How to Assess TVL Quality in DeFi Protocols
Understanding TVL quality DeFi is essential for distinguishing robust protocols from those propped up by temporary incentives or inflated metrics. Total value locked (TVL) alone can be misleading—double-counted liquidity, protocol-owned tokens, and incentive-driven deposits often inflate numbers. This guide provides a systematic framework to assess TVL quality, using real protocols like Aave, Curve, and Lido as examples.
We’ll explore metrics such as token composition, borrowing-to-staking ratios, incentive sustainability, and smart contract risk. By the end, you’ll be equipped to look beyond headline TVL and identify genuinely healthy liquidity in DeFi.
- TVL quality is more important than raw TVL for assessing protocol health and risk.
- Double-counting is rampant; always use de-duplicated TVL from reliable sources.
- High native token share and generous incentive rewards are red flags for TVL quality.
- Borrowing-to-supply ratios over 80% often indicate recursive borrowing, inflating TVL.
- Smart contract risk and centralization (admin keys, proxy upgrades) affect TVL quality.
- Use multiple tools (DeFiLlama, Dune, Token Terminal) to build a complete picture.
Why TVL Quality Matters More Than Raw TVL
Raw TVL is a vanity metric. Two protocols with $1B TVL can have vastly different risk profiles. For example, a lending protocol like Aave where most TVL is supplied by genuine borrowers and lenders is healthier than a yield farm where 90% of deposits come from users chasing token incentives. High TVL quality correlates with sustainable fee revenue, lower impermanent loss risk, and resilience during market downturns. TVL quality DeFi assessment helps you avoid platforms that rely on recursive borrowing or liquidity mined tokens that might crash.
- Genuine liquidity comes from organic user deposits, not from the protocol itself.
- Inflated TVL often uses wrapped versions of the same asset (e.g., wETH on both sides of a DEX) or counts liquidity locked in forks.
- Protocols like Curve and Convex have high quality because their TVL is tied to stablecoin swaps that generate real fees.
Three Layers of TVL: Minted, Borrowed, and Staked
Break down TVL into three categories to assess quality:
- Minted TVL – Assets created by the protocol itself, e.g., DAI in MakerDAO or Lido stETH. This is high quality when backed 1:1 by collateral, but double-counting can occur if stETH is used as collateral elsewhere.
- Borrowed TVL – Assets supplied to lending markets. On Aave, deposits are lent out; if the same asset is borrowed and redeposited (recursive borrowing), it inflates TVL.
- Staked TVL – Assets used for staking, liquidity mining, or vaults. Staked assets are often locked and generate fees, but may be incentivized by the protocol’s own token.
To assess quality, calculate the ratio of borrowed to supplied TVL. On Compound, a high ratio (>80%) suggests heavy recursive borrowing, lowering quality. Tools like DeFiLlama‘s “Breakdown” tab show these splits for major protocols.
Double-Counting: The Hidden Inflator of TVL Quality
Double-counting occurs when the same asset is counted as TVL in multiple protocols or multiple times within the same protocol. Classic examples:
- wETH/ETH liquidity pairs – A Uniswap pool with wETH and USDC counts both sides. If wETH is minted from a bridge, the ETH is counted again in the bridge’s TVL.
- Liquid staking tokens – Lido’s stETH is counted in Lido’s TVL. When stETH is deposited into Aave, it appears in both protocols. DeFiLlama attempts to de-duplicate this, but not all aggregators do.
- Forked protocols – Liquidity on a fork like SushiSwap that is also locked in a MasterChef contract may be counted multiple times if not properly aggregated.
A rule of thumb: subtract any TVL that comes from wrapped or liquid staking derivatives of the protocol’s own asset. For example, Curve’s stETH/ETH pool TVL is partially double-counted with Lido’s TVL.
Assessing Token Composition and Concentration
TVL concentrated in a single asset or a few addresses is low quality. A high share of the protocol’s own token (or governance token) indicates incentive-driven liquidity.
Steps to evaluate:
- Use DeFiLlama‘s “Exposures” tab to see asset composition. For example, a stablecoin swap protocol with 90% USDC is more diversified than one with 70% in a single volatile token like CRV.
- Check the top 10 depositors’ share. Tools like Dune Analytics or Nansen show concentration. If the top 5 wallets control >50% of TVL, the protocol is vulnerable to whale withdrawals.
- Avoid protocols where the native token represents >20% of TVL, as that often signals self-referential liquidity (e.g., lending your own token to earn more tokens).
| Protocol | Native Token % of TVL | Quality Assessment |
|---|---|---|
| Curve | ~5% (CRV in gauges) | High |
| Yearn Finance | ~10% (YFI in vaults) | Medium-High |
| A hypothetical farm | 80% | Low |
Incentive Dependency: Is TVL Sustainable?
Protocols that pay high token incentives (e.g., >50% annualized yield) attract mercenary capital. When incentives drop, TVL often evaporates.
To measure incentive dependency:
- Calculate the ratio of annualized incentive spend (in USD) to protocol revenue. If incentives exceed revenue, the protocol is burning capital to maintain TVL.
- Look at the inflation rate of the governance token. Tokens with >10% annual inflation used for rewards indicate low organic demand.
- Check the age of deposits: older deposits (>30 days) suggest stickier TVL. Tools like Token Terminal provide revenue vs. incentive data.
For example, Convex Finance had high TVL but much of it was driven by CRV rewards. After a CRV price drop, TVL fell significantly. In contrast, Aave’s TVL remained relatively stable during market downturns because borrowing demand is real.
Smart Contract Risk and TVL Quality
TVL locked in unaudited or low-liquidity contracts is inherently lower quality. Even with high TVL, a hack can drain funds instantly.
Assess smart contract risk via:
- Audit history – multiple audits by reputable firms (Trail of Bits, OpenZeppelin, ConsenSys Diligence).
- Bug bounties – high value (>$1M) indicates confidence.
- Time since deployment – established protocols like MakerDAO (since 2017) have proven resilience.
- Code upgrades – proxy contracts with admin keys pose centralization risk. Tools like DefiSafety score protocols on these factors.
For example, a new fork of Uniswap with no audit and a $10M TVL is far riskier than a similarly sized but audited DEX. Higher TVL quality often correlates with lower smart contract risk.
A Framework for Comparing TVL Quality Across Protocols
Use this multi-dimensional framework to compare protocols:
| Metric | High Quality | Low Quality |
|---|---|---|
| Double-counting adjustment | De-duplicated TVL (from DeFiLlama or Dune) | Headline TVL |
| Borrow/Supply ratio | < 0.5 (organic lending) | > 0.8 (recursive borrowing) |
| Native token share | < 10% | > 30% |
| Incentive/revenue ratio | < 1 (profitable) | > 2 |
| Address concentration | Top 10 < 20% | Top 5 > 50% |
| Audit & bug bounty | Multiple audits + bounty > $500k | No audit or no bounty |
Apply this framework to any protocol. For example, Compound scores well on borrow/supply ratio but may have moderate incentive dependency. Lido scores well on smart contract risk but has double-counting issues with stETH deposits elsewhere.
Tools to Assess TVL Quality in Real Time
Several platforms help you evaluate TVL quality DeFi:
- DeFiLlama – Shows adjusted TVL, breakdowns by asset, and historical trends. The “Liquidations” tab also reveals leverage.
- Dune Analytics – Custom dashboards to analyze on-chain metrics like deposit concentrations, incentive payouts, and borrowing ratios. Example: community-built dashboards can surface recursive borrowing rates for lending protocols like Aave.
- Token Terminal – Focuses on revenue and incentive spend. Compare a protocol’s P/E ratio with TVL growth.
- DefiSafety – Scoring for smart contract risk and admin keys.
- Nansen – Tracks whale wallets and smart money flows.
Combine these tools with manual checks: read the protocol’s docs for tokenomics, check liquidity on both sides of pairs, and verify if liquidity is locked in timelocks.
Case Study: High vs Low TVL Quality
Compare two hypothetical protocols to illustrate the framework:
Protocol A (High Quality): Aave on Ethereum. It has $5B TVL, but after adjusting for double-counting (stETH, wETH) it’s ~$4.5B. Borrow/Supply ratio 0.6, native token (AAVE) only 2% of TVL, incentive/revenue ratio 0.8, top 10 wallets hold 12%, multiple audits and $5M bug bounty.
Protocol B (Low Quality): A new yield aggregator with $1B TVL. Headline TVL includes LP tokens from its own liquidity pool (self-referential). Native token (XYZ) is 60% of TVL, borrow/supply ratio 0.95 (most deposits are borrowed and redeposited), incentive/revenue ratio 4.0 (burning tokens to attract liquidity), top 5 wallets hold 70%, no audit.
Protocol A is clearly superior. This exercise shows why raw TVL is insufficient. Always dig deeper.
Step-by-step
- Identify all asset types locked in the protocol and check for wrapped or derivative tokens.
- Use DeFiLlama's 'Adjust TVL' feature to see de-duplicated numbers.
- Calculate the borrow-to-supply ratio on lending protocols via Dune Analytics.
- Determine the share of the protocol's native token in total TVL (look at each pool).
- Compare annualized incentive spending to protocol revenue using Token Terminal.
- Check the concentration of top depositors (top 10 wallets) via Etherscan or Nansen.
- Review audit history and bug bounty programs on DefiSafety or protocol docs.
- Apply the comparison table framework to score the protocol against high-quality peers.
Common mistakes to avoid
- Failing to subtract liquidity that is locked in the protocol’s own liquidity pair (self-referential).
- Ignoring double-counting across multiple protocols, e.g., stETH counted in both Lido and Aave.
- Assuming high TVL always means high liquidity depth; concentrated positions can skew TVL.
- Overlooking governance token inflation that artificially props up yield incentives.
- Not checking if the TVL is dominated by a few whales who can withdraw en masse.
- Relying on a single aggregator’s TVL number without understanding its de-duplication logic.
Frequently asked questions
What is the difference between TVL and adjusted TVL?
TVL counts all assets locked in a protocol's smart contracts. Adjusted TVL subtracts double-counted assets, such as liquidity that is also counted in another protocol (e.g., stETH) or within the same protocol (e.g., LP tokens that represent underlying assets).
How does incentive dependency affect TVL quality?
High incentive dependency means deposits are driven by token rewards rather than organic demand. When reward rates drop or token price falls, TVL can collapse quickly, making the protocol less stable.
Can a protocol have high TVL but low liquidity depth?
Yes. For example, if TVL is concentrated in a few large positions, actual tradable liquidity (depth) may be thin. A pool with $100M TVL but dominated by one whale might have low effective liquidity for large orders.
Which tools are best for checking double-counting?
DeFiLlama's 'Adjusted TVL' and Dune Analytics custom dashboards are best. Token Terminal also provides protocol-level breakdowns that exclude overlapping assets.
Related reading
Track the entities behind the concepts
DeFi Intel maps 11,000+ protocols, tokens and companies to a typed knowledge graph — with live data, incidents and regulation.