputs, the answer also depends on an unlock schedule that has not happened yet.
Counting rules differ, so read them
Providers differ on whether to include staked native tokens, whether to count both sides of a bridge, whether receipt tokens are netted out, and whether a protocol's own treasury holdings count as locked value. Those choices can change a figure by a wide margin, and a change in methodology can move a series without anything happening on-chain. Every revision to the rules used here is recorded, and the revision and backfill entries explain how historical series are treated when a definition changes.
The TVL section shows deposits with their asset composition and the ratios above, and data sources records where each series comes from. The next lesson takes on the flow measure with the worst reputation in the sector: decentralized exchange volume.