What problem is DeFiLlama TVL actually trying to solve
Total value locked, or TVL, exists because DeFi is unusually hard to size. There is no quarterly report, no audited balance sheet, and no central registry of who has deposited what into which smart contract. In 2020, a developer going by 0xngmi built DeFiLlama to scrape on-chain data and convert it into a single comparable number. It became the de facto dashboard of the industry almost by accident, and it is still the first page most analysts open.
The pitch is simple. If you can see, in dollars, how much capital is sitting inside a protocol, you can compare protocols, rank chains, and spot trends. The trouble is that the number is doing more work than it was designed for. People use it as a proxy for adoption, for trust, for safety, and for growth, and none of those things is what it measures.
This matters because TVL is the most quoted metric in DeFi, and it is also the one most often used to justify a bad decision. A protocol can have a billion dollars of TVL and still be hollow. A chain can have a hundred billion and be propped up by a handful of bridges and a yield farm that ends next Tuesday. Before you trust the number, you have to know what it actually counts, and what it deliberately leaves out.
Risks of treating TVL as a trust signal
The first risk is the simplest: TVL is a snapshot, not a verdict. A protocol can show a clean upward line on the chart and be one oracle failure, one governance attack, or one rug pull away from a vertical drop. The number on the dashboard does not know the difference between capital that arrived to use the product and capital that arrived to chase a points program or a token airdrop. The day the rewards stop, that capital is usually gone within hours.
The second risk is the illusion of liquidity. A lending market that lists a billion dollars of supply TVL may have far less usable liquidity once you account for utilization, queued withdrawals, and borrowed balances. The visible number gives a false sense of how easy it is to exit. Several major protocols, including the failed Terra ecosystem in 2022, looked healthy on TVL charts right up until the moment they did not.
The third risk is concentration. A single wallet can move the needle. In the early days of many yield farms, a few addresses provided most of the TVL, which meant one signature could crash the number by 30 percent or more. When you see a small protocol with a sudden TVL spike, ask who is providing it before you assume the demand is real.
How TVL is actually calculated on DeFiLlama
DeFiLlama pulls token balances directly from the smart contracts of each protocol. For a decentralized exchange, that means the balances sitting in the liquidity pools. For a lending market, it means the supply side deposits. For a staking protocol, it means the staked assets. Each balance is then multiplied by a price, usually from a price oracle such as CoinGecko or a chain-native feed, and the result is summed into a single USD figure.
Two details matter here. First, the price is a current spot price, not a moving average or a volume-weighted price. If a governance token gets a 80 percent wick on a low-liquidity exchange, the TVL of any protocol that holds that token will briefly look enormous. DeFiLlama does try to filter outliers, but the number is only as good as the oracle behind it.
Second, the number is double-counted by design. If you deposit USDC into a lending market, then borrow that USDC and deposit it into a yield aggregator, your original dollar is now counted twice. The same dollar is sitting in two contracts. DeFiLlama flags this with a "doublecount" toggle on some pages, and the toggle usually shaves a third or more off the headline figure. When you compare two protocols, always check whether double-counting is on or off, because the comparison is meaningless otherwise.
Borrowed TVL vs supply-side TVL
One of the easiest mistakes is to read a lending market's supply TVL as if it were the size of the business. It is not. Supply TVL is the total deposited. Borrowed TVL is how much of that has actually been lent out. The difference is the available liquidity, and it is the part that can be withdrawn at any moment.
Imagine a market with one billion dollars of supply TVL and 800 million of borrowed TVL. The headline number suggests a billion dollars of liquidity, but only 200 million is actually free. If a wave of withdrawals arrives, the protocol has to either wait for borrowers to repay, liquidate collateral, or pause withdrawals. Several real-world incidents, including parts of the Celsius and BlockFi unwind in 2022, were driven by exactly this gap between supply and available liquidity.
On DeFiLlama, the borrowed side is shown separately. The ratios worth watching are utilization (borrowed divided by supply) and the share of borrows in a single asset. A market where 90 percent of borrows are in one volatile token against a thin collateral pool is a market that can drain in a single liquidation cascade, regardless of what the supply TVL chart looks like.
Chain-level TVL and protocol-level TVL
Chain TVL aggregates every protocol running on a given network. Protocol TVL is the figure for a single application. They answer different questions. Chain TVL tells you where capital is parked across the industry. Protocol TVL tells you whether one application is winning or losing share inside that chain.
This distinction matters when narratives shift. A chain's TVL can rise while every major protocol on it is flat, because a new bridge or a new staking product absorbed inflows. Conversely, a chain's TVL can fall even though its leading DEX grew, because a single large protocol lost share. If you are evaluating a chain, look at the protocol breakdown, not the headline. If you are evaluating a protocol, look at its market share within the chain, not the chain number.
DeFiLlama's chain pages also have a "exclude doublecount" toggle that removes assets which have been bridged and re-deposited. Bridged assets are not new capital arriving on the chain, they are the same capital counted again. Turning the toggle on gives a closer read on organic inflows, and it is almost always the number you should be looking at when comparing two chains of similar size.
The farm-and-dump TVL illusion
The most common pattern in DeFi is also the easiest to fall for. A protocol launches, announces a points program or a token emission, and its TVL climbs on the chart in a near-vertical line. The chart looks like a hockey stick. Headlines call it the next big thing. Six weeks later, the emission gets diluted, the token unlocks, and the TVL drops by 80 percent in a week.
This happened repeatedly between 2020 and 2024, with yield farms on Ethereum layer-2 networks and on Solana. The protocol that was "growing the fastest" turned out to be the protocol that was paying the most for its TVL, and when the budget ran out, the chart inverted. The same capital that arrived to farm the rewards left to farm the next rewards.
DeFiLlama's own charts are good at showing this in hindsight. The trick is to read the chart live, while the program is still running, and ask whether any non-incentivized user would deposit. If the only reason to be there is the reward, the TVL is borrowed time. A useful test is to look at the fees the protocol collects compared to the value of the emissions it pays out. If emissions are an order of magnitude larger than fees, the TVL is being bought, not earned.
Fee, revenue, and active-address metrics as sanity checks
TVL tells you the size of the pool. Fees tell you whether anyone is using it. Revenue tells you whether the protocol is capturing any of the value from that usage. Active addresses tell you how many distinct users are actually transacting. Read together, these four numbers give you a much more honest picture than TVL alone.
DeFiLlama has a dedicated section for fees and revenue on most protocol pages, and the chain-level dashboards now include active-address data sourced from providers such as Artemis. A protocol with steady TVL, growing fees, and a stable or growing address count is doing something real. A protocol with rising TVL, flat or falling fees, and a shrinking address count is buying its chart with emissions and will likely lose the chart when the emissions stop.
A useful minimum routine is to open three tabs for any protocol you are evaluating. First, the protocol's TVL chart, with the double-count toggle on, over a 90-day window. Second, the fees and revenue chart over the same window. Third, the active-address chart over the same window. If the three lines all point in the same direction, the trend is probably real. If TVL is rising while fees and addresses are flat or falling, treat the TVL number as borrowed and assume it will leave.
Case study: a TVL collapse that looked like a blip until it wasn't
One of the clearest examples is the Anchor Protocol on Terra. For most of 2021, Anchor's TVL was quoted in headlines as proof that algorithmic stablecoins and DeFi savings accounts were working. The number climbed past 14 billion dollars at its peak in early 2022, which made it one of the largest protocols by TVL on DeFiLlama.
What the headline number hid was that almost the entire TVL was sustained by a yield subsidy paid out of the Luna Foundation's reserves, not by real borrowing demand. Fees were modest, and the address base was concentrated. When the reserve budget was no longer sustainable, the peg broke, the TVL collapsed by more than 99 percent in days, and the holders lost access to their funds. The TVL chart on DeFiLlama was technically accurate. It just did not, on its own, say anything about whether the capital was there voluntarily.
The lesson generalizes. Any protocol whose TVL is heavily subsidized, whose fees are a small fraction of its emissions, or whose active-address curve is flat, is one bad week away from the same shape on the chart. The dashboard will show you the drop in real time. It will not warn you before it happens. That part is on you.
How to follow DeFi TVL the smart way
DeFi TVL moves fast, and the news around it moves faster. A protocol can spike on a Friday afternoon and be unwound by Sunday morning. Tracking the TVL number alone is a losing game, because the number is silent about the reasons behind the move. Zippfeed surfaces DeFi protocol and chain headlines with sentiment scoring (bullish, neutral, or bearish) and an importance rating, so you can see why a TVL line moved, not just that it did, and decide whether the move reflects real demand or another farm-and-dump cycle starting to unwind.
