Recently, while examining DeFi data, I noticed an interesting phenomenon: the same TVL (Total Value Locked) means completely different things across various projects. The story behind this is quite complex.

It all started when a developer on Solana stacked 11 identities to inflate protocol TVL artificially. After the incident broke out, Defi Llama changed its calculation method in August 2024, defaulting to exclude double counting between protocols. This change directly burst many "bubbles" and gave us a chance to rethink what TVL really is.

Starting from the basic definition, TVL (Total Value Locked) is a commonly used metric to evaluate DeFi projects. On the surface, a higher TVL indicates a larger amount of funds managed by the project. Dividing the total market cap by TVL can quickly help assess the project's valuation. But there's a trap: TVL is static data; today’s number doesn’t change tomorrow with market fluctuations. Cryptocurrency markets are highly volatile, and project incentives and token prices can instantly alter TVL, so relying solely on this metric can be misleading.

More importantly, TVL in different application scenarios is not the same thing at all.

In DEXs, TVL essentially equals liquidity. Uniswap has no liquidity mining and doesn’t require token staking, so TVL directly represents the funds in the pool. But projects like Curve and Sushi allow staking governance tokens to earn fee shares, and those staked tokens can theoretically be counted in TVL. Defi Llama lists them separately under "Staking."

Lending protocols are even more complex. Compound’s TVL refers to the "interest spread," which is the total deposits minus total loans, representing how much liquidity remains in the protocol. Aave adds staking of AAVE and LP tokens for rewards, which are also calculated separately. MakerDAO is different because it lends out DAI issued by the protocol, not using the deposited funds directly, so its TVL equals total deposits.

Looking at DEXs and lending protocols alone generally doesn’t cause double counting of TVL. But once these projects start combining, things get complicated. For example, Aave using Uniswap LP tokens as collateral, or providing aToken liquidity in DEXs, can lead to double counting at the blockchain level. However, such cases are relatively rare.

The real culprit for inflated TVL is projects built on other protocols.

Yield aggregators are typical examples. Yearn and Convex Finance, for instance, put user funds into underlying protocols for yield farming, which were previously double counted in chain-wide TVL. Convex is particularly interesting; it holds large amounts of CRV tokens and stakes them to help users earn higher yields on Curve. Users can swap CRV for CVXCRV to stake and share rewards, but can’t redeem directly, locking CRV more tightly. Convex’s TVL once ranked sixth, with a substantial size. The incident on Solana was similar: developers stacked Sunny (a yield aggregator) with Saber (a DEX). When Solana’s total TVL was only $10.5 billion, these two projects alone accounted for $7.5 billion, with a very high double counting ratio.

Liquidity staking protocols are also hotspots for double counting. Lido is a typical example, with a total TVL of $7.75 billion, of which $7.61 billion is ETH locked on Ethereum. But derivatives like stETH issued by Lido circulate across various DeFi projects. About 21.6% of stETH is used as collateral in Aave, and 14.7% provides liquidity in Curve’s ETH/stETH pool. These amounts are already included in Aave’s and Curve’s TVL. Defi Llama later adjusted its approach, no longer counting funds staked in liquidity staking protocols when calculating chain TVL. But this might lead to underestimation, as some stETH is held at centralized exchanges or institutions, theoretically corresponding to staked ETH on-chain.

Another category is service protocols, such as Instadapp. It acts as middleware, helping users seamlessly connect different protocols and simplify complex operations. Instadapp’s flash loan feature is especially useful for leveraging, deleveraging, or switching debt positions. Its TVL once reached $13.5 billion, but since all funds are stored in other protocols, it was later removed from chain TVL calculations.

Returning to the core question: what is TVL? Simply put, it’s an easily misunderstood but important metric. On the application level, TVL reflects the current fund size of a project and is suitable for cross-project comparison. On the blockchain level, there was previously a lot of double counting. Defi Llama’s recent change caused chain data to drop significantly, bursting the bubble. What remains is a more authentic value. Understanding what TVL means in different scenarios helps us better grasp the true state of the DeFi ecosystem.