TVL Explained: Separating Deposits, Leverage and Real Demand

Total value locked, or TVL, is a valuation of assets included within a defined protocol or network boundary. It can help describe capital exposure, but it does not directly measure new money, revenue, unique users or safety. A higher dollar TVL can result from larger deposits, higher token prices, changes in accounting coverage or multiple claims on the same underlying assets. Interpreting growth requires separating these effects before explaining what caused it.

The most useful TVL analysis is a reconciliation: what assets changed, how their prices changed, which liabilities or receipt tokens are involved, and whether users stayed when incentives or market conditions changed.

Define the Boundary Before Comparing the Number

A TVL figure is incomplete without a methodology. Does it cover a protocol, a particular product, a chain or a group of contracts? Are borrowed balances included? Are governance-token staking, bridges or receipt tokens counted separately? Are assets valued at a market quote or another reference price?

DefiLlama's methodology describes exclusions and treatment choices, including restrictions on some categories of staking and double counting. These are definitions for its dataset, not universal economic laws. Another dataset can use a different boundary and still calculate its chosen measure consistently.

For a reproducible comparison, record the provider, product scope, chain coverage, relevant toggles, observation time and pricing source. If a dashboard begins tracking an additional deployment, reported growth may partly reflect improved coverage. Comparing the old narrow boundary with the new broad one and calling the difference customer inflows is misleading.

The same issue appears when products are grouped. A parent protocol can include several versions or chains. Summing the parent and its children counts activity twice. A clean chart needs a list of included entities, not just a convenient collection of familiar labels.

Separate Asset Quantities From Asset Prices

For included assets, the basic calculation is the sum of quantity multiplied by reference price. That identity immediately produces a useful test: hold prices constant and recalculate. The result is not a perfect flow measure, but it shows how much of the reported change depends on asset repricing.

Consider a hypothetical protocol with 10,000 ETH and 10 million dollar-pegged tokens. For this example only, assume each dollar-pegged token stays at $1. At an ETH reference price of $2,000, reported TVL is $30 million. If ETH rises to $3,000 with balances unchanged, TVL becomes $40 million: an increase of 33.3% without any additional deposited units.

Now let the end-period balances change to 12,000 ETH and 8 million dollar-pegged tokens, with ETH still at $3,000. Reported TVL reaches $44 million. At the original reference prices, those end balances are worth $32 million.

The 46.7% headline and the 6.7% constant-price change answer different questions. Neither should be renamed organic growth without investigating the underlying transactions. Added ETH might come from deposits, reward accrual, reinvestment or an accounting change. Reduced stablecoin units might represent withdrawals, repayment or movement between included and excluded contracts.

The first conclusion is narrower and stronger: most of the dollar increase cannot be explained by the quantity change valued at the starting prices.

Reconcile the Interaction Term

There is a further detail worth making explicit. The $14 million increase in the example can be decomposed into three parts:

• Price changes applied to beginning balances: 10,000 ETH multiplied by a $1,000 price increase equals $10 million.
• Quantity changes valued at beginning prices: 2,000 additional ETH at $2,000, less 2 million stablecoin units at $1, equals $2 million.
• The interaction: the additional 2,000 ETH also experiences the $1,000 difference between the two price bases, adding $2 million.

Together, $10 million plus $2 million plus $2 million equals the $14 million reported change. Ignoring the interaction while claiming a complete decomposition leaves the arithmetic unreconciled.

Alternative index methods can allocate the interaction differently. That is acceptable if the method is disclosed and used consistently. The objective is not to establish one sacred definition of flow. It is to keep the reader from confusing an identity with an explanation of customer behavior.

For a genuine net-deposit measure, inspect transfers and classify them under a stated methodology. A balance snapshot by itself cannot distinguish external deposits from internal transfers, fees, rebasing or accrued yield.

Even asset units need a consistent conversion basis. Imagine a hypothetical receipt whose balance remains at 100 tokens while its redeemable backing rises from 1.0 to 1.1 underlying units per token. The receipt count is unchanged, but the represented assets increase from 100 to 110 units. Under the stated assumptions, that increase could reflect accrued income without any new customer deposit. Record both the receipt quantity and the conversion rate; otherwise a unit-based chart can hide a changing economic claim just as a dollar chart can hide repricing.

Receipt Tokens Can Multiply Claims Without Multiplying Assets

Suppose $100 of underlying assets is deposited into a hypothetical vault. The depositor receives a receipt token representing that position and places it in another application. One dashboard may display $100 in the vault and another $100 in the application holding the receipt.

Adding both numbers produces $200 of reported positions. It does not create $200 of distinct underlying assets. There is still the original $100 backing a chain of claims. Yet ignoring the second application entirely would also hide an additional layer of contract exposure.

This is why it is useful to distinguish underlying assets, gross positions and risk-bearing dependencies. They are different analytical objects. DefiLlama's double-counting explanation describes the dashboard treatment of receipt and liquidity-provider tokens deposited into another protocol. Check the setting before comparing totals.

Borrowing adds another dimension. If a user borrows against a receipt and redeploys the proceeds, gross positions can grow without an equivalent increase in that user's equity. The borrowed assets nevertheless come from somewhere, often another participant's deposited capital. A system-wide calculation must define its boundary before netting the balances; otherwise it can erase real external claims or count the same claim repeatedly.

Our guide to DeFi collateral and liquidations explains why this matters under stress. A nested position can transmit a price or liquidity problem across applications even when each individual dashboard appears well collateralized.

TVL Is a Stock; Fees Are a Flow

A stock is measured at a point in time. A flow is measured over a period. Dividing annual fees by one moment's TVL can produce a ratio, but the choice of denominator may distort its interpretation.

Take another illustrative protocol collecting $3 million of fees over a year with average TVL of $30 million. Gross fee intensity is 10%. If protocol-retained revenue is $600,000, revenue intensity is 2%. Neither percentage is automatically the return earned by depositors.

Depositor returns depend on fee allocation, incentives, operating deductions, inventory changes and losses. Our explanation of where crypto yield comes from traces those different sources. A large fee-to-TVL ratio can signal active utilization, high charges, substantial risk compensation or a small denominator. It is a starting question rather than a verdict.

If end-of-year TVL falls to $15 million, dividing the same annual fees by that snapshot produces 20%. The business did not necessarily become twice as productive. The denominator changed. Use average balances or a time-weighted method suited to the service, and state that choice.

Token Terminal's metrics reference distinguishes deposit, usage and financial measures. Comparing them can be useful precisely because none is a complete substitute for the others. To follow the money beyond activity, examine protocol revenue and tokenholder value.

Investigate the Quality of the Deposit Base

Two protocols can display the same TVL while having very different funding stability. In one, many users might deposit for an ongoing service. In another, a few large accounts might supply capital for a temporary reward program. The totals alone cannot distinguish these structures.

Concentration analysis can help, but an address is not necessarily a person. A single address may aggregate many clients, and one operator may control many addresses. Report observable address concentration separately from any claim about beneficial ownership.

Look at what happens around known changes. Do assets remain after an incentive reduction? Does utilization persist outside a market spike? Are deposits accompanied by an actual service users pay for? These questions suggest observations; they do not justify declaring causation from one before-and-after chart.

Token incentives can also change the market value of the assets being deposited. If a protocol's own token appreciates while rewards encourage it to be locked, TVL may reinforce a narrative that depends on the token price itself. The accounting and ownership questions in our token unlock and dilution analysis help identify that circularity.

An incentive-driven deposit is not inherently worthless. Subsidies can introduce users to a useful service. The analytical distinction is between evidence of temporary attraction and evidence that customers will remain under less generous conditions.

Test the Valuation and the Exit Together

TVL values included balances at reference prices. It does not promise those balances can be liquidated at those prices simultaneously. A thinly traded collateral token can have a visible quote while offering little capacity for a large sale.

A dollar-pegged asset introduces another assumption. In the earlier example, the $1 reference price was deliberately fixed to isolate ETH repricing. In a real assessment, the token's market price and its holder's redemption pathway can diverge. Read how stablecoin pegs depend on redemption and liquidity before treating a stablecoin balance as interchangeable with immediately available bank cash.

Also separate TVL from accessibility. Assets can remain recorded in a contract while withdrawals are paused, constrained or economically expensive. A large balance can describe exposure without establishing that every depositor can exit when needed.

Stress the composition, not just the headline. What happens if the largest collateral asset falls, a receipt token trades below its backing value, or a major depositor withdraws? Record which balances reprice, which liabilities remain and which applications depend on the same exit market. An aggregate percentage drop can conceal very different transmission mechanisms.

A Practical TVL Review

The countercase to excessive skepticism is straightforward: TVL can still be informative. A consistent series, with a stable definition and a clear asset composition, helps show the scale of capital exposed to a service. Combining it with unit balances, user behavior and economic flows can reveal meaningful changes.

The mistake is asking the number to answer questions it was not designed to answer. A robust report states what is included, reconciles the change and names the evidence still missing. The next useful observation is the one that separates the competing explanations: more asset units, persistent utilization, a change in incentives or a different concentration of deposits.