Pith. sign in

REVIEW 4 major objections 5 minor 51 references

Towards Verifiability of Total Value Locked (TVL) in Decentralized Finance

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper argues that published DeFi TVL is only partly verifiable from on-chain data and introduces vTVL, a reconstruction metric showing 23.5% of 400 protocols match nearly perfectly and 46.5% align within an acceptable band.

desk verdict A useful, honest measurement study of how DeFiLlama computes TVL, with a softer vTVL 'verification' result than the abstract suggests. read the letter →

arxiv 2505.14565 v1 pith:NVBUMOBZ submitted 2025-05-20 q-fin.GN cs.CRcs.CYcs.ET

Pietro Saggese , Michael Fröwis , Stefan Kitzler , Bernhard Haslhofer , Raphael Auer This is my paper · ORCID
classification q-fin.GNcs.CRcs.CYcs.ET
keywords TotalValueLockedDeFiverifiabilityon-chaindataLlamavTVLEthereumdoublecounting
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Total Value Locked (TVL) is the main yardstick for the size of DeFi, but the paper argues that as currently computed it is opaque: aggregators rely on community-submitted plugins, and computation methods vary, so published numbers cannot be independently reproduced. The paper studies 939 Ethereum DeFi projects by executing the DeFiLlama plugin code and recording every call it makes, finding that 10.5% of protocols depend partly or wholly on external servers, 68 non-standard balance functions are in use, and 240 balance queries hit the same contracts and tokens for different protocols. To measure how much TVL can actually be verified, the paper introduces verifiable Total Value Locked (vTVL), which rebuilds each protocol's TVL using only on-chain data and standard balance queries. In a case study of 400 protocols, 23.5% match published TVL almost perfectly and 46.5% fall within an aligned band. The point of the exercise is that verifiability is a design choice: with standardized contracts, token lists, and balance methods, TVL could be checked independently rather than trusted.

What carries the argument

The load-bearing instrument is a proxy-based measurement pipeline that executes the DeFiLlama adapter (plugin) code for each protocol and records all HTTP and JSON-RPC calls, so every interaction with Ethereum nodes and external hosts is observable. On top of that sits the vTVL construction: from the recorded eth_getBalance and balanceOf calls it derives the set of addresses and tokens that contribute to a protocol's TVL, queries their historical balances, prices tokens using Uniswap V2 liquidity pools, and sums the values. The Discrepancy Ratio, the average of vTVL over published TVL minus one, is the paper's yardstick for how much of a reported figure can be reproduced. The regular-expression scan for alternative balance functions (e.g., totalAssets, getTotalPooledEther) is what exposes the 68 non-standard methods that make verification harder.

What would settle it

Pick any protocol from the 46.5% whose vTVL is aligned with published TVL, obtain its plugin-queried address-token list, and re-run the balance queries at each monthly snapshot against an independent Ethereum archive node; if even one major protocol's reconstructed value moves outside the reported +/-0.5 discrepancy band, the vTVL alignment claim for that protocol fails. Alternatively, for a protocol like Maker where vTVL diverges after late 2022, check whether the divergence coincides with plugin calls to an external server or with the API folding in a second protocol, which would identify exactly which assumption broke.

Watch

Extended reading notes

Core claim

The paper's central discovery is that TVL's verifiability is empirically low and systematically measurable. By proxying the execution of DeFiLlama's TVL plugins, the authors capture every on-chain and off-chain interaction used to compute TVL for 939 Ethereum protocols. They find that the majority of protocols (78.6%) use standard balance queries, but 68 self-defined functions also feed TVL, 10.5% of protocols rely on external servers, and 240 balance queries are shared across multiple protocols, creating a concrete double-counting risk. The paper then defines vTVL, the value recoverable by replaying only standard balanceOf and eth_getBalance queries on the addresses and tokens the plugins touch, and prices those holdings with Uniswap V2 exchange rates. Comparing vTVL to DeFiLlama's published TVL for 400 protocols, the Discrepancy Ratio is near zero for 23.5% of protocols and aligned within +/-0.5 for 46.5%, while large discrepancies concentrate in smaller protocols. The conclusion is that only part of published TVL is independently verifiable today, and that the rest is not necessarily fraudulent but is beyond what on-chain data alone can confirm.

Load-bearing premise

The vTVL reconstruction assumes that the addresses and tokens the DeFiLlama plugin queries during TVL computation exactly define the protocol's TVL; if a plugin queries addresses that are not locked value, or misses addresses that are, the vTVL estimate drifts from true TVL by an unknown amount.

Editorial extensions

If this is right

  • If vTVL were adopted as a standard, independent auditors could recompute a protocol's TVL from a publicly visible address-token list without trusting the aggregator.
  • The 240 shared balance queries imply a measurable double-counting risk at the infrastructure level: the paper shows the value in those contract accounts peaked near $16 billion at the end of 2021.
  • The decreasing share of alternative balance functions (28.2% in January 2023 to 8.9% in January 2024) suggests that standardization pressure is already shifting TVL computation toward standard on-chain methods.
  • The design guidelines - on-chain sources, published contract and token lists, standard balance methods, published token categorizations, and common protocol selection criteria - would make TVL figures reproducible and explainable if aggregators adopted them.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same proxy-and-replay method could be run continuously on other EVM-compatible chains and on a moving window of plugin commits, turning vTVL from a one-off case study into an automated verifiability monitor for TVL.
  • A protocol could game vTVL by including extra low-value addresses in its plugin or by omitting large holdings, so vTVL is best read as a lower bound on what is verifiable, not as proof that unverified value is absent.
  • Linking vTVL to the TVR distinction between derivative and non-derivative tokens could yield a verifiable redeemable value metric, addressing both double counting and verifiability in one number.
  • The finding that most discrepancy concentrates in small protocols suggests verification effort should be prioritized by TVL size; checking the top protocols would recover most of the ecosystem's verifiable value.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper studies how Total Value Locked (TVL) is computed by the DeFiLlama aggregator for 939 Ethereum DeFi protocols. Using a proxying pipeline over the DeFiLlama plugin runtime, it reports that 10.5% of protocols rely at least partly on external servers, that 68 non-standard balance functions are used, and that 240 balance queries are duplicated across protocols. It then introduces verifiable TVL (vTVL), an on-chain reconstruction of TVL using only standard balance queries (eth_getBalance and ERC-20 balanceOf), and reports that in a case study of 400 protocols, vTVL aligns with published TVL within a discrepancy ratio of ±0.5 for 46.5% of protocols and within ±0.05 for 23.5%. The paper concludes with design guidelines for more standardized and verifiable TVL computation.

Significance. If the main claims hold, this is a timely and useful measurement study. Its strengths are concrete: a transparent, repeatable instrumentation pipeline applied across five repository commits; direct observation of external-server dependencies, non-standard balance functions, and duplicate queries; and publicly available code and data. These descriptive findings are valuable regardless of the vTVL caveat. The proposed vTVL metric is a reasonable operationalization of 'verifiable' TVL, but its independence from the aggregator's own address and token selection is the central weak point, and the quantitative headline (46.5% alignment) currently rests on permissive and post hoc choices. The paper's contribution is therefore better characterized as a careful reproducibility analysis of DeFiLlama's TVL pipeline than as independent verification of published TVL.

major comments (4)
  1. [IV-A, Appendix F, footnote 2] The vTVL reconstruction inherits the address and token sets directly from the DeFiLlama plugin calls: Appendix F states that the authors 'inspect calls that represent balance queries for Ether or tokens, assuming that addresses queried during computation contribute to the project TVL,' and footnote 2 explicitly acknowledges that this assumption may fail. Because the audited system supplies the address set, a match between vTVL and published TVL can indicate faithful reproduction of DeFiLlama's own bookkeeping rather than independent on-chain verification. This directly affects the headline claim that vTVL can 'independently verify just a part of the reported TVL.' The authors should either rephrase vTVL as a plugin-consistent reconstruction, or validate the address lists against an independent source (e.g., protocol documentation, transfer-event analysis, or a second aggregator).
  2. [IV-B and Appendix F] The 46.5% alignment figure depends on a permissive threshold: a discrepancy ratio within ±0.5 means vTVL is between 50% and 150% of the published value. It also depends on several post hoc filtering choices with no sensitivity analysis: the 20% rolling-average outlier threshold, the minimum of 10,000 price points per token, the manual removal of 80 high-volatility tokens, and the 20 mln USD stablecoin market-cap cutoff. Since vTVL is acknowledged to be an underestimation by construction, the reported breakdown across a single threshold set is hard to interpret. The authors should report how the alignment percentage changes across a range of thresholds and filtering choices, or at minimum provide a distribution of the discrepancy ratios rather than only the two summary intervals.
  3. [IV-A] The comparison between vTVL and published TVL uses DeFiLlama API values that aggregate several columns ('Ethereum', 'Ethereum-borrowed', 'Ethereum-pool2', 'Ethereum-staking', 'Ethereum-vesting'), while vTVL only captures standard balance queries on addresses obtained from plugin calls. These components need not correspond to the same economic value, and the version mismatches noted in Appendix F (e.g., Aave v2/v3, Uniswap v2/v3) are corrected only ad hoc for a few large protocols. This can bias the discrepancy ratio for a substantial number of protocols. A systematic reconciliation of API columns with the captured on-chain calls, or an explicit statement of which API components are excluded, is needed before the aggregate alignment statistics can be treated as reliable.
  4. [IV-B] The case study sample is restricted to 400 protocols with at least one recorded eth_getBalance or balanceOf call and available API data. Protocols that rely solely on external hosts, that use only non-standard functions, or that raise errors during collection are excluded from the denominator. As a result, the 46.5% alignment figure describes the subset of protocols already amenable to standard balance queries, not all 939 protocols in the study. The authors should state this clearly and, if possible, report a conservative lower bound for the full population by treating excluded protocols as non-verified.
minor comments (5)
  1. [III-D] The text says the average Jaccard index 'ranges from 0.93% to 0.89%, 0.86% and 0.81%, respectively'; these are similarity values in [0,1], not percentages, so they should read 0.93, 0.89, 0.86, and 0.81.
  2. [Figure 4 caption] The caption refers to 'get_ethBalance' queries; the correct RPC method name used elsewhere is 'eth_getBalance'.
  3. [Figure 2] The figure's internal labels ('729 27 62 1 0 26 72') are visually confusing, especially the '1 0' and '26 72' groupings; a standard flowchart or tree layout with explicit edge labels would make the branching logic clearer.
  4. [Table I] The last column header 'Returns the . . .' is incomplete in the rendered text; each row should end with a complete phrase, e.g., 'Returns the account balance of another account (owner address).'
  5. [Appendix F] The appendix is titled 'TVL reconstruction from on-chain data' but the cross-reference in the main text and the internal narrative sometimes refer to 'Section F'; this should be harmonized to Appendix F.

Circularity Check

1 steps flagged · score 3.0 of 10

vTVL reuses DeFiLlama's own address/token selection, so 'independent verification' is partly a self-consistency check; the descriptive findings are unaffected.

  1. other [Section IV-A, Appendix F, footnote 2]
    "Having access to the functionalities and accounts queried on-chain through the DeFiLlama infrastructure, we can acquire a set of addresses per protocol containing deposited assets and provide further insights on the TVL associated with each protocol, solely relying on blockchain data (assuming that addresses called during computation contribute to the project TVL). ... While we assume that balances queried during computation directly contribute to the projects TVL, we acknowledge the possibility that they don't."

    vTVL is constructed by summing balances only for the addresses and tokens that DeFiLlama's own plugin queries during TVL computation, and the published DeFiLlama TVL is generated from those same plugin queries. The comparison is therefore partly a self-consistency check of the aggregator's address selection rather than an independent determination of which contracts actually hold locked value. If a plugin queries a treasury, an EOA, or a peripheral wrapper, or omits a vault, vTVL inherits that choice, so the 46.5% alignment rate can be read as plugin-replay fidelity.

full rationale

Most of the paper's contribution is not circular. The headline measurements—10.5% of protocols relying on external servers, 68 alternative balance functions, 240 duplicated balance queries, and the decline of non-standard calls over time—are direct observations of instrumented plugin executions and do not depend on any fitted parameter or self-citation. The vTVL case study also does not fit any parameter to the published DeFiLlama TVL: balances are read on-chain and prices are derived from Uniswap V2 liquidity, independently of the API values being compared. However, the vTVL address and token universe is taken from the very DeFiLlama plugin code that produces the published figures (Section IV-A and Appendix F), under the explicit assumption that queried addresses contribute to TVL. This makes the 46.5% alignment partly a measure of whether DeFiLlama's own computation can be replayed with standard balance queries, rather than a fully independent verification of asset lock-up. The paper discloses this assumption in footnote 2, and the independence of balance and price retrieval means the result is not equivalent to its input by construction. No load-bearing self-citation chain or imported uniqueness theorem is present. Overall, the circularity is limited to an overstatement of the word 'independent' in the central claim; the empirical findings and the reconstruction exercise retain substantial independent content.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The vTVL reconstruction rests on the assumption that addresses and tokens queried by DeFiLlama plugins exactly define each protocol's TVL, stated in Appendix F. The standard versus alternative function taxonomy treats only ERC-20 balanceOf and eth_getBalance as standard, despite totalAssets being standardized by ERC-4626. Price cleaning thresholds, the 10,000-point cutoff, manual removal of 80 tokens, and the ±0.5 alignment threshold are hand-chosen and not sensitivity-tested. No new physical entities are introduced; vTVL is a metric with on-chain falsifiability.

free parameters (5)
  • Alignment threshold for 'aligned' vTVL = discrepancy ratio within ±0.5
    Section IV-B: 46.5% of protocols are called aligned using ±0.5, while only 23.5% fall within ±0.05. This post hoc threshold sets the headline claim.
  • Price outlier removal threshold = 20% deviation from 10-day rolling average
    Appendix F: individual price values differing by more than 20% from a centered 10-day rolling average are removed. This threshold is chosen by the authors without sensitivity analysis.
  • Minimum price data points per token = 10,000 points
    Appendix F: tokens with fewer than 10,000 Uniswap V2 price points are excluded from pricing, affecting which tokens contribute to vTVL.
  • Manually removed high-volatility tokens = 80 tokens
    Appendix F: after manual sanity checks, 80 token price series with excessively high volatility are removed. This changes vTVL for protocols holding those tokens.
  • Stablecoin market cap cutoff = 20 million USD
    Appendix F: only stablecoins with more than $20m market capitalization and a USD peg are categorized as stablecoins. This affects token categorization and TVL composition.
assumptions (5)
  • domain assumption Addresses queried by the DeFiLlama plugin during TVL computation exactly constitute the protocol's TVL assets
    Appendix F and the Section III-E footnote state this assumption explicitly. It underpins all vTVL alignment results; if false, discrepancy ratios do not measure verifiability.
  • domain assumption Custom interceptors capture every relevant on-chain and off-chain interaction
    The pipeline uses mswjs/interceptors, but 62 projects were not computed, 28 raised errors, 22 produced no interaction, and older commits are noted as less reliable, so capture may be incomplete.
  • domain assumption Only ERC-20 balanceOf and eth_getBalance count as standard balance queries
    Section III-C and IV-A treat totalAssets and similar methods as alternative. However, totalAssets is standardized by ERC-4626 for vaults, so the standard versus alternative taxonomy is a modeling choice, not a protocol standard.
  • domain assumption DeFiLlama plugins on Ethereum are representative of TVL aggregation generally
    The measurement covers only Ethereum plugins from DeFiLlama (939 of 3494 projects), while the abstract and title speak of TVL in DeFi broadly. Generalization assumes other aggregators compute TVL similarly.
  • domain assumption Uniswap V2 prices plus Coingecko ETH/USD with cleaning thresholds yield accurate USD values
    Prices are obtained for only 942 of 12,246 tokens, 80 high-volatility tokens are manually removed, and stablecoin pegs are assumed. Low-liquidity tokens and NFTs may be mispriced or missing.
invented entities (1)
  • verifiable Total Value Locked (vTVL) independent evidence
    purpose: Metric measuring the portion of a protocol's reported TVL that can be reconstructed using only on-chain data and standard balance queries
    vTVL is an operational reconstruction protocol; anyone with an Ethereum archive node and the protocol's on-chain addresses can compute it. Its comparison with DeFiLlama's published TVL is a falsifiable check.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Verifiability of Total Value Locked (TVL) in Decentralized Finance." pith.science (2026). https://pith.science/paper/NVBUMOBZ

@misc{pith2026250514565,
  author       = {Pith},
  title        = {Pith review of: Towards Verifiability of Total Value Locked (TVL) in Decentralized Finance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NVBUMOBZ}},
  note         = {Machine review of arXiv:2505.14565}
}
read the original abstract

Total Value Locked (TVL) aims to measure the aggregate value of cryptoassets deposited in Decentralized Finance (DeFi) protocols. Although blockchain data is public, the way TVL is computed is not well understood. In practice, its calculation on major TVL aggregators relies on self-reports from community members and lacks standardization, making it difficult to verify published figures independently. We thus conduct a systematic study on 939 DeFi projects deployed in Ethereum. We study the methodologies used to compute TVL, examine factors hindering verifiability, and ultimately propose standardization attempts in the field. We find that 10.5% of the protocols rely on external servers; 68 methods alternative to standard balance queries exist, although their use decreased over time; and 240 equal balance queries are repeated on multiple protocols. These findings indicate limits to verifiability and transparency. We thus introduce ``verifiable Total Value Locked'' (vTVL), a metric measuring the TVL that can be verified relying solely on on-chain data and standard balance queries. A case study on 400 protocols shows that our estimations align with published figures for 46.5% of protocols. Informed by these findings, we discuss design guidelines that could facilitate a more verifiable, standardized, and explainable TVL computation.

Figures

Figures reproduced from arXiv: 2505.14565 by the authors.

Figure 1
Figure 1. Analysis pipeline. For each protocol, we execute the plugin code provided by the TVL aggregator and record all interactions with both on-chain and off-chain sources. Method Count Num. of protocols Returns the . . . balanceOf 102 655 641 . . . account balance of another account (owner address) [32]. eth_getBalance 1696 170 . . . ETH balance of the account of a given address [33]. totalSupply 2577 154 . . . total toke… view at source ↗
Figure 2
Figure 2. represents graphically the space of 939 DeFiLlama projects analyzed. For the largest group (N = 729), the proce￾dure is executed without errors and no external hosts are used. While 73 projects utilized both external hosts and on-chain data sources, for 26 protocols we observe interactions with external hosts but no direct on-chain data could be retrieved; since no errors occurred in their collection procedure, we i… view at source ↗
Figure 3
Figure 3. Pairwise Jaccard similarity, computed for each protocol on the set of balanceOf calls queried in the main and four older commits. Each line reports the values relative to one specific commit. Projects are ranked in descending order based on their similarity score (ranging from 0 to 1). A value of 1 indicates no changes among the compared sets. collection for various commits, we repeat the data gathering at four alte… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Verifiable Total Value Locked (vTVL). We query on-chain information for relevant DeFi protocols — Aave v2 and v3 (a), Compound v3 (b), dYdX (c), Maker (d), Uniswap v3 (e) — and compare it to published off-chain TVL data. Each plot represents the evolution in time of th…
Figure 5
Figure 5. Figure 5: Differences between vTVL and published TVL. Projects are dots ranked in descending order based on the Discrepancy ratio, i.e., the ratio of on-chain estimations to off-chain data, normalized by subtracting one. particular attention to derivative and non-derivative toke…
Figure 6
Figure 6. Figure 6: Jaccard similarity. Panel (a) reports the same infor￾mation of [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: TVL reconstruction for the Ethereum DeFi ecosys￾tem. The plot shows the evolution in time of the vTVL value for all protocols in the case study. To provide a broader picture of the ecosystem, in addition to the individual protocols described in Section IV-A, we show in…
Figure 8
Figure 8. Figure 8: Ratio R for different protocol categories. We compute R = T V Ra/T V L as the ratio between the value of assets used to measure TVR and the value of TVL, and compare it across protocol categories. Derivtives and staking protocols have the highest R values. Derivatives …
Figure 9
Figure 9. Figure 9: shows two panels conveying additional information on the ratio R, with protocols further divided by size (upper panel) and time (lower panel). More precisely, on top we divide protocols in small size (N = 117), medium size (N = 232), and large size (N = 63), respective…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 47 canonical work pages

  1. [1]

    The technology of decentralized finance (defi),

    R. Auer, B. Haslhofer, S. Kitzler, P. Saggese, and F. Victor, “The technology of decentralized finance (defi),”Digital Finance, vol. 6, no. 1, pp. 55–95, 2024

    work page 2024

  2. [2]

    GitHub repository - DeFiLlama-Adapters,

    DeFiLlama, “GitHub repository - DeFiLlama-Adapters,” available at: https://github.com/DefiLlama/DefiLlama-Adapters

    work page

  3. [3]

    Why we don’t accept APIs,

    ——, “Why we don’t accept APIs,” available at: https://github.com/D efiLlama/DefiLlama-Adapters/discussions/432

    work page

  4. [4]

    Master of anons: How a crypto developer faked a defi ecosystem,

    D. Nelson and T. Wang, “Master of anons: How a crypto developer faked a defi ecosystem,” https://www.coindesk.com/layer2/2022/08/04 /master-of-anons-how-a-crypto-developer-faked-a-defi-ecosystem/

    work page 2022

  5. [5]

    Understanding defi through the lens of a production-network model,

    J. Chiu, T. Koeppl, H. Yu, and S. Zhang, “Understanding defi through the lens of a production-network model,” Bank of Canada, Tech. Rep., 2023

    work page 2023

  6. [6]

    Data provider defillama de-emphasizes double-counted crypto deposits after saber revelation,

    Coindesk, “Data provider defillama de-emphasizes double-counted crypto deposits after saber revelation,” 2022, available at: https://ww w.coindesk.com/business/2022/08/05/data-provider-defillama-de-emp hasizes-double-counted-crypto-deposits-after-saber-revelation/

    work page 2022

  7. [7]

    Understanding total value locked (tvl),

    L. Nuzzi, A. L. Calvez, and K. Waters, “Understanding total value locked (tvl),” 2021, available at: https://coinmetrics.substack.com/p/coin-metri cs-state-of-the-network-0c0

    work page 2021

  8. [8]

    Piercing the veil of tvl: Defi reappraised,

    Y . Luo, Y . Feng, J. Xu, and P. Tasca, “Piercing the veil of tvl: Defi reappraised,” inInternational Conference on Financial Cryptography and Data Security. Springer, 2025

    work page 2025

Show all 51 references
  1. [9]

    Detecting token systems on ethereum,

    M. Fröwis, A. Fuchs, and R. Böhme, “Detecting token systems on ethereum,” inInternational Conference on Financial Cryptography and Data Security. Springer, 2019, pp. 93–112

    2019

  2. [10]

    Identification of token contracts on ethereum: standard compliance and beyond,

    M. Di Angelo and G. Salzer, “Identification of token contracts on ethereum: standard compliance and beyond,”International Journal of Data Science and Analytics, vol. 16, no. 3, pp. 333–352, 2023

    2023

  3. [11]

    Sok: A classification frame- work for stablecoin designs,

    A. Moin, K. Sekniqi, and E. G. Sirer, “Sok: A classification frame- work for stablecoin designs,” inInternational Conference on Financial Cryptography and Data Security. Springer, 2020, pp. 174–197

    2020

  4. [12]

    Sok: Decentralized exchanges (dex) with automated market maker (amm) protocols,

    J. Xu, K. Paruch, S. Cousaert, and Y . Feng, “Sok: Decentralized exchanges (dex) with automated market maker (amm) protocols,”ACM Computing Surveys, vol. 55, no. 11, pp. 1–50, 2023

    2023

  5. [13]

    Defi protocols for loanable funds: Interest rates, liquidity and market efficiency,

    L. Gudgeon, S. Werner, D. Perez, and W. J. Knottenbelt, “Defi protocols for loanable funds: Interest rates, liquidity and market efficiency,” in Proceedings of the 2nd ACM Conference on Advances in Financial Technologies, 2020, pp. 92–112

    2020

  6. [14]

    Sok: Yield aggregators in defi,

    S. Cousaert, J. Xu, and T. Matsui, “Sok: Yield aggregators in defi,” in 2022 IEEE International Conference on Blockchain and Cryptocurrency (ICBC). IEEE, 2022, pp. 1–14

    2022

  7. [15]

    Leverage staking with liquid staking derivatives (lsds): Opportunities and risks,

    X. Xiong, Z. Wang, X. Chen, W. Knottenbelt, and M. Huth, “Leverage staking with liquid staking derivatives (lsds): Opportunities and risks,” arXiv preprint arXiv:2401.08610, 2023

    2023 arXiv

  8. [16]

    A short survey on business models of decentralized finance (defi) protocols,

    T. A. Xu and J. Xu, “A short survey on business models of decentralized finance (defi) protocols,” inInternational Conference on Financial Cryptography and Data Security. Springer, 2022, pp. 197–206

    2022

  9. [17]

    DeFi Rankings,

    DappRadar, “DeFi Rankings,” available at: https://docs.dappradar.com /rankings/defi-rankings

  10. [18]

    Total Value Locked (TVL) in DeFi protocols,

    Stelareum, “Total Value Locked (TVL) in DeFi protocols,” available at: https://www.stelareum.io/en/defi-tvl.html

  11. [19]

    DeFiPulse Total Value Locked (TVL) Methodology,

    DeFiPulse, “DeFiPulse Total Value Locked (TVL) Methodology,” avail- able at: https://docs.defipulse.com/methodology/tvl

  12. [20]

    What Total Value Locked (Tvl) and Why Users Monitor This Metric,

    Coingecko, “What Total Value Locked (Tvl) and Why Users Monitor This Metric,” see: https://www.coingecko.com/learn/total-value-locked

  13. [21]

    DeFiLlama website,

    DeFiLlama, “DeFiLlama website,” available at: https://defillama.com/

  14. [22]

    Review of decentralized finance applications and their total value locked,

    V . Stepanova and I. Erin, š, “Review of decentralized finance applications and their total value locked,”TEM Journal, vol. 10, no. 1, p. 327, 2021

    2021

  15. [23]

    Return factors of ether cryptocurrency: On chain metrics and defi,

    K. Shilov and A. V . Zubarev, “Return factors of ether cryptocurrency: On chain metrics and defi,”Available at SSRN 4432586

  16. [24]

    Understanding digital bubbles amidst the covid-19 pandemic: Evidence from defi and nfts,

    Y . Maouchi, L. Charfeddine, and G. El Montasser, “Understanding digital bubbles amidst the covid-19 pandemic: Evidence from defi and nfts,”Finance Research Letters, vol. 47, p. 102584, 2022

    2022

  17. [25]

    Sok: Decentralized finance (defi) attacks,

    L. Zhou, X. Xiong, J. Ernstberger, S. Chaliasos, Z. Wang, Y . Wang, K. Qin, R. Wattenhofer, D. Song, and A. Gervais, “Sok: Decentralized finance (defi) attacks,” in2023 IEEE Symposium on Security and Privacy (SP). IEEE, 2023, pp. 2444–2461

    2023

  18. [26]

    The return of (i) defix,

    F. ¸ Soiman, G. Dumas, and S. Jimenez-Garces, “The return of (i) defix,” arXiv preprint arXiv:2204.00251, 2022

    2022 arXiv

  19. [27]

    Towards understanding governance tokens in liquidity mining: a case study of decentralized exchanges,

    S. Fan, T. Min, X. Wu, and C. Wei, “Towards understanding governance tokens in liquidity mining: a case study of decentralized exchanges,” World Wide Web, pp. 1–20, 2022

    2022

  20. [28]

    Decentralized finance (defi) projects: A study of key performance indicators in terms of defi protocols’ valuations,

    D. Metelski and J. Sobieraj, “Decentralized finance (defi) projects: A study of key performance indicators in terms of defi protocols’ valuations,”International Journal of Financial Studies, vol. 10, no. 4, p. 108, 2022

    2022

  21. [29]

    Decentralized finance: the possibilities of a blockchain “money lego

    T. Katona, “Decentralized finance: the possibilities of a blockchain “money lego” system,”Financial and Economic Review, vol. 20, no. 1, pp. 74–102, 2021

    2021

  22. [30]

    Where do defi stablecoins go? a closer look at what defi composability really means,

    K. Saengchoteet al., “Where do defi stablecoins go? a closer look at what defi composability really means,”Available at SSRN 3893487, 2021

    2021

  23. [31]

    Ethereum Signature Database,

    4byte, “Ethereum Signature Database,” see https://www.4byte.directory

  24. [32]

    ERC-20: Token Standard Improvement Proposal,

    Ethereum.org, “ERC-20: Token Standard Improvement Proposal,” see https://eips.ethereum.org/EIPS/eip-20

  25. [33]

    Ethereum JSON-RPC API,

    ——, “Ethereum JSON-RPC API,” see https://ethereum.org/en/develo pers/docs/apis/json-rpc/#eth_getbalance

  26. [34]

    Functionalities documentation,

    Uniswap, “Functionalities documentation,” see https://docs.uniswap.org /contracts/v2/reference/smart-contracts/pair

  27. [35]

    Factory contract documentation,

    ——, “Factory contract documentation,” see https://docs.uniswap.org/c ontracts/v2/reference/smart-contracts/factory

  28. [36]

    AaveV2Provider contract,

    Etherscan, “AaveV2Provider contract,” see %https://etherscan.io/addre ss/0x7ac6859e69d6549b39a8367097d7ae5fbff5951e#readContract

  29. [37]

    Vyper contract,

    ——, “Vyper contract,” see https://etherscan.io/address/0xba3cfea6514 cf5acddeff3167df0b7a4337751bc#code

  30. [38]

    Disentangling decentralized finance (defi) compositions,

    S. Kitzler, F. Victor, P. Saggese, and B. Haslhofer, “Disentangling decentralized finance (defi) compositions,”ACM Transactions on the Web, vol. 17, no. 2, pp. 1–26, 2023

    2023

  31. [39]

    Mea- surements, analyses, and insights on the entire ethereum blockchain network,

    X. T. Lee, A. Khan, S. Sen Gupta, Y . H. Ong, and X. Liu, “Mea- surements, analyses, and insights on the entire ethereum blockchain network,” inProceedings of The Web Conference, 2020, pp. 155–166

    2020

  32. [40]

    Behavior of liquidity providers in decentralized exchanges,

    L. Heimbach, Y . Wang, and R. Wattenhofer, “Behavior of liquidity providers in decentralized exchanges,”arXiv preprint arXiv:2105.13822, 2021

    2021 arXiv

  33. [41]

    DeFiLlama, “APIs,” available at: https://defillama.com/docs/api

  34. [42]

    Defi protocol risks: The paradox of defi,

    N. Carter and L. Jeng, “Defi protocol risks: The paradox of defi,” Regtech, suptech and beyond: innovation and technology in financial services” riskbooks–forthcoming, vol. 3, 2021

    2021

  35. [43]

    Stablecoin devaluation risk,

    B. Eichengreen, M. T Nguyen, and G. Viswanath-Natraj, “Stablecoin devaluation risk,”WBS Finance Group Research Paper, 2023, available at SSRN: http://dx.doi.org/10.2139/ssrn.4460515

    2023 doi

  36. [44]

    Assessing the solvency of virtual asset service providers: are current standards sufficient?

    P. Saggese, E. Segalla, M. Sigmund, B. Raunig, F. Zangerl, and B. Haslhofer, “Assessing the solvency of virtual asset service providers: are current standards sufficient?”Applied Economics, pp. 1–16, 2024

    2024

  37. [45]

    Defi risks and the decentralisation illusion,

    J. F. Doerr, A. Kosse, A. Khan, U. Lewrick, B. Mojon, B. Nolens, and T. Rice, “Defi risks and the decentralisation illusion,”BIS Quarterly Review, vol. 21, 2021

    2021

  38. [46]

    Decentralized crypto governance? transparency and concentration in ethereum decision-making,

    C. Fracassi, M. Khoja, and F. Schär, “Decentralized crypto governance? transparency and concentration in ethereum decision-making,”Trans- parency and Concentration in Ethereum Decision-Making, 2024

    2024

  39. [47]

    The governance of decentralized autonomous organizations: A study of contributors’ influence, networks, and shifts in voting power,

    S. Kitzler, S. Balietti, P. Saggese, B. Haslhofer, and M. Strohmaier, “The governance of decentralized autonomous organizations: A study of contributors’ influence, networks, and shifts in voting power,” in International Conference on Financial Cryptography and Data Security. ...

    2024

  40. [48]

    Scale-free networks are rare,

    A. D. Broido and A. Clauset, “Scale-free networks are rare,”Nature communications, vol. 10, no. 1, pp. 1–10, 2019

    2019

  41. [49]

    Power-law distributions in empirical data,

    A. Clauset, C. R. Shalizi, and M. E. Newman, “Power-law distributions in empirical data,”SIAM review, vol. 51, no. 4, pp. 661–703, 2009

    2009

  42. [50]

    Top Crypto Categories By Market Cap,

    Coingecko, “Top Crypto Categories By Market Cap,” available at: https: //www.coingecko.com/en/categories

  43. [51]

    Cryptocurrency Sectors by 24h Price Change,

    Coinmarketcap, “Cryptocurrency Sectors by 24h Price Change,” avail- able at: https://coinmarketcap.com/cryptocurrency-category/. 9 APPENDIX A. Implementation Details To investigate how TVL is technically computed in practice, we devised a data extraction pipeline that captures...

    2021

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.