DeFi Total Value Locked: What It Measures and Its Limits

Research question and scope

This article examines what Total Value Locked (TVL) actually measures, how it compares to two other commonly cited DeFi adoption proxies, and where each can mislead. It does not present any live TVL, revenue, or user-count figures for any specific protocol — the goal is understanding what these metrics mean, not tracking their current values.

Background and context

TVL became DeFi's dominant headline metric largely because it was easy to compute and reflected real, verifiable on-chain deposits rather than a purely speculative figure. As DeFi's ecosystem grew more interconnected -- the same token wrapped, supplied as collateral, borrowed against, and redeposited into a second protocol -- the metric's simplicity became its weakness: aggregators like DefiLlama sum each protocol's reported deposits independently, with no built-in way to recognize that one unit of underlying capital is responsible for several of those deposits at once. This is the specific, structural concern a 2024 academic paper set out to measure directly rather than describe qualitatively.

Key findings

• TVL measures deposited capital, not usage, revenue, or genuine demand — a protocol can have substantial TVL with little real transactional activity, or the reverse.
• The same underlying capital can inflate TVL across multiple protocols simultaneously through re-deposit chains, a form of double-counting specific to this metric.
• No single commonly cited DeFi metric (TVL, protocol revenue, or active addresses) captures adoption completely on its own — each measures a genuinely different thing and has its own blind spot.

What TVL measures

Total Value Locked (TVL) generally refers to the total value of crypto assets deposited into a specific DeFi protocol's smart contracts — for example, funds supplied to a lending protocol or deposited into a liquidity pool.

Why it's widely used

TVL is commonly used as a rough proxy for a protocol's scale and adoption, since it reflects real capital that users have chosen to deposit rather than a purely speculative metric like token price alone.

Real limitations of the metric

• Double-counting: the same underlying capital can be counted multiple times if it's deposited, borrowed against, and redeposited across multiple protocols in a chain.
• Denomination effects: TVL is often measured in the underlying assets' current market value, so it can rise or fall simply due to price movement rather than any actual change in deposited quantity.
• Incentive-driven deposits: temporary token reward programs can inflate TVL with capital that may leave quickly once incentives end, rather than reflecting durable adoption.

TVL compared to two other common adoption proxies

Evidence: quantifying the double-counting problem

Luo, Feng, Xu & Tasca's paper proposes "total value redeemable" (TVR) as an alternative metric designed to strip out double-counting, and directly measures the gap between it and conventional TVL. Their headline, directly-confirmed finding: the gap between TVL and TVR reached $139.87 billion at the peak of DeFi activity (December 2, 2021), with a TVL-to-TVR ratio of roughly 2 -- meaning headline TVL at that peak was overstating genuinely distinct, redeemable capital by a factor of approximately two. The paper also finds TVL is meaningfully more unstable than TVR during downturns: a 25% decline in Ether's price produced a $1 billion larger non-linear decrease in TVL than in TVR, which the authors attribute to liquidations triggered by derivative tokens amplifying TVL's apparent decline beyond the underlying capital's actual change. This gives the "double-counting" and "denomination effects" limitations already described above concrete, quantified magnitudes rather than only a qualitative description.

Independent critical analysis

The TVR framework is a genuine methodological improvement, but it does not fully dissolve the underlying measurement problem -- it relocates it. Correctly computing TVR requires reliably identifying which tokens are derivative/wrapped claims on other deposited capital versus genuinely new capital, a classification problem that gets harder, not easier, as DeFi's composability deepens and new wrapped-asset and leveraged-vault designs are introduced. Separately, secondary reporting (not independently verified in this pass) describes DefiLlama itself offering a "double count" toggle that most users reportedly leave in its default, non-adjusted state, and describes DefiLlama's founder publicly disputing specific double-counting claims about individual protocols' reported figures. Taken together, this suggests the practical gap between TVL and a more accurate figure like TVR is not just a matter of an academic metric existing -- it also depends on whether aggregators' adjusted figures are actually used by the readers citing them, which the available evidence suggests they usually aren't.

Limitations of this comparison

• This describes each metric's documented definition and known blind spot in general, not a scored or ranked comparison of which metric is "best."
• No live figures for TVL, revenue, or active addresses for any specific protocol are presented here — verify current values directly before relying on them.
• A protocol can report these metrics with differing methodologies; comparing two protocols' self-reported numbers directly can itself be misleading without checking how each was calculated.
• The $139.87 billion figure and the TVL-to-TVR ratio of roughly 2 are specific to the December 2, 2021 peak the paper analyzed -- they describe a historical snapshot, not a current or constant gap between the two metrics.
• This pass directly read the paper's own abstract to confirm its headline figures and methodology description, but did not independently re-derive its underlying on-chain dataset or regression results from the full paper.
• The claims about DefiLlama's own double-count toggle and public response to specific inflation claims come from secondary industry reporting (a 2026 trade-press article), not independently verified by this research against DefiLlama's own statements.

Research conclusion

TVL is a genuinely useful, widely used proxy for capital deposited in DeFi, but it has real, documented, and now quantified limitations — double-counting that reached roughly $140 billion and a TVL-to-TVR ratio near 2 at DeFi's late-2021 peak, plus denomination effects that make TVL meaningfully more volatile than a double-counting-adjusted alternative during downturns. These make TVL most useful alongside other metrics, including a double-counting-adjusted figure where available, not in isolation.

References

• Primary (read directly, abstract): Luo, Y., Feng, Y., Xu, J. & Tasca, P., "Piercing the Veil of TVL: DeFi Reappraised", arXiv 2404.11745 -- arxiv.org/abs/2404.11745. The $139.87B double-counting gap, ~2x TVL-to-TVR ratio at the December 2, 2021 peak, and the 25%-ETH-decline/$1B-differential stress-test finding all come directly from this paper's own abstract.
• Secondary (industry commentary, found via WebSearch, not independently verified): trade-press coverage (2026) of DefiLlama's own "double count" toggle and its founder's public response to specific protocol-level TVL-inflation claims.