DeFi TVL analysis needs more than a dollar total. A six-day Aave V3 Ethereum snapshot shows how valuation changes and token balances can pull in opposite directions.
Reading DeFi TVL growth starts with asking what actually increased. Existing tokens can become more valuable without anyone depositing another coin. More tokens can also be held while their combined dollar value falls. We separated price and quantity effects in Aave V3’s Ethereum data to test the capital-flow interpretation, not to award the protocol a safety score.
Observation window: 18 October 2026 at 00:00 UTC to 24 October 2026 at 00:00 UTC. The source is a DeFiLlama public API response retrieved on 25 October. These are fixed historical observations, not a live counter or a total across all Aave deployments.
TVL is a stock; an inflow is a movement
TVL values the assets included in a measurement at a particular moment. Each token quantity is multiplied by its price, then the values are added. A rising total alone says nothing conclusive about user growth, revenue or investor returns.
Imagine a pool holding 10 ETH when ETH costs $2,000. Its value is $20,000. At the next observation, the balance falls to 9 ETH while the price rises to $2,500. The total becomes $22,500: TVL has increased 12.5%, although the ETH balance has fallen 10%. Growth in value is established; fresh deposits are not.
DeFi protocol revenue needs a separate examination. Assets held, fees paid by users and revenue retained by a protocol measure different things. A large balance can coexist with weak economic activity.
A real snapshot: Aave V3 on Ethereum
We used the API’s Ethereum branch, excluding the borrowed branch, Aave V2 and other chains. The source methodology excludes borrowed coins from core TVL. The result should therefore not be equated with the gross value of all user claims.
| Measurement | Result | Interpretation limit |
|---|---|---|
| 18 October 2026 TVL | $13,982,777,170 | Starting observation |
| 24 October 2026 TVL | $13,969,430,630 | Ending observation |
| Total change | −$13,346,540 / approximately −0.0954% | Not a direct measure of withdrawals |
| Price effect | −$71,363,479 | Starting quantities held constant |
| Quantity effect | +$58,016,939 | Balance changes valued at ending prices |
All 55 token entries matched across the two dates. The components reconcile to the reported TVL change with less than $1 of rounding residual. The headline chart shows a decline. The decomposition shows a negative price effect and a positive quantity effect, largely offsetting each other.
The largest positive quantity contributions were approximately $66.26 million from USDT and $60.97 million from WETH. WstETH contributed about −$30.45 million and USDC −$19.47 million. These are balance effects valued at endpoint prices, not individual investment amounts or a sum of traced deposit and withdrawal transactions.
How to reproduce the calculation
For each token, let q₀ and q₁ be starting and ending quantities, and p₀ and p₁ their prices. Dividing the token’s API dollar value by its quantity recovers the implied valuation price at that observation. The two components are:
- Price effect: q₀ × (p₁ − p₀).
- Quantity effect: p₁ × (q₁ − q₀).
Their sum equals q₁p₁ − q₀p₀. Changing price first assigns the interaction between price and quantity to the quantity component. Reversing the order changes the components, but not the total change. Two tables labelled “flows” should not be compared without checking their conventions.
In the 10 ETH illustration, the price effect is 10 × $500 = $5,000. The quantity effect is $2,500 × (9 − 10) = −$2,500. The net increase is $2,500. We applied the same identity to the 55 matched tokens in the real snapshot. New or missing token entries would require a separate unexplained component rather than silently disappearing.
The Aave V3 API response contains tokens, tokensInUsd and tvl series under chainTvls → Ethereum. Select the stated dates and match token names and timestamps. Historical data may later be revised; our calculation uses the response archived on 25 October.
Why a positive quantity effect is not proven deposit inflow
Tokens held by a lending contract can change through borrowing and repayment as well as deposits and withdrawals. Interest accrual, token conversions and adapter changes also need examination. Repayment of an existing loan can increase the contract balance without a new saver making a deposit.
DeFiLlama’s USD Inflows metric attempts to separate price changes using daily asset-balance differences. Our six-day endpoint-price decomposition is not that daily series. A claim about actual deposit flows requires reconciling deposit, withdrawal, borrowing and repayment events against onchain records. We have not performed that transaction-level reconciliation.
The lending model and its possible transmission to AAVE holders are examined separately in our Aave analysis. Treating every TVL change as token demand or an expected price gain would go beyond the measurement.
Three ways a comparison can mislead
| Observation | Alternative explanation | Check |
|---|---|---|
| TVL grows rapidly | Collateral tokens appreciated | Token quantities and fixed-price changes |
| The total jumps overnight | A chain, version or asset entered coverage | Adapter notes and consistent scope |
| Combined protocol totals look large | Underlying collateral appears more than once | Receipt tokens and aggregation filters |
Double counting in DeFi concerns underlying economic collateral appearing in more than one protocol. The same value can carry several protocol exposures without being independent fresh capital each time. Comparing a dashboard with staking or borrowed enabled against another with those categories excluded can also distort rankings.
Stablecoin-heavy balances are not entirely free of valuation effects. Depeg and collateral risk make a permanent one-token-equals-one-dollar assumption unsafe. Keep the source’s actual valuation separate from a fixed-dollar modelling assumption.
What would support a stronger growth claim?
I would look for sustained asset growth under consistent coverage alongside usage and revenue, rather than a rising TVL line alone. Deposit and withdrawal records help identify the source; concentration reveals dependence on a few large wallets; behavior after incentives end helps test durability. Each measure has limits: addresses are not people.
The Aave snapshot supports a narrower finding: a small decline in dollar TVL coincided with a positive quantity effect. It does not establish net user flows, contract safety or token returns. Our DeFiLlama review explains the dashboard and filter choices that can help readers make their own comparisons.
Sources and method
DeFiLlama data definitions, filter explanations and the Aave V3 adapter were checked on 25 October 2026. The decomposition is our calculation from reported quantities and dollar values. It is not a contract audit, complete onchain flow study or investment recommendation.
