Source-linked AI summary
Dissecting Ponzi schemes on Ethereum: identification, analysis, and impact
Massimo Bartoletti, Salvatore Carta, Tiziana Cimoli, Roberto Saia
TL;DR
Ponzi schemes on Ethereum combine dependence on new investors with smart contracts that can appear trustworthy while enabling anonymous, persistent fraud. The paper surveys these schemes through a constructed dataset and analyses their code, transactions, impact, users, and temporal behaviour. It finds recurring vulnerabilities and characteristic transaction patterns, while noting that the overall economic impact was limited and scammer profits cannot be precisely quantified.
Problem
The paper addresses the need to understand how many victims and how much money Ethereum Ponzi schemes involve, how they evolve, and whether their execution reveals recognizable fingerprints.
Method
The authors conduct a comprehensive Ethereum survey using a constructed Ponzi dataset, source-code analysis, transaction extraction, and analyses of impact, users, time, and payment inequality.
Results
Ethereum Ponzi experiments showed recurring vulnerabilities, while transaction logs commonly exhibited short lifespans, highly unequal payouts, and few users receiving more money than they contributed.
Takeaways & Limitations
Transaction-log patterns can support analysis of potential Ponzi schemes, although no single listed feature alone appears sufficient to distinguish them from other contracts.
Takeaways & Limitations
The study cannot precisely quantify scammers’ profits because scammer identities and addresses cannot be reliably separated from legitimate users’ money flows.
Abstract
from arXiv · showhide
Ponzi schemes are financial frauds which lure users under the promise of high profits. Actually, users are repaid only with the investments of new users joining the scheme: consequently, a Ponzi scheme implodes soon after users stop joining it. Originated in the offline world 150 years ago, Ponzi schemes have since then migrated to the digital world, approaching first the Web, and more recently hanging over cryptocurrencies like Bitcoin. Smart contract platforms like Ethereum have provided a new opportunity for scammers, who have now the possibility of creating "trustworthy" frauds that still make users lose money, but at least are guaranteed to execute "correctly". We present a comprehensive survey of Ponzi schemes on Ethereum, analysing their behaviour and their impact from various viewpoints.
1. Introduction
Ethereum smart contracts create new opportunities for Ponzi schemes by combining anonymity, persistence, and code-enforced execution with a false sense of trustworthiness. This paper surveys Ethereum Ponzis through a dataset and analyses their code, economic impact, user gains and losses, temporal behaviour, payment inequality, and user-protection guidelines.
- Motivation: Smart contracts can keep Ponzi initiators anonymous and make schemes difficult for authorities to terminate or reverse.Permissionless blockchains reinforce contract persistence and decentralization.
- Motivation: Public, immutable code and automatic execution may give investors a false sense that the scheme is trustworthy and fair.Investors may believe owners cannot misuse their money and that the scheme will run indefinitely.
- Scope and contributions: The paper is presented as the first comprehensive survey of Ponzi schemes on Ethereum, combining a dataset with analyses from multiple perspectives.Its stated goal is to support understanding and intervention against these frauds.
- Scope and contributions: The authors define schemes using contract logic, construct a public dataset, and provide an open-source tool for extracting scheme transactions and money movements.The dataset expands from 138 source-code-identified schemes to 184 using bytecode similarity and manual inspection.
- Scope and contributions: Source-code analysis finds recurring patterns, unfair payout distributions, and security vulnerabilities that adversaries could exploit to steal money.The paper also develops a rough taxonomy based on payout redistribution patterns.
- Scope and contributions: The survey measures economic impact, user gains and losses, temporal dynamics, payment inequality, and guidelines for protecting users.Its user analysis commonly finds a few large winners and a majority of losing users.
2. Ethereum in a nutshell
Ethereum is a decentralized platform where users and contracts hold and transfer ether through programs called smart contracts. The section illustrates these concepts with a wallet contract whose functions, storage, balance, and execution rules govern money transfers.
- Ethereum architecture: Ethereum executes Turing-complete programs called contracts, which have permanent storage and functions callable by users or other contracts.Contracts and users can own, send, and receive ether.
- Transactions: User transactions can create contracts, invoke contract functions, or transfer ether, while contract-triggered internal transactions can also affect balances.External transactions are recorded on the public blockchain.
- Transactions: A decentralized network of nodes processes transactions without a trusted central authority and uses consensus to address mismatches.The described consensus protocol is based on proof-of-work.
- Wallet example: AWallet models a personal wallet whose owner can pay recipients, while hashtables record outgoing and incoming addresses and amounts.Ether received by the contract remains in its balance, and incoming ether triggers a fallback function.
- Wallet example: The wallet’s pay function sends a specified amount of wei to a recipient, with exceptions reverting side effects when authorization or invocation conditions fail.One wei equals 10^-18 ETH, and an invalid caller loses the execution fee even when transferred ether is returned.
3. Collection of Ponzi schemes
The paper classifies Ethereum Ponzi schemes using four requirements applied exclusively to contract logic, then builds a dataset through verified source-code inspection and bytecode-similarity searches. The resulting public collection contains 184 schemes, while explicitly excluding some related or difficult-to-classify contracts.
- Classification criteria: The classification requires a contract to distribute money to investors according to some internal logic.This requirement alone also covers non-Ponzi mechanisms such as gambling, lotteries, insurance, and bonds.
- Classification criteria: The money distributed to investors must come only from investors rather than external sources.This excludes mechanisms funded by institutions such as banks or bookmakers.
- Classification criteria: Each investor must profit when new investors continue sending money to the contract, so profits arise through later investments.This distinguishes the criterion from gambling, betting, and lotteries, where profit is not guaranteed.
- Classification criteria: The risk of losing an investment must increase with the time at which the investor joins.This criterion captures the collapse risk associated with difficulty recruiting new investors.
- Dataset construction: The authors inspect verified Solidity contracts manually, identify 138 schemes, and expand the collection using bytecode similarity searches.The similarity procedure uses normalized Levenshtein distance and a threshold below 0.35, compared with an estimated arbitrary-contract distance of 0.79.
- Dataset construction: The final public dataset contains 184 Ponzi schemes, but it does not include every Ponzi scheme deployed on Ethereum.The collection is therefore a studied dataset rather than a complete census.
4. Anatomy of Ponzi schemes
Ethereum Ponzi schemes use distinct user-ordering and payout structures, but all depend on later investments to fund earlier returns and expose participants to increasing loss risk. Their implementations also contain bugs and owner-controlled functions that can harm investors or disable the scheme.
- Taxonomy of Ponzi schemes: Tree-shaped schemes route each user’s investment among inviter ancestors, while user gains depend on recruiting descendants and the owner alone is guaranteed profit.The amount shared decreases with ancestor distance, and users cannot predict returns from their own entry alone.
- Money redistribution: Users in doubling schemes may wait for exponentially expanding groups of later investors, while unusually large investments, fees, or multipliers can further delay payouts.Tree-shaped schemes are slowed because new users may not all become descendants of a given node.
- Taxonomy of Ponzi schemes: Handover schemes pay the former user after one new user joins, but rising entry tolls make later participation increasingly unattractive and risky.The toll increases as the scheme progresses, so later users are more likely to lose their money.
- Money redistribution: Across identified scheme types, investors earn only if sufficient later money arrives, while late investors face greater risk of losing their investment.This dependence is the common behavioral pattern captured by requirements R3 and R4.
- Security issues: Implementation vulnerabilities include failed refund handling, escalating fees that can exceed deposits, payout-iteration errors, owner-controlled parameter changes or withdrawals, and self-destruction.These defects can leave users unpaid, concentrate payments on one user, drain investor funds, or terminate the contract.
- Security issues: Scheme-level failures include denial-of-service or blackmail through reverted payouts, and gas limits can leave a legitimate jackpot winner unable to receive payment.HYIP can be attacked through send failures, while Government15 became stuck when clearing its user array exceeded the transaction gas limit.
5. Impact of Ponzi schemes
The study measures Ethereum Ponzi schemes through transaction flows, user participation, and transferred value, finding substantial economic impact across 184 contracts. Most schemes are chain-shaped, and their implementations show low variation within categories.
- Measurement: 184 contracts were analyzed using incoming and outgoing transactions, transferred ETH and USD values, and counts of paying and paid users.External transactions generally contribute to inflow, while payouts from the contract contribute to outflow.
- Measurement limits: Scammer profit cannot be precisely quantified because the scammer's addresses and payments to legitimate users cannot be reliably separated.The study therefore uses total scheme inflow as a rough over-approximation of scammer profit.
- Economic impact: 630662 USD was collected from 2378 distinct users across the studied Ponzi schemes.ETH values alone are less informative because the ETH/USD exchange rate was highly volatile.
- Scheme taxonomy: 151 of 184 contracts were chain-shaped, while 12 belonged to waterfall, tree-based, or handover kinds and 21 were uncategorized.The uncategorized contracts were mostly experiments, variants, or singular cases.
- Implementation similarity: Contracts within the same category differed only slightly, commonly in multiplication factors, fees, or auxiliary owner functions.Their average normalized bytecode Levenshtein distance was 0.54, below the 0.79 average distance for the comparison set.
6. Measuring gains and losses
The gain analysis finds the characteristic Ponzi pattern: a small number of users receive substantial returns while most users recover little or none of their investment. This pattern persists across contracts despite differences in scheme type, user count, and transaction volume.
- Distribution of outcomes: The gain diagrams show that contracts generally redistributed nearly all received money while concentrating gains among very few users.For Doubler2, the gain graph's integral was close to zero despite a peak gain of 486 ETH.
- Doubler2: 486 ETH was the highest Doubler2 gain, earned by a user who invested only 1 ETH.Other high-gain users had ratios below 2 and generally invested large amounts or invested more than once.
- Etheramid1: Etheramid1 showed a similar pattern: most users lost money, a few gained up to 5 ETH, and exactly one user received 30 ETH.One user received 7 ETH after investing 1 ETH, while another received 30.6 ETH without investing.
- Cross-contract pattern: Across the selected contracts, users who received nothing or less than their investment were the most numerous, with non-gaining users averaging around 70%.Doubler had the highest reported non-gaining share at 88%, followed by ShinySquirrels at 87%.
- Exceptional gains: One or two users per contract typically earned exceptionally high revenues, often without investing, and these gains generally belonged to owners through contract fees.EthereumPyramid had two owners and Rubixi had six; Rubixi's case was described as singular because of a code bug.
7. Evolution over time
Ponzi schemes typically concentrate activity in short bursts, with inflows and outflows closely linked according to the payout structure. Across the collection, activity and scheme creation peaked in early 2016 before declining.
- 7.1. Inflow and outflow correlation: Most contracts show strong inflow–outflow correlation, but chain-shaped schemes delay payouts whereas tree-shaped schemes distribute investments immediately.Doubler2 aggregates smaller inflows before paying one user, while Etheramid distributes new investments among users as they arrive.
- 7.2. Lifetime: Most contracts have a short lifespan marked by an intense activity peak followed by stagnation.The authors note that isolated later transactions and inactivity can make the final transaction a coarse measure of actual activity.
- 7.2. Lifetime: ∼60% of Ponzi schemes have lifetimes close to 0 days, indicating deployment without attracting users.Lifetime is measured from the first to the last inflow or outflow transaction.
- 7.2. Lifetime: 91 new public Ponzi schemes were created in April 2016, after which creation fell to an average of ∼3 per month.The paper conjectures that the decline may reflect schemes becoming harder to classify rather than disappearing.
- 7.3. Volume of payments: Most value was exchanged from February to May 2016, with three transaction-volume peaks between March and April.The authors consider the later activity decline plausibly related to the contemporaneous drop in scheme creation.
- 7.3. Volume of payments: Payment timing reflects scheme mechanics: DynamicPyramid nearly stopped after one high-volume day, while Etheramid showed near-perfect inflow–outflow overlap.DynamicPyramid received almost 60,000 USD on 11 March; Etheramid immediately forwarded incoming funds to users’ ancestors.
8. Measuring payment inequality
Payment distributions are generally unequal, with payouts more concentrated than investments. Some schemes strongly favor large investors or recipients, while others are closer to equal investment distributions.
- Lorenz curves: Lorenz curves compare cumulative users with cumulative payments, using the 45-degree line as perfect equality.Curves nearer the perfect-inequality boundary indicate stronger concentration among fewer users.
- Investment inequality: 10% of victims invested more than 90% of the money in the most unbalanced schemes, Government and ProtectTheCastle.Etheramid1 was comparatively close to perfect equality for investments.
- Payout inequality: Payout distributions are generally more unequal than investment distributions, including for Etheramid1 despite its balanced investments.The payout Lorenz curves are more compressed toward the right than the investment curves.
- Gini coefficients: DynamicPyramid exceeds 80% inequality for both investments and payouts and is identified as the most lucrative scheme.Gini coefficients relate investment and payout inequality to the total amounts invested and received.
9. Conclusions
The survey finds that Ethereum enabled numerous smart-contract Ponzi experiments, though their overall economic impact remained limited. It recommends examining advertisements, contract code, and transaction logs, while extending analysis to harder-to-classify crypto scams.
- Conclusions: During Ethereum’s first three years, many smart-contract Ponzi experiments appeared, but they represented only a small fraction of blockchain transactions and value.The paper frames these schemes as opportunities for fraud enabled by blockchain anonymity and decentralization.
- Recommendation #1: check the advertisements: Advertisements often promise high returns while omitting risks, payout discrepancies, and owner powers to alter conditions or destroy contracts.The paper recommends checking advertisements and consulting scam blacklists before sending money.
- Recommendation #2: analyse the contract code: Contract-code analysis revealed vulnerabilities caused by poor programming or possibly intentional design, undermining claims of trustworthiness.The collection’s contracts were generally small, with most containing fewer than 100 Solidity lines.
- Recommendation #3: analyse the transaction logs: Transaction logs show that few users receive more than they invest, lifetimes are usually short, and payout Gini coefficients often exceed 80%.These features alone may not distinguish Ponzis, but together they can support automated classifiers.
- Future works: Future research should study ICOs and crypto-collectibles as possible “pseudo” Ponzis that share characteristics with, but do not satisfy, the paper’s pure-Ponzi requirements.The paper notes that some ICOs were unmasked as Ponzi schemes by financial authorities.
Appendix A. Appendix: general charts
Appendix charts provide per-contract views of user gains, gain-ratio groups, inflow and outflow timing, high-gain users, and users who received no net benefit.
- User gains: Figures A.24–A.26 plot each user’s gain in ether per contract after ordering users by increasing gain.These charts support contract-level inspection of how gains are distributed across users.
- Gain-ratio groups: Figures A.27–A.29 group users by gain ratio, including users receiving nothing, ratios from 0 through higher ranges, infinite ratios, and other cases.An infinite ratio denotes users who received money without sending anything.
- Transaction timing: Figures A.30–A.32 show inflow and outflow timing, with transaction date on the x-axis and ether amount on the y-axis.Blue denotes inflow and red denotes outflow.
- High-gain users: Figure A.33 lists each contract, user investment, receipt, net difference, and gain ratio for users with gain ratio > 10, infinite ratio, or net gain > 100.Values are truncated to the first non-zero decimal digit.
- Users without gains: Figure A.34 reports, per contract, the percentage of participating users who gained nothing.Its second column counts entrants, while its third counts entrants with gain ratio less than or equal to 1.