BLOCKCHAIN TRANSACTION ORDERING SSRN-id4187752
BLOCKCHAIN TRANSACTION ORDERING SSRN-id4187752
BLOCKCHAIN TRANSACTION ORDERING SSRN-id4187752
In so doing, the Article confronts how basic notions of fairness and trust
play out differently in a world of discretionary transaction ordering in crypto
markets compared to the first-come first-serve world of traditional finance.
Behaviors that might seem outrageous off-chain look very different when
examined in light of how blockchains actually work.
Nonetheless, this Article argues that some forms of MEV extraction entail
a significant risk of market manipulation liability. Focusing on sandwiching in
particular, we provide novel arguments showing that there is a route for courts
that adopt a moralized lens, focused on behavior that exploits privileged control
over financial infrastructure, to find sandwiching impermissibly manipulative.
We argue, further, that the legal hazards are even greater when it comes to
sandwiching private transactions, which more clearly involves a heightened
trust relationship, as well as disruptive schemes like oracle manipulation,
wherein MEV is part of an independently manipulative strategy. Nonetheless,
we argue, this alone does not mean a sweeping ban on these forms of MEV is
necessarily a desirable policy. It remains unclear whether a strict ban on MEV
sandwiching, for instance, would be prudent, given the unknowns about the net
effects of MEV extraction and behavioral impact that a ban on MEV
sandwiching would entail.
I. INTRODUCTION ....................................................................................... 3
II. WHAT IS MEV EXTRACTION? TECHNICAL AND ECONOMIC
BACKGROUND ............................................................................................. 10
A. The Journey of an Ethereum Transaction ................................ 11
1. From wallet software to “the mempool”............................. 11
2. Block production................................................................. 12
B. What is MEV? ........................................................................... 13
C. The MEV Extraction Ecosystem ............................................... 15
D. The Importance of Order Flow: Publicness vs. Exclusivity of
Ethereum Transactions............................................................. 16
E. Strategies for Generating Profit through MEV ........................ 18
F. The Techniques for Executing MEV Strategies ........................ 20
III. THE LAW OF MARKET MANIPULATION ................................................ 21
A. Scope of Legal Analysis ............................................................ 21
B. Anti-Market Manipulation Rules .............................................. 22
1. Price Manipulation ............................................................... 23
2. Fraud-Based Manipulation ................................................... 26
C. Insider Trading ......................................................................... 31
D. Front-running ........................................................................... 33
IV. LEGALITY OF MEV EXTRACTION......................................................... 34
A. Sandwiching Public Transactions ............................................ 35
1. Price Manipulation .............................................................. 36
2. Fraud-Based Manipulation.................................................. 37
(i) First simple theory: the sandwiched user is at least
recklessly misled .......................................................... 39
(ii) Second simple theory: the market is at least recklessly
misled ........................................................................... 41
(iii) More sophisticated theory: artificial effect on prices
created at least recklessly ............................................. 44
(iv) Rejoinders .................................................................... 49
(v) Front-running ............................................................... 53
B. Sandwiching Private Transactions ........................................... 54
1. Explicit Private Order Flow ................................................ 55
2. Non-Explicit Private Order Flow and Payment For Order
Flow .................................................................................... 60
C. Other Ways to Extract MEV: Oracle Manipulation ................. 62
1. Oracle manipulation to create loan liquidation opportunities
63
2. Fraud-based manipulation liability ..................................... 64
V. A NOTE OF CAUTION ON THE POLICY QUESTION ................................. 65
VI. PROPOSALS AND CONCLUSIONS ........................................................... 68
I. INTRODUCTION
Applying the assumptions of traditional finance to the radically different
infrastructure of crypto markets is risky business. On February 16, 2022, an
Ethereum user we will call 0x61 (based on her address) appears to have noticed
a profit opportunity. Since the beginning of February, the price of ETH1 (the
native token of the Ethereum blockchain) had recovered somewhat from a prior
drop, and 0x61 likely decided that the recovery wouldn’t last much longer. 0x61
used the decentralized exchange Uniswap V2 to sell 79 ETH, then worth around
$250,000, in exchange for DAI (roughly put, 1 DAI corresponds to 1 dollar),
and logged off perhaps feeling content about the day’s work.2 Little did 0x61
1 Strictly speaking, this was Wrapped Ethereum (WETH). See, e.g., Ivan Cryptoslav, What Is
https://etherscan.io/tx/0x9760b7dedcbfbc37e6feb491a9bdf33c98c99d8d339fde49f2c3e97828cd4
b6b. The transaction can also be analyzed on ZeroMEV:
https://www.zeromev.org/block?txh=0x9760b7dedcbfbc37e6feb491a9bdf33c98c99d8d339fde49f
2c3e97828cd4b6b.
know that others were watching in the “dark forest”3 of Ethereum’s public
“mempool,”4 where submitted transactions sit waiting to be executed by being
built into blocks and recorded on the blockchain. It is only once the submitted
transaction is executed in this way that 0x61’s effort to capitalize on the profit
opportunity she saw would be complete.
However, things did not turn out exactly as 0x61 may have expected. A
few minutes after pressing the button on her transaction, it became clear that
someone else had interfered in her plans. The transaction she likely had thought
would provide her about $225,000 (adjusted for $25,000 in transaction fees) in
DAI, ended up garnering just $179,000.5
Where did the extra $46,000 that 0x61 expected go?6 It had been captured
by what has become known in the crypto community as a MEV extractor.7
“MEV” stands for Maximal Extractable Value8 and paradigmatically involves
the direct or indirect exploitation of the ability to control the order in which
transactions are executed, a power which is possessed by those who construct
the blockchain on which crypto assets are traded.9 Despite its very recent
formalization in 2019,10 MEV extraction has quickly become a major market
phenomenon, with a conservative estimate at roughly $600 million in profit
between 2020 and 2022 alone, while affecting tens of billions of dollars in
transaction value.11
More specifically, what happened to 0x61 was a sandwich,12 the classic
example of MEV extraction. Upon spying her transaction pending in the
3 Dan Robinson & Georgios Konstantopoulos, Ethereum is a Dark Forest, PARADIGM.XYZ (2020),
https://www.paradigm.xyz/2020/08/ethereum-is-a-dark-forest.
4 See infra Section II.A.
5 See supra note 2.
6 Had 0x61 not been front-run by the bot’s trade, they would have ended with $46,083-worth more
of DAI. For the source of that calculation see the Ethereum block 14217123 in the Zeromev
explorer: https://www.zeromev.org/block?num=14217123.
7 Maximal Extractable Value, https://ethereum.org/en/developers/docs/mev/ (last accessed, 31 Jan.,
2023).
8 Id.
9 Philip Daian et al., Flash boys 2.0: Frontrunning, transaction reordering, and consensus
[sic] from the Salt Mines: Ethereum Miners Extracting Value, (2022),
https://arxiv.org/abs/2203.15930.
13 See supra note 2.
14 For the purpose of clarity, we assume here that the searcher bot was operated by a validator-
proposer, meaning that it would net all 18 ETH as its own profit. Yet, as we will soon discuss, in
Section II.C., if the searcher were not operated by a proposer, it would have to give some of those
profits up to validators in exchange for the ability to have its own transactions ordered directly
before and directly after that of 0x61.
15 Ari Juels, Ittay Eyal & Mahimna Kelkar, Miners, Front-Running-as-a-Service Is Theft,
16 We focus our analysis in this Article on Ethereum, as this is where MEV extraction is most
prevalent and most discussed. With this said, much of our analysis is likely to apply to other public,
smart contract-enabling blockchains (e.g., Solana).
17 See infra Section II.
18 There have been valuable analyses of how securities law applies to fraud and other forms of
abuse in crypto markets in general, although they do not discuss MEV techniques like sandwiching
or oracle manipulation, which we confront in depth. See, e.g., Menesh Patel, Fraud on the Crypto
Market, FORTHCOMING IN HARVARD J. L. & TECH. (2023). Moreover, there have also been quick
comments on the legality of MEV, though these only scratch the surface and do not systematically
examine the depth of technical and legal nuance that these issues present; see, e.g., Auer et al.,
supra note 11; Mikołaj Barczentewicz, MEV on Ethereum: A Policy Analysis, INTERNATIONAL
CENTER FOR LAW & ECONOMICS WHITE PAPER 2023-01-23 (2023),
https://ssrn.com/abstract=4332703.
19 For example, MEV is mentioned in the crypto-regulation bill introduced by Senator Elizabeth
Warren: Digital Assets Anti-Money Laundering Act of 2022, S.5267, 117th Cong. §3(a) (2022).
20 See, e.g., Kristin N Johnson, Decentralized finance: Regulating cryptocurrency exchanges, 62
WM. & MARY L. REV. 1911 (2021); Chris Brummer, Disclosure, Dapps and DeFi, 5.2 STAN. J. OF
BLOCKCHAIN L. & POL’Y 137 (2022); Dirk A Zetzsche, Douglas W Arner & Ross P Buckley,
Decentralized finance, 6 J. OF FIN. REG. 172 (2020); Proceedings of the 2021 Spring Conference:
The Impact of Blockchain on the Practice of Law, 17 NYU J. OF L. & BUS. 681 (2021).
21 Gina-Gail S Fletcher, Legitimate yet manipulative: The conundrum of open-market manipulation,
68 DUKE L. J. 479 (2018); Tom CW Lin, The new market manipulation, 66 EMORY L. J. 1253
(2017); Jakob Arnoldi, Computer algorithms, market manipulation and the institutionalization of
high frequency trading, 33 THEORY CULT. SOC. 29 (2016); MERRITT B FOX, LAWRENCE GLOSTEN
& GABRIEL RAUTERBERG, THE NEW STOCK MARKET: LAW, ECONOMICS, AND POLICY (2019).
they can extract rents) of ordering and including transactions. 22 Thus, the central
question we confront in this Article is whether the sort of rent extraction MEV
arguably involves, flowing from privileged control over or access to key
financial infrastructure in which users place their trust, might make it count as a
form of manipulation.
We argue in Section IV that there are narrow contexts in which MEV
extractors plausibly can be seen as occupying a position of heightened trust and
thus we contend that existing anti-manipulation law would be violated by some
forms of MEV. But especially because of the differences from traditional
finance, we also find that much MEV extraction – especially when carried out
through open market trades at arms’ length – fit within familiar economic
rationales that the courts widely accept as legitimate within legacy finance.
Furthermore, we argue in Section V, at a policy level, that because of the open
empirical questions about whether MEV behavior has a net positive impact on
the efficiency of crypto markets (e.g., through properly incentivizing essential
financial functions on which all crypto market participants rely), regulators and
lawmakers should be very careful about introducing blanket MEV prohibitions.
We argue that prohibitions are likely to be defensible, if at all, only in narrowly
circumscribed cases such as those involving i) express representations that are
false or misleading,23 ii) a special relationship of trust between MEV extractors
and particular traders (such as when private order flow is involved),24 or iii)
harmful kinds of benchmark manipulation (known as “oracle manipulation,”25
and which DOJ, CFTC, and SEC are already prosecuting26).27
MEV extraction thus forces us to re-examine the normative assumptions
and principles underlying market manipulation law in order to determine
22 For instance, blockchain users rely on validators to maintain the integrity of the network by
ensuring that “double spending” does not occur. See, e.g., Satoshi Nakamoto, Bitcoin: A Peer-to-
Peer Electronic Cash System, (2008), https://bitcoin.org/bitcoin.pdf.
23 See generally infra Section IV.
24 See infra Section IV.B.
25 See infra Section IV.C; an “oracle” is essentially a communication channel which provides
external data to closed blockchain systems. While this data can come from on-chain or off-chain
sources, price oracles in decentralized finance– the focus of our discussion here – are a specific
kind of oracle which pull data exclusively from on-chain sources to determine prices for DeFi
protocols. For more information regarding oracles see Cryptopedia, Blockchain Oracles Explained:
Decentralized Oracles in DeFi (Feb. 4, 2022), https://www.gemini.com/cryptopedia/crypto-oracle-
blockchain-overview#section-inbound-versus-outbound-oracles (discussing oracles which bring
off-chain data to blockchain systems).
26 Press release, CFTC Charges Avraham Eisenberg with Manipulative and Deceptive Scheme to
Misappropriate Over $110 million from Mango Markets, a Digital Asset Exchange,
COMMODITY FUTURES TRADING COMMISSION (Jan. 9, 2023),
https://www.cftc.gov/PressRoom/PressReleases/8647-23; Press release, SEC Charges Avraham
Eisenberg with Manipulating Mango Markets’ “Governance Token” to Steal $116 Million of
Crypto Assets, SECURITIES AND EXCHANGE COMMISSION (Jan. 20, 2023),
https://www.sec.gov/news/press-release/2023-13.
27 See infra Section IV.C.
whether the effects of MEV extraction are legitimate and compatible with
market integrity or whether they fall afoul of basic notions of fairness and well-
ordered markets. In our analysis, the key ends up being to determine the extent
to which various types of MEV extractors, involved in different types of trading
strategies and techniques, can be said to occupy positions of trust that carry
special responsibilities to avoid interfering with the reasonable expectations of
other market participants.28 A core contribution of this Article thus is to frame
the central legal and normative questions to be answered – whether by courts,
regulator or legislators – and we argue for an approach to this question that takes
a holistic look at the costs and benefits of MEV extraction and allocates legal
duties on that basis.29
The remainder of this Article will proceed as follows. Section II introduces
the technical background necessary for a legal understanding of MEV. We
explain the process of transaction validation on Ethereum and show how it
allows MEV extraction. We describe the most common forms of MEV
extraction, including sandwiching, arbitrage, and liquidations. Section III
introduces relevant aspects of US law governing market manipulation in
securities and commodities markets. Beyond sketching the statutory and
regulatory backdrop of anti-manipulation law, we discuss the law’s treatment of
particularly salient categories of manipulative trading like open market
manipulation, insider trading, and front-running.
In Section IV, we apply the law of market manipulation to core cases of
MEV extraction. We begin with MEV sandwich trades targeting public
transactions, addressing the main arguments for and against market
manipulation liability for this practice. We provide novel arguments showing
that a court drawing on moralized conceptions of market fairness may have a
route to concluding that the sandwiching of public transactions is manipulative
in violation of CFTC Rule 180.1 or SEC Rule 10b-5 on the grounds that a
sandwicher at least recklessly creates an artificial price effect by unfairly
exploiting their position of privilege and control over essential financial
infrastructure. That said, we note that the practical hurdles for succeeding with
this cause of action make it unlikely to be pursued as a high regulatory priority
in the near term. Turning then to sandwiching of private transactions, we find
that a broader scope for market manipulation liability exists here as actors trusted
to act in confidence are more likely to end up behaving in misleading ways when
handling private transaction information. Finally, we examine trading strategies
that involve MEV only incidentally or as a means to further increase the
profitability of other schemes, including harmful oracle manipulation. As we
will argue, a number of these techniques are likely to attract market manipulation
liability.
Section V then moves from analyzing liability for MEV under existing law
to considering the policy question of how to respond to such practices. We
caution that it remains far from clear that a flat ban on questionable forms of
MEV sandwiching would be good policy. There are at present too many
unknowns about the net impact on market efficiency and social welfare of MEV
sandwiching, as well as the behavioral effects that a sweeping sandwiching
prohibition would have, to be confident that this is a desirable way forward at
present. More empirical research on the issue is imperative. We conclude in
Section VI by offering four recommendations to policymakers, courts,
researchers, and the Ethereum community. In these ways, we hope to shed light
on the dark art of MEV extraction on Ethereum.
CRYPTOLOGY EPRINT ARCHIVE (2022); Piet, Fairoze, and Weaver, supra note 12.
33 A “block” is an ordered batch of transactions which is added to a blockchain. Ethereum
group/extractable-value-7b0d4356a843.
36 Our description omits a large amount of technical detail not immediately required to understand
the mechanisms of MEV extraction. For more technical detail see, e.g., Daian et al., supra note 9;
Zhou et al., supra note 32.
2. Block production
37 “RPC” stands for “remote procedure call.” See, e.g., JSON-RPC API, Ethereum.org,
https://ethereum.org/en/developers/docs/apis/json-rpc/.
38 See, e.g., Lucianna Kiffer et al., Under the hood of the ethereum gossip protocol, in FINANCIAL
users participate in the validation process (as validators) of Proof of Stake (PoS) blockchains like
divided in 12-second “slots,” and in every slot one specific proposer has the
power to tell the network what the next block should be.43 If she decides to
construct the block on her own, then she assembles as many transactions from
the mempool as will “fit” in the block, maybe adding some of her own
transactions.
However, the proposer has an option of outsourcing the block-building
process to specialist block builders. Those block builders also monitor the
mempool and also can submit any transactions they want from outside the
mempool. Typically, block builders do not have direct relations with
proposers—both sides rely on an intermediary, a “relay.”44 The relay process,
popularized by the Flashbots organization,45 involves two auctions. First, block
builders compete for their block to be chosen by a relay as the one block the
relay will submit to the second auction. Second, relays compete for their block
to be chosen by the proposer, so that the proposer declares that this block is to
be added to the blockchain. In both auctions, to win one must offer the highest
fee.46
Whoever produces a block can decide not only which transactions will be
included, but also in what order will they be executed.47 Both of those things
can be valuable, because many profitable opportunities – like the example of
sandwiching introduced at the beginning of this Article – depend on the order of
transactions.
B. What is MEV?
This brings us to the question: what is MEV? One way to think about it is
that MEV is like maximal extractable ground rent, but for blockspace. Because
the validator-proposer controls her piece of blockspace (a block), and because
users have rivalrous, valuable uses for the scarce space, the proposer can charge
users proportionately to the value they attach to what can be done in the space.
Note, that the power of the proposer is not only to choose which transactions
will be included in a block, but also in what order. And both inclusion and
Ethereum. On Ethereum, any network participant who seeks to become a validator must lock up
(stake) 32 ETH in a sort of escrow account in order to propose blocks to the rest of the network.
See Proof-of-stake (POS), Ethereum.org, https://ethereum.org/en/developers/docs/consensus-
mechanisms/pos/.
43 Id.
44 Introduction - What is MEV-Boost, Flashbots Docs, https://docs.flashbots.net/flashbots-mev-
boost/introduction. The main relays are include those operated by Flashbots, BloXroute,
Blocknative, and Eden. Both block builders and block proposers can connect to multiple competing
relays. Anish Agnihotri, MEVBOOST.ORG, https://www.mevboost.org/ (last visited 1/23/2023).
45 Id.
46 Id.
47 Daian et al., supra note 9.
ordering constitute sources of value that may accrue to the proposer. Consider
the following example of arbitrage.
Ethereum is the largest public permissionless blockchain by market
capitalization to enable smart contract48 functionality. This has given rise to a
vibrant ecosystem of decentralized applications (dApps) on Ethereum but is also
largely the cause of MEV on Ethereum. To illustrate, consider a smart contract
which stores multiple crypto assets and provides Ethereum users with automated
crypto asset exchange services. This describes a popular kind of dApp called a
decentralized exchange (DEX). Notably, DEXs are magnets for MEV
extraction.49 There are many such DEXs operating on the Ethereum blockchain
(as well as other blockchains), and the prices of a particular crypto asset on
different DEXs may diverge. Accordingly, arbitrage opportunities arise where it
becomes profitable for a strategic actor to buy the crypto asset on one DEX
which offers it for a lower price and sell the crypto asset on another DEX which
offers it for a higher price. This describes a common MEV extraction strategy
known as “DEX arbitrage,” in which the arbitrageur earns a riskless profit
because of a cross-DEX price discrepancy.
Given that, on its face, this profitmaking strategy seems independent of
transaction ordering and inclusion, one may wonder what exactly makes DEX
arbitrage a form of MEV extraction. In a blockchain ecosystem with many
sophisticated users, arbitrage opportunities have become very competitive.50
Accordingly, in practice, multiple arbitrageurs will almost always be competing
for a particular arbitrage opportunity. Because competition for a particular profit
opportunity necessarily implicates transaction ordering and inclusion, it is the
existence of competition that transforms an otherwise ordinary trading practice
into a form of MEV extraction. To put it simply, only those who can process
their arbitrage transactions earliest will be able to exploit the price discrepancy
before others erase it through similar trades. Accordingly, the transaction
ordering characteristic of MEV becomes necessary to exploit arbitrage
opportunities. Thus, one way to look at MEV is as the theoretical maximum
profit that a block proposer can extract through the strategic ordering and
placement of transactions in a block.51
48 Smart contracts are segments of code and data used to program money to autonomously perform
functions or series of transactions upon the occurrence of predefined conditions. Smart contracts
can even be grouped or linked together to execute increasingly complex transaction chains and/or
create decentralized applications (dApps).
49 See, e.g., Jiahua Xu et al., SoK: Decentralized Exchanges (DEX) with Automated Market Maker
incorporate both permissionlessness and the potential for a MEV opportunity to require some
amount of initial capital).
52 In proof-of-stake Ethereum it is known some time in advance who will have the right to propose
(control) which block in a given “epoch,” which may facilitate either cooperation or single-actor
strategies. According to Barczentewicz: “To be randomly selected as a proposer of two consecutive
blocks once a month may currently require running around 1,250 validators—i.e., staking 40,000
ETH (over $53 million)”; Barczentewicz, supra note 18 at 13–14.
53 See infra Section IV.C.
54 Jeff Kauflin, The Secretive World Of MEV, Where Bots Front-Run Crypto Investors For Big
57 The other one is bundle atomicity, see infra Section II.E. One qualification to this promise is that
transactions routed through privacy RPCs may become public due to risks endemic to the protocol.
See Flashbots, Uncle Bandit Risk, ғʟᴀsʜʙᴏᴛs ᴅᴏᴄs, https://docs.flashbots.net/flashbots-
protect/rpc/uncle-bandits.
58 See, e.g., Alchemy, How to Send Private Transactions on Ethereum, ᴀʟᴄʜᴇᴍʏ ᴅᴏᴄs (Sept. 7,
2022), https://www.alchemy.com/overviews/ethereum-private-transactions.
59 Barczentewicz, supra note 18 at 21.
60 See Sebastian Bürgel, DERP Example 3: Uniswap MEV, ᴍᴇᴅɪᴜᴍ (March 2022),
https://medium.com/hoprnet/derp-example-3-uniswap-mev-c2a8d3417c8.
Cases (ii) and (iii) cover situations where service providers do not control
the broadcasting of pending transactions, but may exclusively possess earlier or
more complete information about the user’s trading intent than other network
participants. Hence, these groups may possess material non-public information
about their users’ trades. For instance, those in group (ii) can know which of
their users’ blockchain queries did not end up in submitted transactions, giving
those nodes an information advantage regarding their users’ potential upcoming
trades. Likewise, those in group (iii) may have the ability to track the activity of
their users on their front-end interfaces, equipping them with valuable data about
their users’ trading preferences and behaviors.
Moreover, when a user (or, more likely, a wallet provider) submits a
transaction to an RPC operator in group (i), this operator is in a privileged
position as they possess non-public information until the moment where they
rebroadcast the transaction to other nodes in the network. While reputational
considerations tend to incentivize nodes to act honestly (i.e., to rebroadcast the
mempool transactions they receive in a timely manner), there may be cases in
which these incentives are not strong enough. For instance, a node who is the
first recipient of a mempool transaction may treat that transaction as private
order flow (POF), intentionally delaying rebroadcasting that transaction for just
fractions of a second such that it would be very difficult for external observers
to detect their misbehavior. With a delay of several hundred milliseconds, the
misbehaving node could assess whether the transaction presents a profitable
MEV extraction opportunity and, if so, submit the node’s own transactions to
take advantage of the opportunity. This series of events could all occur before
the transaction actually becomes public (i.e. accessible to other nodes in the
network).
In sum, transactions are not assured to be public or private on the basis of
their purported routing. It is important to understand the nuance involved in
transaction routing on Ethereum prior to classifying transactions as public or
non-public, given the significant legal consequences associated with the
publicness of information. We suggest that a meaningful standard of publicness
for a pending transaction would be the following: a transaction is public when
an actor, who did not receive the transaction directly from a user who submitted
the transaction, can access it in an unencrypted state without too much delay and
without special arrangements with the node that originally received the
transaction.61 This standard would be satisfied even if reliably detecting public
transactions requires maintaining “watcher” nodes simultaneously in several
because transactions may be transmitted in an encrypted state, e.g., in commit-reveal schemes; see
M. Arulprakash & R. Jebakumar, Commit-reveal strategy to increase the transaction
confidentiality in order to counter the issue of front running in blockchain, 2460 AIP CONFERENCE
PROCEEDINGS 020016 (2022). To the extent a transaction is encrypted, it does not give access to
material information. There may be encryption schemes that only encrypt some parts of a
transaction: in that case, some – but not all – information about a pending transaction may be public.
62 Sandwich trades are often referred to as “sandwich attacks.” Instead, throughout the rest of this
paper, we refer to this practice as a “sandwich trade” or a “sandwich”. In doing so, we seek to avoid
unhelpful and premature normative implications, allowing for genuine debate regarding the
normative and legal character of sandwich trades.
63 See supra Section I.
64 Just-in-time (“JIT”) liquidity provision may be structured as a sandwich where the sandwiched
transaction is front-run by a transaction providing more liquidity to a given smart contract market
(“liquidity pool”), thus improving the price of execution for the sandwiched transaction, and then
back-run by removing the liquidity added earlier and realizing profits. JIT is profitable if the
liquidity provider can obtain sufficient trade fees for providing a large proportion of liquidity during
the sandwiched trade. See Robert Miller (@bertcmiller), Twitter (Nov. 12, 2021, 4:04 PM),
https://twitter.com/bertcmiller/status/1459175377591541768. By providing this momentary
liquidity, JIT reduces the share of fees collected by “passive” liquidity providers and thus reduces
incentives to engage in “passive” liquidity provision; See, e.g., Chainsight (@ChainsightLabs),
Twitter (Nov. 9, 2021, 7:30 AM),
https://twitter.com/ChainsightLabs/status/1457958811243778052.
65 This is more likely to happen if the asset that is expected to rise in price cannot be leveraged for
a riskless back-run from a different on-chain market (e.g. is not available in any other “liquidity
pool”). See, e.g., Stalkopat, Flashbots Discord server (Aug. 8, 2022, 7:53 PM); Hasu (@hasufl),
Twitter (Sep. 26, 2021, 5:14 PM), https://twitter.com/hasufl/status/1442145582978674713.
66 EIGENPHI RESEARCH, supra note 11 at 5.
67 Id. at 26.
68 Id. at 27.
69 See supra Section II.C.
70 Amber Group, supra note 35.
71 SIRIO ARAMONTE ET AL., DeFi lending: intermediation without information?, (2022); Kaihua
Qin et al., An empirical study of DeFi liquidations, in PROCEEDINGS OF THE 21ST ACM INTERNET
MEASUREMENT CONFERENCE (2021), https://doi.org/10.1145%2F3487552.3487811.
72 We refer to what Qin et al. call “fixed spread liquidation” (used, e.g., by Aave, Compound, and
dYdX) as distinguished from “auction liquidations” (used, e.g., by MakerDAO); only a fixed spread
liquidation “allows to extract value in a single, atomic transaction”; Qin, Zhou, and Gervais, supra
note 49 at 5.
73 ARAMONTE ET AL., supra note 71.
74 See infra Section IV.C. For an introduction to the concept of an “oracle,” see supra note 25.
competitive valuable use of blockspace can be identified. This may include, e.g.,
NFT “minting” and trading opportunities.75
75 One example of Long tail MEV is the practice of NFT sniping. NFT sniping involves a MEV
extractor identifying a NFT offered at a price which is significantly lower than both i) the previous
lowest offer for the collection and ii) the average of prices for NFTs within the collection listed
before and after the instance of NFT sniping. The MEV extractor would then offer a high bribe
payment to validators in order to front-run other potential purchasers and “snipe” the NFT at the
low price. See, e.g., Zhou et al., supra note 32; Amber Group, supra note 35.
76 Carsten Baum, James Hsin-yu Chiang, Bernardo David, Tore Kasper Frederiksen & Lorenzo
Gentile, SoK: Mitigation of Front-running in Decentralized Finance, ɪᴀᴄʀ ᴄʀʏᴘᴛᴏʟ. ᴇᴘʀɪɴᴛ ᴀʀᴄʜ.
(2021) at 4, https://eprint.iacr.org/2021/1628.pdf.
77 Peyman Momeni, Sergey Gorbunov, and Bohan Zhang, FairBlock: Preventing Blockchain
Front-running with Minimal Overheads, ɪᴀᴄʀ ᴄʀʏᴘᴛᴏʟ. ᴇᴘʀɪɴᴛ ᴀʀᴄʜ. (2022) at 21-22,
https://eprint.iacr.org/2022/1066.pdf.
78 See Zach, Miner-Extractable Value, Oracle Front-running, and the Rise of Arbitrage Bots, sᴍᴀʀᴛ
some sandwich attacks take place where the front-run and back-run are separated by more than 200
transactions; Qin, Zhou, and Gervais, supra note 49 at 4–5.
82 See A JOINT REPORT OF THE SEC AND THE CFTC ON HARMONIZATION OF REGULATION (2009)
(“Since the 1930s, securities and futures have been subject to separate regulatory regimes.”).
class of crypto assets.83 Both the SEC and CFTC have opted for an adjudication-
based (as opposed to rule-based) approach to policymaking with respect to
crypto assets, with both agencies currently pursuing enforcement actions
grounded in claims that particular crypto assets belong within their respective
jurisdictional domains.84 Commentators have also joined this debate to offer
important practical, legal, and technical considerations implicated in this issue.85
We do not attempt to resolve these tensions and proceed throughout this paper
on the assumption that either the SEC’s or CFTC’s regulatory regime may apply
to crypto assets affected by the MEV extraction strategies we discuss.
Conveniently, as we’ll see, the substance of most of the applicable standards
relating to market manipulation relevant to MEV are the same regardless of
whether the asset is classified as a security or a commodity.86
83 Compare SEC Chair Gary Gensler, Speech: “Kennedy and Crypto”, September 8, 2022, available
13, 2022) (treating relevant the digital assets as commodities); Complaint at 6, SEC v. Eisenberg,
No. 1:23-cv-503 (S.D.N.Y. filed Jan. 20, 2023) (treating other digital assets, specifically
governance tokens on Mango Markets, as securities).
85 See, e.g., Cohen, Lewis R., Strong, Gregory, Lewin, Freeman & Chen, Sara, The Ineluctable
Modality of Securities Law: Why Fungible Crypto Assets are Not Securities (Nov. 10, 2022),
https://dlxlaw.com/wp-content/uploads/2022/11/The-Ineluctable-Modality-of-Securities-Law-
%E2%80%93-DLx-Law-Discussion-Draft-Nov.-10-2022.pdf (discussion draft); Thomas L.
Hazen, Tulips, Oranges, Worms, and Coins – Virtual, Digital, or Crypto Currency and the
Securities Laws, 20 N.C. J.L. & Tech. 493 (2019) (“under most, if not all, circumstances, crypto
currencies are likely to be securities”).
86 See, e.g., Prohibition on Manipulative and Deceptive Devices, 76 Fed. Reg. at 41399 (“The
language of CEA section 6(c)(1), particularly the operative phrase ‘manipulative or deceptive
device or contrivance, is virtually identical to the terms used in section 10(b) of the Securities
Exchange Act of 1934”) (internal quotation marks omitted).
87 See 7 U.S.C. § 6(c) and 9(a)(2) and 15 U.S.C. § 78(i) and 78(j).
to include any operation of the … market that does not suit the gentleman who
is speaking at the moment”.88 As we’ll see this challenge is posed in especially
stark terms by the complex phenomenon of MEV extraction. As such, MEV is
an ideal vehicle for illuminating operative assumptions and crystallizing issues
that require clarity.
Despite the jurisdictional differences of the SEC and CFTC, the purpose
motivating each agency’s anti-manipulation enforcement is the same. According
to Professor Gina-Gail Fletcher, market manipulation, if left unchecked, “can
eventually lead to the demise of the market” because it i) “[interferes] with price
accuracy” by injecting false information into the market and creating false
impressions of liquidity, and ii) “adversely impacts market integrity” by harming
the actual and perceived fairness of the market.89 Accordingly, the SEC and
CFTC are concerned with market manipulation in their respective markets for
the same reason: because it undermines the efficiency (including, but not limited
to, price accuracy)90 and integrity of the markets which it is their role to protect.
They root out manipulative behavior which harms price accuracy by prohibiting
price manipulation, and that which harms market integrity by prohibiting fraud
and misstatements with respect to the asset class they regulate. We address each
of these broad prohibitions in turn.91
1. Price Manipulation
Both section 9(a)(2) of the SEA and Section 6(c)(3) of the CEA prohibit
price manipulation.92 Historically, the CFTC has been more active than the SEC
in exercising their price manipulation authority under CEA s6(c)(3), codified by
the agency as Rule 180.2,93 because it was largely their only means of anti-
manipulation enforcement prior to the passage of the Dodd-Frank Wall Street
88 Craig Pirrong, Commodity Market Manipulation Law: A (Very) Critical Analysis and a Proposed
Alternative, 51 Wash. & Lee L. Rev. 944, 949 (1994), quoting 2 FEDERAL TRADE COMM'N,
THE COTTON TRADE, S. Doc. No. 100, 68th Cong., Ist Sess. 148 (1924), mtcroformed on CIS
No. 8242 (Congressional Info. Serv.).
89 Fletcher, supra note 21.
90 Id. at 490.
91 The SEC and CFTC also prohibit "fictitious trades”, another broad category of manipulative
practices. We focus our discussion here on price manipulation and fraud/misstatements as these
classes of manipulative behavior are most relevant to our legal analysis of MEV extraction
practices. For a discussion of fictitious trades, see Id. at 499.
92 See SEA s9(a)(2) [15 U.S.C. s78i(a)(2)] (prohibiting transactions in a security which “creat[e]
actual or apparent active trading” or “rais[e] or [depress] the price of such a security, for the purpose
of inducing the purchase or sale of such security by others”) and CEA s6(c)(3) [7 U.S.C. s9(3)]
(making it unlawful to “manipulate or attempt to manipulate the price of any swap, or of any
commodity in interstate commerce, or for future delivery on or subject to the rules of any registered
entity”).
93 17 C.F.R. § 180.2 (2012).
Reform and Consumer Protection Act (Dodd-Frank).94 Meanwhile, the SEC has
tended more often to pursue price manipulation cases under SEA s10(b) and
Rule 10b-5,95 the agency’s longstanding authority to police fraud-based
manipulation, when possible.96 Accordingly, and as reflected below, much of
the defining features of price manipulation are expounded in the case law of
enforcement actions brought by the CFTC.
CFTC Rule 180.2 renders it unlawful for “any person, directly or indirectly,
to manipulate or attempt to manipulate the price of any swap, or of any
commodity in interstate commerce”.97 There are four requisite elements to a
successful claim for price manipulation: (1) an artificial price existed; (2) the
accused caused the artificial price; (3) the accused had the ability to influence a
market price; and (4) the accused specifically intended to cause the artificial
price.98
(1) An artificial price is a price which “does not reflect the market or
economic forces of supply and demand”.99 A price is considered artificial where
it is “affected by a factor which is not legitimate.”100 Price artificiality is often
called the sine qua non of price manipulation,101 yet, no binding tests exists for
determining which forces or factors informing a price are legitimate and which
are not.102 Thus, some scholars question the meaningfulness of an artificiality-
based standard.103 As such, determinations of price artificiality generally depend
on a variety of considerations including, but not limited to, i) the competitiveness
94 Merritt B Fox, Lawrence R Glosten & Gabriel V Rauterberg, Stock market manipulation and its
July 15, 1987); see also SEC v. Resch-Cassin & Co., 362 F. Supp. 964, 978 (S.D.N.Y. 1973)
(finding manipulation of the price of a security in violation of SEA §9(a)(2) because defendant
made “it appear to be the product of the independent forces of supply and demand when... in reality,
it was completely a creature of defendants’ subterfuge”).
100 In re Cox, ¶ 23,786 at 26,060.
101 See, e.g., Pirrong, supra note 88 at 956.
102 In re Indiana Farm Bureau Coop. Ass'n, Inc., [1982-1984 Transfer Binder] Comm. Fut. L. Rep.
(CCH) 21,796, at 80-81,281(CFTC Dec. 17, 1982)(Johnson, C., concurring) (“Legitimacy with
respect to supply and demand is undefined in law and economics, unless the sole question is
whether the forces were put in motion by an illegal act”).
103 Frank H. Easterbrook, Monopoly, Manipulation, and the Regulation of Futures Markets, 59 J.
BUS. S103, S117 (1986) (“An effort to isolate which “forces of supply and demand” are “basic”
and which are not is doomed to failure. (…) Economists think of supply and demand as givens.
(…) There is no way to say what demand is real and what is artificial.”); Matthijs Nelemans,
Redefining Trade-Based Market Manipulation, 42 VAL. U. L. REV. 1169 (2008)(arguing that
“prohibitions to counteract traders who cause artificial prices” are problematic because they “lack
a precise delineation of ‘non-artificial price’ versus ‘artificial price’”).
104 See, e.g., United States CFTC v. Donald R. Wilson & Drw Invs., No. 13 Civ. 7884, 2018 LEXIS
207376, at *40 (S.D.N.Y. Nov. 30, 2018) (“a price is artificial when it has been set by some
mechanism which … prevent[s] the determination of those prices from free competition alone”)
(internal citations omitted).
105 See, e.g., In re Tether & Bitfinex Crypto Asset Litig., 576 F. Supp. 3d 55 (S.D.N.Y. 2021)(
Plaintiffs sufficiently alleged price manipulation on the basis of defendants’ fraudulent issuances
of unbacked Tether (USDT), which defendants’ publicly stated were backed by the US dollar);
Resch-Cassin & Co., 362 F. Supp. at 964, 977 (S.D.N.Y. 1973)(defendants engaged in price
manipulation because they “create[d] a false appearance of activity in the over-the-counter market
[which tended] to support the price at an inflated level” by using their “dominion and control of the
market”); Easterbrook, supra note 103 at 118 (“manipulation is a form of fraud [in which]…the
profit flows solely from the trader's ability to conceal his position from other traders”).
106 In re Amaranth Natural Gas Commodities Litig., 587 F. Supp. 2d 513, 535 (S.D.N.Y 2008) (“If
a trading pattern is supported by a legitimate economic rationale, it cannot be the basis for liability
under the CEA because it does not send a false signal”).
107 In re Cox, ¶ 23,786, at 35-36,060 (CFTC July 15, 1987) (“Once the Division of Enforcement
shows that the respondents had the ability to influence prices and that the prices in question were
artificial, it must then show that the respondents caused the artificial prices”).
108 See, e.g., CFTC v. Parnon Energy Inc., 875 F. Supp. 2d 233, 246 (S.D.N.Y. 2012) (applying
but-for test in this context); In re Cox, ¶ 23,786 at 11,060 (“accused lacks the ability to influence
prices if other market participants can bypass his demands and extinguish their obligations
elsewhere”).
109 In re Cox, ¶ 23,786 at 12-13, 060 (“the acquisition of market dominance is the hallmark of a
long manipulative squeeze”); Resch-Cassin & Co., 362 F. Supp. at 977 (“dominion and control of
the market for the security” are factors establishing causation of an artificial price).
110 In securities price manipulation cases under SEA s9(a)(2), the language used in reference to this
element is sometimes different. In the context of securities price manipulation, courts often use
terms like “purpose” (see Resch-Cassin & Co., 362 F. Supp. at 977), “motive”, and “willfulness”
(see Crane Co. v. Westinghouse Air Brake Co., 419 F.2d 787, 795 (2d Cir. 1969)) when referring
to the requisite scienter for a violation.
liability requires a showing that the defendant “acted (or failed to act) with the
purpose or conscious object of causing or effecting a price or price trend in the
market that did not reflect the legitimate forces of supply and demand”.111 The
CFTC’s recent defeat in CFTC v DRW & Wilson re-emphasized that “mere
intent to affect prices is not enough” to establish a price manipulation claim, but
the defendant must have “intended to cause artificial prices”.112
Price manipulation liability alone has proved inadequate as a vehicle for
protecting the efficiency and integrity of markets.113 Given the stringent
requirements of establishing price artificiality and intent to manipulate prices,
the SEC has consistently strayed away from exercising its anti-price
manipulation authority in securities market manipulation cases, opting instead
to rely on the fraud-based manipulation prohibition under SEA s10(b) and Rule
10b-5.114 More remarkably, the CFTC – who was until recently left with no other
choice but to police commodities market manipulation through price
manipulation charges – tried time and again to bring price manipulation claims,
but has only a single court victory to show for it.115
Realizing the inadequacy of this approach, Congress imbued the CFTC
with expanded authority, modeled explicitly after the SEC’s Rule 10b-5, to
effectively police commodities market manipulation in 2010, through the
passage of s753 of Dodd-Frank and codified in CFTC Rule 180.1, to which we
now turn.116
2. Fraud-Based Manipulation
S753 of Dodd-Frank amended CEA s6(c) to give the CFTC the authority –
long exercised by the SEC for securities under SEA s10(b) and Rule 10b-5 – to
prohibit the use of any “manipulative or deceptive or contrivance” in
contravention of CFTC rules in connection with commodities, swaps, or
111 In re Indiana Farm Bureau Coop. Ass'n, Inc., [1982-1984 Transfer Binder] Comm. Fut. L. Rep.
*39, quoting In re Amaranth Natural Gas Commodities Litig., 587 F. Supp. 2d 513, 535 (S.D.N.Y
2008).
113 Rosa M. Abrantes-Metz, Gabriel Rauterberg, & Andrew Verstein, Revolution in Manipulation
Law: The New CFTC Rules and the Urgent Need For Economic and Empirical Analyses, 15 Penn.
J. Bus. L., 357 (2013); Jerry W. Markham, Manipulation of Commodity Futures Prices-The
Unprosecutable Crime, 8 Yale J. On Reg. 281 (1991) (noting that price manipulation is “virtually
unprosecutable” as “Plaintiffs must establish a manipulative intent that is conceptually and
doctrinally among the most demanding mental state requirements anywhere in financial law.”).
114 Maxwell K. Multer, Open-Market Manipulation Under SEC Rule 10b-5 and its Analogues:
futures.117 CEA s6(c)(1), codified through CFTC Rule 180.1,118 empowered the
CFTC to police market manipulation even in the absence of evidence
establishing a defendant’s specific intent to manipulate prices or the existence
of an artificial price.119 In relevant part, Rule 180.1 makes it unlawful for those
engaged in commodities trades “to intentionally or recklessly”:
(1) Use or employ, or attempt to use or employ, any manipulative device, scheme,
or artifice to defraud;
(2) Make, or attempt to make, any untrue or misleading statement of a material
fact or to omit to state a material fact necessary in order to make the statements
made not untrue or misleading;
(3) Engage, or attempt to engage, in any act, practice, or course of business, which
operates or would operate as a fraud or deceit upon any person[.] 120
In the adopting release accompanying the CFTC’s enactment of Rule
180.1, the agency clarified that its application would be “guided, but not
controlled, by the substantial body of judicial precedent” interpreting the
Securities and Exchange Commission’s Rule 10b-5.121 That is, the interpretation
of Rule 180.1 in the context of commodities markets draws explicitly from Rule
10b-5 precedent in securities markets.122
The requisite elements of a successful 180.1 enforcement action include
evidence of: i) reckless or intentional conduct by the accused, and ii) a
“manipulative device, scheme, or artifice to defraud.”123 Like Rule 10b-5, Rule
180.1 is intended be a “broad catch-all provision” capturing all instances of
fraud-based manipulation, and it has been applied as such.124 Rule 180.1, in its
relatively few years of existence, has been used by the CFTC to prosecute
conduct ranging from insider trading in commodities125 to the alleged corporate
misconduct of Samuel Bankman-Fried and related entities in the FTX
debacle.126
Devices and Prohibition on Price Manipulation, 76 Fed. Reg. 41398, 41403 (July 14, 2011)
(codified at 17 CFR pt. 180).
120 17 C.F.R. §180.1 (2012).
121 Supra note 119.
122 Gregory Scopino, The (questionable) legality of high-speed pinging and front running in the
Watkins, Insider Trading in Commodities Markets: An Evolving Enforcement Priority, Client Alert
White Paper (March 11, 2021),
https://www.lw.com/admin/upload/SiteAttachments/Alert%202827.v5.pdf.
126 See Complaint at 2, CFTC v. FTX Trading et al, No. 1:22-cv-10503 (S.D.N.Y. filed Dec. 13,
2022).
Yet, while capacious, Rule 180.1 still has discrete limits in its application.
Most importantly, Rule 180.1 parallels SEC Rule 10b-5 in that it is “described
as a catchall provision, but what it catches must be fraud”.127 With this said, it is
important to note that fraud in the context of fraud-based manipulation is not
just fraud in its common sense, as express misrepresentation or deceptive
omission.128 Rather, fraud-based manipulation under Rule 10b-5 and Rule 180.1
can include both claims of fraud by misleading statements or deceptive
omissions, and manipulative action which send a “false pricing signal to the
market.”129 A promising means for establishing fraud-based manipulation is the
fraud-on-the-market (FOTM) theory.130 While the FOTM theory has long been
used in the securities fraud context, the advent of Rule 180.1 suggests that it may
have some success in commodities’ manipulation cases as well.131 More
specifically, Gregory Scopino proposes a variant of FOTM which he calls, and
we will refer to, as the manipulation-as-fraud legal theory, which provides:
market participants are entitled to rely on the assumption that the securities
market is free of manipulation and they are therefore deceived when,
unbeknownst to them, a wrongdoer manipulates the market and distorts the way
that the market prices securities.132
It is important to note here that the courts themselves disagree as to whether
trading activity alone is a form of manipulative action that can constitute fraud-
based manipulation in the absence of express misrepresentations or other
independent unlawful acts.133 This raises a concept that will figure into our
analysis below: open-market manipulation. 'Open-market manipulation’ refers
127 See Chiarella v. United States, 445 U.S. 222, 235-236(1980) (describing SEC’s Rule 10b-5) ;
United States CFTC v. Kraft Foods Grp., Inc., 153 F. Supp. 3d 996, 1010 (N.D. Ill. 2015) (“this
Court finds that Section 6(c)(1) and Regulation 180.1 prohibit only fraudulent conduct”).
128 ATSI Commc’ns, Inc. v. Shaar Fund, Ltd., 493 F.3d 87, 100 (2d Cir. 2007) (“Section 10(b), in
proscribing the use of a ‘manipulative or deceptive device or contrivance,’... prohibits not only
material misstatements but also manipulative acts”).
129 In re Tether & Bitfinex Crypto Asset Litig., 576 F. Supp. 3d 55, 114 (S.D.N.Y. 2021), quoting
ATSI Commc'ns, 493 F.3d at 100 (observing that both SEA 10(b) and the CEA plus Rule 180.1
“‘prohibit[] not only material misstatments but also manipulative acts,’ including ‘a transaction that
sends a false pricing signal to the market.’”).
130 In the securities context, the FOTM theory “establishes a rebuttable presumption in private rights
of action under Exchange Act 10(b) and SEC Rule 10b-5 that in an efficient market for a security
a plaintiff can be held to have relied on a defendant’s fraudulent misrepresentation or omission in
connection with the purchase or sale of a security—even if the plaintiff was not aware of the
misrepresentation or omission—by virtue of the plaintiff’s reliance on the fact that a security’s
price reflects the fraudulent misrepresentation and omission”. Supra note 119 at 41402 n.50.
131 In the enacting release of Rule 180.1, the CFTC “decline[d] to adopt comments recommending
outright rejection of the potential application of the ‘fraud-on-the-market’ theory under final Rule
180.1.” Id. at 41403.
132 Scopino, supra note 122 at 672.
133 See Fox, Glosten, and Rauterberg, supra note 94 at 119–122 (discussing the circuit split with
respect to whether trading activity -- i.e. “open market manipulation” -- on its own can constitute
fraud-based manipulation under Rule 10b-5).
134 For a more in-depth discussion of open-market manipulation, see Fletcher, supra note 21.
135 While many instances of open-market manipulation could theoretically be brought as price
manipulation charges, the aforementioned challenges of successful price manipulation actions
make the agencies’ fraud-based manipulation prohibitions (10b-5 and Rule 180.1) better suited to
open-market manipulation.
136 In re Amaranth Natural Gas Commodities Litig., 587 F. Supp. 2d 513, 535 (S.D.N.Y 2008) (“a
Brewing Co. Securities Litigation, 613 F. Supp. 1286, 1292 (N.D. Ill. 1985)(“[r]egardless of
whether market manipulation is achieved through deceptive trading activities or deceptive
statements….it is clear that the essential element of the claim is that inaccurate information is being
injected into the marketplace”).
138 CFTC v. Amaranth Advisors, L.L.C., 554 F. Supp. 2d 523, 528 (S.D.N.Y. 2008) (“purchasing a
substantial number of futures contracts leading up to the closing range on expiration day, followed
by the sale of those contracts several minutes before the close of trading[] is known as “marking
the close”); id. at 534 (“there is no doubt that marking the close or any other trading practices,
without an allegation of fraudulent conduct, can also constitute manipulation in contravention of
the CEA, so long as they are pursued with a manipulative intent”).
139 CFTC, ‘Banging the Close‘, CFTC Glossary,
https://www.cftc.gov/LearnAndProtect/EducationCenter/CFTCGlossary/glossary_b.html#:~:text=
Banging%20the%20Close%3A%20A%20manipulative,position%20in%20an%20option%2C%2
0swap (last visited Jan. 25, 2023); Koch v SEC (finding market manipulation in violation of SEA
s10(b) for marking the close).
pricing of a different, but related position.140 These trading strategies have been
ruled out as deceptive forms of manipulation.141
In its paradigmatic sense, banging the close constitutes what has been
called a covered open-market manipulation scheme.142 Covered open-market
manipulation derives its name from the fact that the trading activity involves (is
covered by) a structural mechanism or arrangement that generates heightened
duties of trust and honest dealing. Thus, covered open-market manipulation
typically involves trades in some financial instrument X with the purpose of
moving the price of financial instrument Y, where the pricing of Y is explicitly
determined (i.e. through a benchmark or other formal dependent pricing
mechanism) by reference to the price of X.143 It is this pricing mechanism for Y,
which explicitly references X, that creates the relationship of trust which entails
heightened duties to avoid manipulation (e.g. an implied expectation that one
will avoid affecting Y’s price through exploiting trades in X, rather than simply
allowing Y’s price to be set independently by uninterested third parties’ good
faith demand for X).
By contrast, naked open-market manipulation is not covered by any trust
relationship resulting in heightened duties. A paradigmatic example would be a
simple “buy-low, sell-high” profitmaking strategy, with the added element that
the trader has “some way of preventing the price from increasing as she
purchased, decreasing as she sells, or both.”144 Covered open-market
manipulation schemes involve some explicit or implicit agreement – that is, an
agreement not to manipulate a predetermined formal pricing mechanism – which
is interfered with by the manipulator.145 By contrast, naked open-market
manipulation schemes involve no such agreements or relations of trust, whether
express or implied, which would be breached by their occurrence. As such,
naked open market manipulation schemes tend to be both more difficult to pull
off, at least profitably, and predictably will be more difficult to prosecute than
covered open-market manipulation. We return to these issues in discussing
sandwiching public versus private transactions in Sections IV.A-B.
140 CFTC v. Wilson, No. 13 Civ. 7884 (RJS), 2018 LEXIS 207376 (S.D.N.Y. Nov. 23, 2018) at
*57 (defining ‘banging the close’ as involving “someone putting in a disproportionate number of
trades to push the price up or down to affect the closing price, typically in a noneconomic fashion,
to benefit a position that they held elsewhere”).
141 See supra notes 138, 139.
142 Covered open-market manipulation has also been referred to as open-market manipulation with
an external interest and contract-based manipulation. See Fox, Glosten, and Rauterberg, supra note
94 at 75.; Daniel R. Fischel & David J. Ross, Should the Law Prohibit “Manipulation” in Financial
Markets?, 105 Harv. L. Rev. 503, 523 (Dec. 1991).
143 See Fletcher, supra note 21 at 503.
144 Id.
145 Id. at 403.
C. Insider Trading
The SEC, CFTC, and Department of Justice (DoJ) each prohibit insider
trading in their respective markets.146 While there has been scholarly debate
about the applicability of insider trading laws to crypto assets,147 the SEC and
DoJ recently clarified their stance when charging Ishan Wahi (a former Coinbase
employee) and associates with wire fraud in the first ever crypto asset insider
trading case.148 We focus on the SEC and CFTC’s insider trading enforcement,
as the analogous criminal law regime is beyond the scope of this paper.
Neither the SEC nor the CFTC’s authority to police insider trading comes
from a statutory or regulatory prohibition. Rather, each agency views insider
trading as a form of fraud and uses their respective anti-fraud provisions to
pursue cases of insider trading. That is, insider trading cases brought by the SEC
are charged as Rule 10b-5 violations,149 those brought by the CFTC are charged
as Rule 180.1 violations.
Much of the major precedent in the realm of insider trading law comes from
securities regulation,150 simply because SEA s10(b) is the oldest vehicle for
pursuing insider trading actions. Notably, the CFTC lacked the authority to bring
insider trading actions in commodities markets, with few exceptions,151 until the
146 See Andrew Verstein, Crypto Assets and Insider Trading Law’s Domain, 105 IOWA L. REV. 1,
13–17 (2019).
147 Compare Id. (arguing that insider trading law should apply to crypto assets) with Mihailis E.
Diamantis, The Light Touch of Caveat Emptor in Crypto’s Wild West, 104 Iowa L. R. 113 (2020)
(noting that “there exists a strong argument that insider trading laws would be unconstitutionally
void for vagueness as applied to cryptocurrency insiders,” and arguing for a light-touch approach
to the enforcement of insider trading laws against crypto asset traders).
148 Press Release, SEC, SEC Charges Former Coinbase Manager, Two Others in Crypto Asset
of the SEA – like SEA s16 and SEA s14e-3. Yet, the SEC pursues most insider trading cases under
SEA s10(b) and Rule 10b-5, because these provisions equip the agency with the broadest authority.
See Verstein, supra note 146 at 14.
150 See Chiarella v. United States, 445 U.S. 222, 234 (1980) (rejecting the notion that the possession
of insider information by traders in an open-market creates any general duty absent a specific duty
to disclose); Dirks v. SEC, 463 U.S. 646, 661 (1983) (holding that, in cases where an insider does
not themselves trade on material nonpublic information, but provides an insider tip to a “tippee”
who then trades on the material nonpublic information, the tippee is liable for insider trading only
where the insider tipper breached their fiduciary duty to the source of the inside information);
United States v. O'Hagan, 521 U.S. 642, 666 (1997) (upholding the misappropriation theory).
151 The only insider trading prohibition enforced by the CFTC prior to Dodd Frank were those
against misuse of information by the CFTC’s own staff and employees of the exchanges and self-
regulatory organizations overseen by the agency (7 U.S.C. § 13(d) (2008); U.S.C. § 13(e) (2008)).
152 In enacting Rule 180.1, the CFTC “recognize[d] that unlike securities markets, derivatives
markets have long operated in a way that allows for market participants to trade on the basis of
lawfully obtained material nonpublic information.” The agency then stated that the new rule may
prohibit “trading on the basis of material nonpublic information in breach of a pre-existing duty”.
Supra note 119 at 41403.
153 Id. at 41402 ; see also O'Hagan, 521 U.S. 642; Verstein, supra note 146 at 15 (“the
misappropriation theory holds that a trader who feigns loyalty to a company or person to gain access
to secrets ultimately defrauds his source out of information when he misuses the information for
trading”).
154 Id.
155 See, e.g., Dirks v. SEC, 463 U.S. 646 (1983); Salman v. United States, 137 S. Ct. 420, 428
(2016).
156 CFTC v. EOX Holdings L.L.C., No. H-19-2901, 2021 WL 4482145, at *45 n.112 (S.D. Tex.
always…necessary, and none of the Supreme Court opinions require a fiduciary relationship [for]
an actionable securities claim under s10(b)); CFTC v. EOX Holdings LLC, No. 19-cv-02901 (S.D.
Tex. Sept. 26, 2019) at *713 (misappropriation theory is not limited to fiduciary relationships).
D. Front-running
In closing, we set aside a source of confusion for crypto markets: “front-
running.” A familiar critique of MEV techniques like sandwiching (discussed in
IV.A-B) is that they involve front-running.160 This follows a colloquial usage
likely derived from Michael Lewis’s influential book Flash Boys, which referred
to a form of High Frequency Trading latency arbitrage (focused on colocation
and other speed advantages) as “electronic front-running.”161 However,
commentators pointed out that latency arbitrage should not be confused with
illegal front-running, insofar as high frequency trading involves only public
information accessed through superior infrastructure.162
In legal contexts, front-running prototypically refers to the illegal practice
of a trusted person (usually, a broker or investment advisor) “trading a security,
option, or future while in possession of non-public information regarding an
imminent block transaction that is likely to affect the price of the stock, option,
or future.”163 Neither the SEC nor the CFTC has promulgated any rule generally
prohibiting front-running (outside of specific, highly regulated contexts),164 but
159 SEC v. Cuban, 634 F. Supp. 2d 713, 726 (N.D. Tex., 2009) (holding that a “duty sufficient to
support liability under the misappropriation theory can arise by agreement absent a preexisting
fiduciary or fiduciary-like relationship”), vacated on other grounds in SEC v. Cuban, 620 F.3d 551,
559 (5th Cir. 2010).
160 See, e.g. Auer, supra note 11.
161 Mɪᴄʜᴀᴇʟ Lᴇᴡɪs, Fʟᴀsʜ Bᴏʏs (2014) (“They created a taxonomy of predatory behavior in the
Mᴀʀᴋᴇᴛ: Lᴀᴡ, Eᴄᴏɴᴏᴍɪᴄs, ᴀɴᴅ Pᴏʟɪᴄʏ 96-99 (noting that “electronic front-running” is different
form front-running in its traditional, legal sense, and that the name is “inapt because the HFT is not
even accused of taking a position in anticipation of another trader’s order,” thus suggesting
“anticipatory order cancelation” as a more accurate label).
163 Christopher Gibson, Initial Decision Release No. 1398 (ALJ March 24, 2020) (initial decision),
37.203(a) requires Swap Execution Facilities (SEFs) to prohibit abusive trading practices including
front-running on their markets. 17 CFR § 37.203 (2013). Similarly, SEC Rule 17(j)-1 has been
interpreted to prohibit portfolio managers from front-running their clients. 17 C.F.R. § 270.17j-I
(2005).
165 See FINRA, RULE 5270 (2013). For the early history of efforts to regulate front-running, see
Markham, "Front-Running" - Insider Trading Under the Commodity Exchange Act, 38 Cath. U. L.
Rev. 69, 72-83 (1988).
166 See Order Instituting Proceedings, In the Matter of Arya Motazedi, CFTC No. 16-02 (Dec. 2,
2015) (insider trading claim in violation of CEA s6(c)(1) for front-running one’s employers).
167 Markham, supra note 113 at 86.
168 See supra Section II.A.
169 See supra definition and description of sandwich trades in Section III.C.
170 See supra discussion of CFTC Rule 180.1 and SEC Rule 10b-5 in Section II.B.
171 For a definition of oracles, see supra note 25. For in-depth discussions regarding the mechanics
of how price oracles can be manipulated in decentralized finance, see infra note 284.
172 See generally Complaint, CFTC v. Eisenberg, No. 23-cv-00173 (S.D.N.Y. filed Jan. 9, 2023).
173 See supra our explanation of sandwiches in Section II.E.
174 See supra for our definition of publicness of pending transactions in Section II.D.
175 See infra Section IV.B.
transaction.176 We also noted that transactions “public” by this standard are only
meaningfully public to professional operators, not to an average user.
In what follows, we will argue that there is a plausible case for thinking
that sandwich attacks constitute a manipulative act within the meaning of the
CFTC Rule 180.1(a)(3) or SEC Rule 10b-5 – at least if courts or regulators were
to adopt a moralized form of inquiry focused on protecting particular (if
widespread) notions of fairness in financial markets, which of course is a big if.
Should courts decide to crack down on the exploitation of privileged positions
of control of the machinery involved in processing transactions on Ethereum
(i.e., control over “blockspace”) in order to extract rents, the courts could
determine that execution prices in sandwiched trades have been subject to an
illegitimate and therefore artificial price effect, amounting to a manipulative
practice. Nonetheless, we argue that this conduct, to the extent it constitutes a
form of naked open market manipulation (not covered by an agreement or other
relationship), is unlikely to rise to the top of the list of the regulatory agencies’
enforcement priorities. Our focus here, nonetheless, is on the theoretical case, as
much can be learned about the legally salient features of this paradigmatic form
of MEV extraction.
1. Price Manipulation
Although we will focus on the SEC and CFTC’s fraud-based manipulation
prohibitions, we start by applying the agencies’ largely dormant price
manipulation prohibitions to the sandwiching of public transactions, if only to
demonstrate why we do not discuss this cause of action further. Recall that a
successful claim for price manipulation requires proof of four elements: i) an
artificial price, ii) the intent to cause that artificial price, iii) having the ability to
cause the artificial price (typically through market dominance), and iv) actually
causing the artificial price.177 Yet, the evidentiary burdens of proving both
specific intent and price artificiality have posed too high a bar for both the SEC
and CFTC to experience much luck in enforcing their anti-price manipulation
authorities. One scholar notes the cause of action has become “virtually
unprosecutable”:178
Plaintiffs must establish a manipulative intent that is conceptually and doctrinally
among the most demanding mental state requirements anywhere in financial law.
Moreover, the evidence for such intent is typically only highly ambiguous public
behavior.179
In a sandwich targeting public transactions, the sandwicher’s actions – that
is, the front-run and back-run of the sandwiched transaction – are open-market
transactions. In other words, they involve the “transacting party simply
2. Fraud-Based Manipulation
7 U.S.C. § 9(1) prohibits any person to “use or employ, or attempt to use
or employ, in connection with any swap, or a contract of sale of any
commodity…any manipulative or deceptive device or contrivance,” in
contravention of CFTC Rules. The CFTC clarified this prohibition with its Rule
180.1, which mirrors the content of SEC rule 10b-5 (prohibiting securities
fraud).184 Rule 180.1, like SEC 10b-5, makes it unlawful for those engaged in
commodities trades to “to intentionally or recklessly”:
(1) Use or employ, or attempt to use or employ, any manipulative device, scheme,
or artifice to defraud;
(2) Make, or attempt to make, any untrue or misleading statement of a material
fact or to omit to state a material fact…;
(3) Engage, or attempt to engage, in any act, practice, or course of business, which
operates or would operate as a fraud or deceit upon any person[.]185
180 Michael A. Asaro, ‘Masri’ and Open-market Manipulation Schemes, 239 N.Y. L. J., May 12,
Commission deems it appropriate and in the public interest to model final Rule 180.1 on SEC Rule
10b-5.” See supra note 119 at 41399.
185 17 CFR § 180.1; 17 CFR § 240.10b-5.
Violating Rule 180.1 requires only reckless action, which is defined for
CEA purposes as “one that departs so far from the standards of ordinary care
that it is very difficult to believe the [actor] was not aware of what he was
doing.”186 This means “the Commission need not prove that the
defendant's…primary motive was to interfere with the forces of supply and
demand.”187
Our analysis will focus on liability for manipulative devices or acts under
(a)(1) and (a)(3), and analogous provisions of SEC Rule 10b-5, as it is unlikely
that sandwiching public transactions will involve untrue or misleading
statements such as under 180.1(a)(2). By itself, sandwiching does not clearly
involve an express false statement. Searchers typically do not make any
representations to the public or to specific traders. Matters might be different, of
course, if some actors who facilitate sandwiching were to make express
representations that they know to be falsehoods. For example, a block builder
could publicly promise not to engage in sandwiching (e.g., by including their
own front- and back-running transactions), but then break that promise.
Similarly, there may be a risk of liability if a relay operator or a block builder
promised to “try its best” not to include transactions attempting to sandwich but
then either (1) they did not try or (2) their efforts were manifestly below the
reasonable level under the circumstances.188
Instead, our focus will be the idea that a standard sandwich involves
prohibited manipulation: either a “manipulative device,” CFTC 180.1(a)(1) and
SEC 10b-5(a), or an act that “would operate as a fraud or deceit,” CFTC
180.1(a)(3) and SEC 10b-5(c).189 That is, we will explore the manipulation-as-
fraud theory, which maintains that:
[W]hen a person engages in manipulative trading practices in the markets and
does not let others know of his manipulative acts, the fraud derives from the
failure to inform the other market participants, who are entitled to rely on their
belief that the market is free of such improper behavior.”190
186 Drexel Burnham Lambert Inc. v. CFTC, 850 F.2d 742, 748 (D.C. Cir. 1988).
187 In the Matter of JPMorgan Chase Bank, 2013 WL 6057042, at *11 (“even if a trader were
motivated by a desire to obtain compensation rather than by a desire to affect a market price, if the
trader recklessly effected the manipulative trades, he will be held liable”).
188 As will be discussed in section B, liability based on express misrepresentations may also arise
in private order flow arrangements where users are promised that their transactions will not be
sandwiched as an inducement to receive exclusive access to their order flow, but where the promise
is broken. Still, this is not the case we are concerned with here, namely sandwiching public
transactions.
189 It’s arguable that some of the manipulative acts in question here could be recast as deceptive
omissions in contravention of 180.1(a)(2). For example, one might maintain that the failure to
disclose a potentially manipulative act is itself a deceptive omission for (a)(2) purposes. However,
this makes the omission theory of liability parasitic on the theory of manipulative act, which is
more fundamental. Therefore, we set aside this conceptual possibility as it is not especially
important in practice. For clarity, we focus chiefly on manipulative acts.
190 Scopino, supra note 122 at 673.
For example, this covers deceptive trading practices such as “banging the close”
– executing many trades at the end of the trading day e.g. to give a false
impression of high trading volume.191 Even though such practices do not involve
affirmative false statements, they count as fraud because they “artificially affect
the price of securities without informing other market participants, who
justifiably rely on the assumption that the market for those securities is
functioning normally and not being manipulated.”192 (This theory of liability is
similar in spirit to a “fraud on the market” theory, applicable in other
contexts.193)
As the Second Circuit explained the securities context, a manipulative act
“‘refers generally to practices…that are intended to mislead investors by
artificially affecting market activity,’ and ‘connotes intentional or willful
conduct designed to deceive or defraud investors by controlling or artificially
affecting the price of securities.’”194 For assessing manipulation, “[t]he critical
question then becomes what activity ‘artificially’ affects a security's price in a
deceptive manner.”195
There are several theories for why sandwiching might contravene CFTC
Rule 180.1 or SEC 10b-5 that are worth exploring – the most plausible of which,
we argue, is that it involves manipulation through having an artificial impact on
prices. But first, let’s focus on deception. Is sandwiching an act or business
practice that “would operate as a fraud or deceit” under 180.1(a)(3) and 10b-
5(c)? Sandwiching might be seen as fraudulent if it misleads either the
sandwiched user or the market generally. We consider each in turn.
(i) First simple theory: the sandwiched user is at least recklessly misled
First, one might ask whether sandwiching is likely to involve knowingly or
at least recklessly misleading the sandwiched user herself. The idea would be
191 CFTC v. Amaranth Advisors, L.L.C., 554 F. Supp. 2d 523, 528 (S.D.N.Y. 2008) (“purchasing a
substantial number of futures contracts leading up to the closing range on expiration day, followed
by the sale of those contracts several minutes before the close of trading[] is known as “[banging]the
close”); id. at 534 (“there is no doubt that [banging] the close or any other trading practices, without
an allegation of fraudulent conduct, can also constitute manipulation in contravention of the CEA,
so long as they are pursued with a manipulative intent”).
192 Scopino, supra note 122 at 674. See also GREGORY SCOPINO, ALGO BOTS AND THE LAW:
market” presumption, under which “the market price of shares traded on well-developed markets
reflects all publicly available information, and, hence, any material misrepresentations,” and as a
result “the typical ‘investor who buys or sells stock at the price set by the market does so in reliance
on the integrity of that price’”; therefore, “whenever the investor buys or sells stock at the market
price, his ‘reliance on any public material misrepresentations ... may be presumed for purposes of
a Rule 10b–5 action.’”).
194 Set Capital LLC v. Credit Suisse Grp. AG, 996 F.3d 64, 76 (2d Cir. 2021).
195 ATSI Communications, Inc. v. Shaar Fund, Ltd., 493 F.3d 87, 100 (2d Cir. 2007).
that when a searcher sandwiches another user, the first trade within the sandwich
(which front-runs the original user) affects the natural price of the underlying
crypto asset.196 Specifically, it worsens the sandwiched user’s execution price –
and a competent searcher will inevitably be aware of this effect, since it is
integral to making the sandwich more profitable. For example, if it is a buy order
being sandwiched, the front-run will slightly raise the asset’s price before
building the original buy order into the relevant block, and so the sandwiched
user purchases the asset at a slightly higher price than she otherwise would have.
Trades on DEX-es like Uniswap V2 are only executed if the execution price is
within the specified slippage limit, as competent sandwichers will be aware. One
might think that because the searcher fails to disclose this practice to the
sandwiched user, thereby harming her by increasing the slippage of her trade,
this involves at least recklessness (awareness of a substantial risk) as to
interfering with the original user’s justifiable reliance on the integrity of the
market, i.e. the reasonable assumption that prices accurately reflect only the
forces of supply and demand.
Setting aside whether sandwiching has an artificial effect on prices (which
we consider in a moment), we think this simple theory of liability is unlikely to
succeed, since we doubt the sandwiched user is likely to be misled by the
sandwicher. This means recklessness by the sandwicher is unlikely to
exist.197 There are two reasons to expect that the sandwiched user will not be
misled by the sandwicher’s conduct. First, sandwiching and other MEV
extraction strategies generally are common on Ethereum at present, and so it is
arguable that users may reasonably be expected to know that being sandwiched
is a substantial risk for any trade she executes, given how the relevant markets
work.198 However, as Wang et al have shown based on their interviews of DeFi
users, not all users potentially affected by sandwiches are aware of the
phenomenon of sandwiching, and some users who are aware of sandwiching are
not able to recognize whether their transaction has been sandwiched.199 That
said, the authors also noted that “[w]hen the financial loss from a sandwich
attack is not significant, traders do not care whether they are being attacked.”200
196 Depending on the size of the searcher’s purchase order as well as other conditions (amount of
liquidity in the liquidity pool, details of the algorithm with which the AMM calculates prices), the
price impact may vary: from very small, likely imperceptible for the sandwiched user, to very large.
Thus, it could be that appreciable harm to the sandwiched user happens only in some, but not all,
sandwiches. See supra II.C.
197 Drexel Burnham Lambert Inc. v. CFTC, 850 F.2d 742, 748 (D.C. Cir. 1988) (observing that
recklessness is made out in the CEA context when “it is very difficult to believe the [actor] was not
aware of what he was doing” – i.e. when there is an obvious risk).
198 See supra Section II.
199 Over the second half of 2021, Wang et al conducted interviews with 15 DeFi users; Ye Wang et
al., Impact and User Perception of Sandwich Attacks in the DeFi Ecosystem, in Proceedings of the
2022 CHI Conference on Human Factors in Computing Systems (2022) at 6-12,
https://doi.org/10.1145/3491102.3517585.
200 Id. at 8.
It is likely that traders whose trading volume is large enough for sandwiching to
become noticeable are also more likely to be aware of sandwiching. Thus,
because sandwiching is likely common knowledge among those traders who are
potentially meaningfully affected by sandwiching, it seems not very likely to
interfere with users’ expectations of what is likely to happen in the market.201
Second, and more importantly, in each trade that is carried out on a DEX
like Uniswap, users are alerted to the prospect of their transaction executing at a
price significantly different from the currently estimated price because they are
asked to specify a slippage limit beyond which their trade will not be executed
(a slippage limit may also be suggested by default). We shall assume that the
sandwicher only worsens the sandwiched user’s execution price without
preventing the underlying trade from being executed altogether. This could
happen if the front-run moves the price to a level beyond the target transaction’s
slippage limit, but a sandwicher rationally would not want to cause that, since
this means he or she will not profit from the sandwich. Assuming the sandwiched
trade goes through despite the increased slippage from the front-run in the
sandwich, the sandwiched user arguably will have consented to the trade being
processed with this worse but still acceptable execution price. In this way,
because users are alerted to the prospect of slippage particularly through the need
to set a slippage limit on their trades, the sandwich is unlikely to mislead or
deceive sandwiched users in fact. As a result, it is unlikely that a sandwicher
would be found to have the mental state of recklessness as to the prospect of
misleading the sandwiched user.
201 This is not to say that users have normatively positive attitudes towards sandwiching, but only
that it is a risk known to them. As Wang et al noted, some of their “interviewees identified sandwich
attacks as malicious towards traders”; id. at 9.
other words, if the front-run is a purchase of 1,000 DAI for 1 ETH, then a
standard back-run would sell 1,000 DAI. Qin et al. operationalized this intuition
with a notion of “perfect” sandwiching.202 The authors found that 80% of
sandwiches in their sample were “perfect,” and 100% would count as “perfect”
on a definition using a wider 90-110% range.203
A sandwicher maximizes their economic return from a sandwich by
completely reversing the front-run transaction.204 In fact, if sandwiching is the
sandwicher’s only goal, there is no economic reason for them to do anything
other than strive for a full reversal of their position. If a sandwicher does not
manage fully to reverse their position before any other market activity occurs,
then their strategy is not entirely market neutral and they take a kind of
“inventory” risk, because it is possible that the price will later move in a way
that reduces the profitability of their sandwich.
Atomic way of executing sandwiches. If a sandwich is executed atomically,
the MEV extractor is guaranteed that all the transactions they submit will be
executed in the precise set order or none of the transactions will be executed at
all. Atomicity may be guaranteed if the MEV extractor controls block production
(because they are a validator or non-validating block builder) or if the MEV
extractor uses a private relay accepting atomic bundles of transactions (like the
Flashbots Auction). This method of sandwiching is much less financially risky
than non-atomistic approaches, where there is a risk of other users’ transactions
being included between the front-run, the sandwiched trade(s), and the back-run.
When an actor controls the contents of a block and can order the transactions in
the “perfect” manner that fully reverses the front run, this will eliminate the risk
that others might profit from carrying out an interfering trade before the MEV
extractor has realized her profit through the back-run. As a result, atomistic
sandwiches are preferred by MEV extractors compared to non-atomistic
alternatives, which are riskier in financial terms. We do not know precisely what
proportion of sandwiches use the less risky atomistic method today. As
measured by Qin et al., from late 2018 to late 2021, 32% of sandwiches were
privately relayed to miners or added by miners themselves, which strongly
suggests that those MEV extractors had guarantees of atomicity.205
Unsurprisingly, nearly all the privately relayed sandwiches had zero
intermediate, unrelated transactions (submitted by other users) between the
front-run, the sandwiched trade, and the back-run.206
Most sandwiches don’t affect non-sandwiched trades. However, even
when – for any reason – sandwichers cannot rely on a guarantee of atomicity,
they seem mostly to succeed in executing the strategy without having any
intermediate, unrelated transactions “within” a sandwich. Taking all sandwiches
in the Qin et al. study, both privately relayed and not, most of them had no
intermediate transactions.207
Impossibility of adverse price effect on non-sandwiched traders. Even if
most sandwiches do not include intermediary non-sandwiched transactions, a
non-trivial proportion of sandwiches still might. It is possible that some affected
transactions in this minority group are transactions interacting with the same
DEX, which means that these non-sandwiched trades would be affected by the
sandwich. This involves two possibilities: either a) these other trades go in the
same direction as the sandwich’s front-run (buy/sell) or b) they go in the
opposite direction.
a) If those trades are in the same direction as the sandwich’s front-run, then
they are effectively sandwiched even if that was not intended by the sandwicher.
It will be so, because same-direction trades following a front-run receive worse
price execution than they would have otherwise in the absence of the front-run.
This situation persists until someone (the sandwicher – in their back-run – or
someone else, likely an arbitrageur) will execute an opposite-direction trade. In
those cases, our discussion from the previous subsection applies. There we found
there was no risk of misleading sandwiched users themselves, as they are likely
to be aware of the risk of sandwiches hurting their execution price through
setting their slippage tolerance. Still, as discussed in the next section, it’s
conceivable that courts might find such users to have been affected by an
artificial price to the extent sandwiching is deemed to have an unnatural impact
on price. Hence, the situation for such users remains complex and we defer the
conclusion as to this situation until the next sub-section.
b) On the other hand, if the intermediate trades are in the opposite direction
than the front-run, then they benefit from the front-run’s price effect. The
front-run’s price effect allows the opposite-direction trade to sell higher (or buy
lower) than they otherwise would have. Moreover, those intermediate opposite-
direction trades reduce the profitability of the sandwich. This is the reason why
an economically rational sandwicher strives to reverse their position before
anyone else does that. Thus, crucially, non-sandwiched users are not adversely
affected. And if they are not adversely affected – in fact, they benefit – then it
would be unlikely that they would be deemed to have been recklessly misled by
the sandwicher. As a result, a successful cause of action for fraud-based
manipulation likely could not be brought against the sandwicher on this theory.
By contrast, if a sandwicher – besides their aim to sandwich – also intends
to take a longer-lasting position in the market (and thus intentionally does not
fully reverse the front-run), then there would be little to doubt that the
sandwicher’s lasting market position is a legitimate trade, just with the use of a
front-running transaction (akin to arbitrage with the use of front-running208). Of
course, if this lasting market position is part of a manipulative scheme, then that
207 Id.
208 See infra Section IV.D.
This suggests that the most likely way in which a sandwich strategy – for
both “standard” sandwiches (economically optimal) or non-standard ones –
could generate liability under Rule 180.1 is this: it could be found to constitute
manipulation via at least recklessly affecting the price for the asset at issue in
the sandwiched user’s trade in an artificial manner. The Second Circuit clarified
that, in determining what constitutes a manipulative act, “[f]or market activity
to ‘artificially’ affect a security’s price, we generally ask whether the transaction
or series of transactions ‘sends a false pricing signal to the market’ or otherwise
distorts estimates of the ‘underlying economic value’ of the securities traded.”209
The most obvious implication of a sandwich front-run is the worse
execution price received by the sandwiched user. Yet, this alone is by no means
indicative of whether the practice is “artificial” – trading in traditional financial
markets is characterized by zero-sum games in which one trader’s gain is another
trader’s loss.210 Rather, courts generally define an artificial price as a price which
“does not reflect basic forces of supply and demand.”211 The court in CFTC v.
Cox held that a price is considered artificial where it is “affected by a factor
209 Set Capital LLC v. Credit Suisse Grp. AG, 996 F.3d 64, 79 (2d Cir. 2021).
210 What is a zero-sum game between the two directly involved traders may at the same time be
positive-sum for the market (or more broadly: from the perspective of social welfare), e.g., by
contributing to effective price discovery.
211 Cargill, Inc. v. Hardin, Secretary of Agriculture, 452 F.2d 1154, 1163 (8th Cir. 1971).
which is not legitimate.”212 Yet, no binding tests exists for determining which
forces or factors informing a price are legitimate and which are not.213
This suggests that in considering whether sandwiches create artificial
prices, or have an artificial price effect, it is conceivable that courts or regulators
might draw on background moral assumptions in order to decide whether the
prices obtained by sandwiched users – although technically consented to by way
of the user’s preselected slippage limit – reflect only the forces of supply and
demand or rather are artificial prices because they are influenced by a force that
is not legitimate. This would be an instance of an approach to conceptualizing
market manipulation which gives pride of place to moral criticisms. As Fox et
al put it, in such cases “the normative criticism of the relevant conduct is doing
all the work in identifying exactly what kind of behavior is supposed to be
prohibited.”214
To take one example of how such moralized reasoning could play out, the
court in Set Capital found that plaintiffs had adequately alleged that the
defendant bank (an issuer and seller of the relevant securities) had “knowingly
or recklessly exacerbated [a] liquidity squeeze” in futures market through its
trading behavior, which it knew would affect prices of certain derivatives that
its hedge positions depended on, and the court further held that this was enough
to constitute evidence of “conscious misbehavior or recklessness,” as required
for a finding of “manipulative intent.”215 Thus, the court reasoned, in effect, that
the defendant bank’s actions had an illegitimate price effect because they
“exacerbated” a liquidity squeeze, which falls on the wrong side of the
distinction between creating a liquidity disturbance for one’s own benefit and
merely taking advantage of such a disturbance that already existed. Professor
Stuart Green argues that this distinction between creating a disturbance versus
merely taking advantage of one that already exists is key to separating unfair or
illegitimate trading practices from those that do not violate norms of fair play.216
Although it is not certain a court or regulators will approach the inquiry by
applying this sort of moralized reasoning about legitimate versus illegitimate
price factors, it is conceivable that decisionmakers might take the sandwicher’s
212In re Cox, [1986-1987 Transfer Binder] Comm. Fut. L. Rep. (CCH) ¶ 23,786 at 26,060 (CFTC
(CCH) 21,796, at 80-81,281(CFTC Dec. 17, 1982)(Johnson, C., concurring) (“Legitimacy with
respect to supply and demand is undefined in law and economics, unless the sole question is
whether the forces were put in motion by an illegal act”).
214 FOX, GLOSTEN, AND RAUTERBERG, supra note 21 at 339 n.10.
215 Set Capital LLC v. Credit Suisse Grp. AG, 996 F.3d 64, 79 (2d Cir. 2021).
216 STUART P. GREEN, LYING, CHEATING, AND STEALING: A MORAL THEORY OF WHITE COLLAR
CRIME 242 (2006) (discussing, in the insider trading context, “[t]he distinction between creating an
unfair informational disparity and exploiting an informational disparity that already exists. (…) The
disclose or abstain rule should be construed so as not to apply to cases in which an investor comes
across non-public information fortuitously (say by overhearing it in an elevator or on the train)”).
gas settings for their transactions, to place them in front or behind some target
transaction, as needed. Such a searcher still pays a rent to the proposer (in the
form of the “tip” part of a transaction fee), but the rent may not be proportionate
to the value of the profitable opportunity. Instead, the rent is based the
transaction fee set by the target transaction (e.g., the transaction to be
sandwiched). However, this is unlikely to be significant as profitable
opportunities – especially for sandwiching – tend to be competitive. This
competition may take place through the auction process described above or
through so-called priority gas auctions.218 In both cases, the proposer’s rent will
tend toward the total value of the opportunity (although more efficiently in overt
auctions than in priority gas auctions). (Note that validators-proposers in
principle could attempt to avoid participating in, or facilitating, MEV
extraction.219
Illegitimacy of relying on control of block production for sandwiching. It
could be argued that, insofar as sandwiching leverages control over blockspace
as described above, one might see the validator-proposer’s involvement in
sandwiches as exploiting a key part of the financial infrastructure on Ethereum
to give some market participants an advantage that others. In this way, their
position as being in charge of the machinery of processing transactions gives
validators an unfair advantage, which users generally do not and cannot have
unless they too become validators in charge of building blocks. We propose that
there are two main ways in which courts and regulators engaged in a moralized
form of inquiry to determine where manipulation exists might deem the
exploitation of one’s control over blockspace through sandwiching to constitute
an illegitimate price factor, thus artificially affecting prices. One builds on
intuitions about market power (or dominance), while the other flows from
intuitions about conflicts of interest.220
Market power (dominance) argument. First, one might see facilitation of
sandwiching by validator-proposers as exploitation of a key position of
(temporary) monopoly power within the machinery of processing transactions,
giving those exploiting the power an unfair vantage point in the ecosystem from
which to extract rents for themselves. While it’s true that anyone could in
principle become a validator (just as anyone could in theory become a landlord
who charges rents), this doesn’t change the fact that validators engaged in
sandwiching exploit their privileged position within the
ecosystem/infrastructure of the financial structure. Given that market power or
proposer’s power in cooperation with the proposer (relays, block-builders, searchers) for
sandwiching, may also be liable.
or immoral queue-jumping, this, we think, would be a clear mistake. The reason is that before the
relevant transactions are selected and ordered by a validator in order to become one block, there is
no pre-existing transaction ordering which could allow the sandwiched user to claim that the
sandwicher cut in line in front of them. That is, transaction order in this context is not established
through a first in time rule, but rather through the auction process, whereby higher bids lead to
greater prospects of having one’s transactions executed earlier. This was by design, as a first in
of intuitions provide the strongest basis for the manipulation argument in the
present context.
(iv) Rejoinders
Let us close this section by considering some rejoinders. First, and most
fundamentally, courts and regulators might eschew the form of moralized
inquiry outlined above in their approach to determining what constitutes
prohibited manipulation. Thus, they might instead prefer to take an empirical
look at what behaviors produce efficient markets in traditional economic terms
in order to decide what behaviors should be deemed manipulative.225
However as argued earlier,226 much more empirical work is needed to
determine the extent to which MEV sandwiching is a behavior with net positive
effects for the ecosystem construed in the aggregate. Hence, the liability
outcome on this approach remains uncertain.
The second sort of rejoinder one might raise would arise within the
moralized inquiry we gestured at above and argue that the domination and
conflict of interest arguments sketched a moment ago are not decisive because
of other conflicting normative principles. Let us consider a few of these, though
we do not propose to resolve the matter conclusively.
First, then, one might respond that the price impact of a sandwich is quite
similar to Uber’s ‘surge pricing’ model, wherein Uber’s pricing model
algorithmically increases prices at times where potential Uber customers may be
willing to pay more.227 So, to the extent that a sandwicher changes the execution
price of a sandwiched user’s transaction through their front-run, they do so
solely on the basis of a user’s publicly expressed price preferences (i.e. their
slippage limit), rather than any false or deceptive information. Nonetheless, one
might see a difference between price effects due to sandwiching and the Uber
surge pricing model. Surge pricing benefits are meant to be redistributed to
drivers to ensure there is enough driver supply to meet rider demand.228 In the
time rule for transaction ordering would have invited the sort of race to be first that one sees in
latency arbitrage from high frequency traders. Accordingly, given the current architecture for
transaction ordering through auctions, there is no plausible argument that the sandwicher’s front-
run is unfair or sends an illegitimate price signal due to “cutting in line” in front of the sandwiched
user’s trade.
225 Many scholars have advocated for such an approach. See, e.g., Pirrong, supra note 88 (arguing
for an approach to manipulation determinations which focuses on the economic effects of market
manipulation and statistical analysis); Abrentez-Metz, supra note 113 at Part III (proposing the
use of empirical economic and statistical tools, such as econometric screens, in the evaluation of
manipulation claims, particularly at the pleading stage).
226 See also Barczentewicz, supra note 18.
227 Brett Helling, Surge Pricing: What It Is & How It Works For Riders & Drivers, Ridester (Jan.
sandwiching case, the extra profit generated in the sandwich is shared among
those involved in sandwiching (searcher, block-builder, relay, proposer-
validator), with the most value like to go to the proposer, i.e., the actor who
controls a key piece of financial infrastructure and extracts rents for themselves
through it. On one view, that would be as if Uber raised prices in times of great
rider demand such as a natural disaster or other crisis solely because the market
would bear it – arguably a form of price gouging.229 Outside of disaster
situations, however, one might argue that even if all the monetary value from
competition for space (cars) accrues to Uber or to sandwichers, users may derive
a significant non-monetary benefit in being able to express the strength of their
preference for the use of space and thus have their preferences satisfied
according to their strength. Of course, this in turn raises questions about the
accuracy of willingness-to-pay as a proxy for well-being effects, particularly for
actors with limited ability to pay.230 But the point is that one might bring duelling
normative principles to bear in order to debate the legitimacy of the relevant
form of rent extraction.
Second, and more importantly, one might counter the domination and
conflict of interest arguments we introduced above by contending that the profit
motive with which MEV sandwiching is pursued provides a legitimate economic
rationale for sandwiching by those who control the construction of blocks. As
some courts have maintained, trading with the purpose of receiving the best
possible price for oneself is a “legitimate economic rationale,” even where it
detrimentally impacts another trader.231 Still, it is clear that this principle must
have limits, as any paradigmatically manipulative trading strategy like
spoofing232 or banging the close233 is also pursued primarily from a profit
229 Ryan Lawler, Uber Caps Surge Pricing During Emergencies Nationwide (Jul 8th 2014)
re Indiana Farm Bureau Coop. Ass'n, Inc., [1982-1984 Transfer Binder] Comm. Fut. L. Rep. (CCH)
21,796, at 24,281(CFTC Dec. 17, 1982). (“It is imperative that each [side] of the market seek the
best price in order for price discovery to occur and that ‘best’ price is, of necessity, at the expense
of the other side”).
232 United States v. Coscia, 4 F.4th 454, (7th Cir. 2021) (upholding conviction for commodities
market, prices reflect the competing judgments of buyers and sellers as to the fair price of a
commodity or, in this instance, swaps. Here, acting on behalf of JPMorgan, the…traders’
activities…constituted a manipulative device in connection with swaps because they sold enormous
volumes of [a credit default index] in a very short period of time at month-end.”); CFTC v. Wilson,
No. 13 Civ. 7884 (RJS), 2018 LEXIS 207376 (S.D.N.Y. Nov. 23, 2018) at *57 (defining ‘banging
motive, and this alone does not suffice to bless the relevant form of activity.
Something more than the appeal to a profit motive for the specific transaction in
question is needed to separate permissible transactions from those that form part
of a manipulative scheme premised on illegitimately produced artificial prices,
though it is no easy matter to specify with perfect generality precisely what more
is required. In the end we suspect that the matter will be decided by the extent
to which courts wish to protect notions of market integrity that exclude
domination based on control of infrastructure that most traders do not have
access to but all rely on.234 While we think the argument will surely continue to
rage on, this strikes us as the most plausible going theory of liability for open
market sandwiching in general sounding in manipulation premised on
illegitimate price effects.
Nonetheless, there is a final problem, which is of particular practical
importance: we suspect that liability for standard open market sandwiching is
unlikely to be pursued as a regulatory priority in the near term. In principle,
someone could argue that an atomic sandwich constitutes a naked open-market
manipulation scheme because the atomic condition of the sandwich equips the
sandwicher with a “good reason to believe” that the average price at which they
purchased (or sold) a crypto asset (in the front-run) would be larger than that
when they sold (or purchased) the crypto asset (in the back-run) due to the price
impact of the sandwiched transaction.235 Nonetheless, very few cases – to our
knowledge, only one – have found fraud-based manipulation liability on the
basis of naked open market manipulation schemes alone.236
Our main explanation for why this is the case simply boils down to a matter
of enforcement discretion. The limited precedent with respect to CFTC or SEC
enforcement against naked open-market manipulation could simply stem from
the fact that these practices are low on the totem pole of enforcement priorities
given the high evidentiary costs associated with establishing the requisite
scienter which differentiates legitimate from manipulative naked open market
the close’ as involving “someone putting in a disproportionate number of trades to push the price
up or down to affect the closing price, typically in a noneconomic fashion, to benefit a position that
they held elsewhere”).
234 Lindsay Farmer, Taking Market Crime Seriously, 42 Legal Studies 508, 523-24 (2022).
235 Fox, Glosten, and Rauterberg, supra note 94 at 74 (stating that naked open market manipulation
schemes should be the subject of legal sanctions “only where it can be proved that [the trader] had,
at the time of her purchase, good reason to believe that" an asymmetric price reaction would occur
on the basis of their trades).
236 See Markowski v. SEC, 274 F.3d 525 (D.C. Cir. 2001).
trades.237 This would explain why practices like predatory trading,238 which
seem to meet the definition of naked open-market manipulation, are left
unchallenged by the agencies.239 As a result, a claim of fraud-based
manipulation liability on the grounds that sandwiching constitutes an open
market manipulation scheme involving at least recklessly creating artificial
prices would be relatively unlikely to be brought in practice even if theoretically
there is a plausible argument for this conclusion to be tested.
Of course, more debate always remains possible. As explained above,240
naked open-market market manipulation is characterized by the lack of relations
of trust between the party accused of manipulation and other traders, while
covered market-manipulation (such as paradigmatic banging the close or oracle
manipulation) do involve some such trust relations, which are at least implied
(e.g. based on a formal mechanisms for pricing one asset that references
another). The considerations adduced above, about block builders’ control and
exploitation of crypto market infrastructure, on which other traders rely, might
similarly be seen as generating implied relations of trust, which could move
MEV sandwiching of public trades more in the direction of covered open-market
237 See, e.g., Id. at 530 (“We cannot find the Commission’s interpretation to be unreasonable in
light of what appears to be Congress’s determination that manipulation can be illegal solely because
of the actor’s purpose”); SEC v. Masri, 523 F. Supp. 2d 361, 373 (S.D.N.Y. 2007) (“The court
concludes, therefore, that if an investor conducts an open-market transaction with the intent of
artificially affecting the price of the security, and not for any legitimate economic reason, it can
constitute market manipulation”); In re Amaranth Natural Gas Commodities Litig., 587 F. Supp.
2d 513, 540 (S.D.N.Y. 2008) (“Entering into a legitimate transaction knowing that it will distort
the market is not manipulation – only intent, not knowledge, can transform a legitimate transaction
into manipulation”).
238 Predatory trading involves a strategic trader becoming aware (on the basis of public or private
information) of an impending large transaction by another trader. For instance, a hedge fund with
a nearing margin call may need to make a large liquidation. In response, the strategic trader first
trades in the same direction as the distressed trader and subsequently reverses his position following
the distressed trader’s large transaction. As such, the strategic trader effectively profits by forcing
the distressed trader to suffer a worse price on their large transaction. For an in-depth look at
predatory trading, see Markus K. Brunnermeier and Lasses Heje Pederson, Predatory Trading, 60
J. of Fin. 1825, 1825 (August 2005).
239 See Procedures To Establish Appropriate Minimum Block Sizes for Large Notional Off-Facility
Swaps and Block Trades, 78 Fed. Reg. 32870, n.46 (May 31, 2003) (discussing predatory trading
as a consequence of a regulatory regime requiring the ”publication of detailed information
regarding ‘‘outsize swap transactions’’); Alternatively, it may be the case that courts are hesitant
to find fraud-based manipulation liability on the basis of purely naked trades because there is some
implicit theory that a contractual relationship, or other kind of relationship that generates
expectations between parties, makes some kinds of conduct manipulative which otherwise are not.
As applied to open-market trades, it would follow from this theory that covered open-market
manipulation schemes are more likely to be manipulative because they generally involve some
contractual relationship or expectation-setting between parties to a transaction, while the same
cannot be said for naked open market manipulation.
However, we explore this issue in more depth below.
240 Supra Section III.C.
(v) Front-running
and “to employ reasonable care to avoid misleading” clients.244 Such duties may
apply not just to clients with personal relationships with an adviser. In Capital
Gains Research Bureau mere recommendation of the adviser’s own investments
in a newsletter, without appropriate disclosure (scalping), was held to be a
breach of the advisers’ legal duties. Much more could be said about the special
cases of investment advisers and brokers but given that they do not appear to be
prominent among MEV extractors, we will not do so here.
An MEV searcher who relies only on public information about a pending
transaction and has no other knowledge about the transaction seems unlikely to
have the special duty.
Unclear whether sandwiching would breach the duty in all cases. It could
be that being involved in sandwiching of a user to whom one does have the
special duty discussed here, would constitute a breach of that duty and thus
illegal front-running. But this need not be so, if it could be shown that the
practice is fully disclosed and that it is in the best interest of the user. The
consideration of best interest could include aspects like “MEV rebates” or other
services offered to a user (like faster execution) in return for assent to
sandwiching.
Front-running as misappropriation of private information. By definition,
sandwiching a public transaction does not involve the use of private information,
hence the “insider trading” version of front-running is not applicable here. We
will, however, come back to this point while discussing sandwiching of private
transactions below.
248 Id.
249 Id.
guarantee that a portion of the MEV extracted from users’ POF will actually be
given back to the user.250 Alternatively, a user may route their transaction
through a privacy RPC out of fear that sending their transaction to the mempool
may serve as blood in the water, luring searcher bots who seek to exploit their
trade and cause them to suffer a worse execution price. Rebate programs also
tend to promise users that, in addition to MEV redistribution, the users will be
protected from front-running (including sandwiching), and will receive “best
execution.”251 POF arrangements generally involve off-chain communication
wherein searchers and block-builders with access to POF make promises (either
explicitly or implicitly, through their integration with a privacy RPC which
advertises itself as offering privacy, front-running-protection, etc. as a service)
to users in exchange for exclusive access to their order flow. Hence, POF is
appealing for users because it provides them with both i) monetary rewards and
ii) certain non-monetary guarantees with respect to their transaction such as pre-
execution privacy and front-running protection.
When a user exclusively routes their order flow to a particular block builder
(or set of builders), they are also not publicizing their transaction to the rest of
the network. If the basic guarantee of a privacy RPC is not broken,252 and the
user herself does not submit the same transaction to a non-privacy (ordinary)
RPC, then the information about a pending transaction submitted to a privacy
RPC remains non-public information until the transaction is included on the
blockchain.
The implicit and explicit promises made by block builders and searchers to
the users who provide them with POF, we argue, create a relationship of trust
between those operators and those users.253 That is, users providing their POF to
network operators trust that their transactions will remain private, will not be
front-run (including sandwiching), and more. The involvement of non-public
information and the existence of a relationship of trust thus paves a route to
insider trading liability for block builders and searchers who sandwich user
trades they received as POF.
Recall that both the CFTC and SEC police insider trading in their
associated markets as a form of fraud-based manipulation prohibited under Rule
180.1 and Rule 10b-5, respectively.254 Both agencies rely on the
misappropriation theory of insider trading, which finds liability for trades made
accessed Jan. 30, 2023) (promising that “Ethermine will never leak nor act on the information
received via this relay”); Frontrunning Protection; BLOXROUTE DOCUMENTATION,
https://docs.bloxroute.com/apis/frontrunning-protection.
253 For an analysis of the trust relationships endemic to blockchain based systems - including that
defendants’ argument that the misappropriation theory applies only “when an individual owes a
fiduciary duty to the principal whose information was allegedly misappropriated”).
259 Verstein, supra note 146 at 32.
260 See, e.g., supra note 252.
261 EOX Holdings L.L.C., 405 F. Supp. 3d at 713 (discussing the conception of “duty” defined by
the court in SEC v. Cuban (Cuban I) -- undisturbed by the Fifth Circuit in Cuban II -- that “insider
trading could arise where there was an express … agreement to maintain the confidentiality of
material, nonpublic information and not to trade on or otherwise use the information”).
262 Verstein, supra note 146 at 15. See also United States v. O'Hagan, 521 U.S. 642, 670(1997)
(holding that “"the deception essential to the misappropriation theory involves feigning fidelity to
the source of information”).
263 SEC v. Cuban, 620 F.3d 551, 555 (5th Cir. 2010); CFTC v. EOX Holdings L.L.C., No. H-19-
2901, 2021 WL 4482145, at *53 (S.D. Tex. Sept. 30, 2021) (“Defendants’ argument that ‘there is
no need for an exegesis on the law of duty in this context,’ fails to recognize that courts have been
struggling with this precise issue for some time”).
264 See 17 C.F.R. §180.1(a)(2) (2012).
265 The court in CFTC v. Kraft noted that fraud-based manipulation claims “based upon a
misstatement” require the plaintiff to allege a material misrepresentation (or omission), scienter,
connection with the purchase or sale of a commodity, reliance, economic loss, and causation.
United States CFTC v. Kraft Foods Grp., Inc., 153 F. Supp. 3d 996, 1012 (N.D. Ill. 2015), quoting
Dura Pharms., Inc. v. Broudo, 544 U.S. 336, 341 (2005).
266 Ploss v. Kraft Foods Grp., Inc., 197 F. Supp. 3d 1037, 1058 (N.D. Ill. 2016)(“The court agrees
p and is broadcast globally, a competitive game naturally ensues among arbitrage bots to be the
first to execute an atomic transaction that exploits the opportunity”).
268See Quintus, Order flow, auctions and centralisation I: a warning,
https://collective.flashbots.net/t/order-flow-auctions-and-centralisation-i-a-warning/258.
269 In re Cox, ¶ 23,786 at 12-13, 060 (“the acquisition of market dominance is the hallmark of a
long manipulative squeeze”); Resch-Cassin & Co., 362 F. Supp. at 977 (“dominion and control of
the market for the security” are factors establishing causation of an artificial price).
270 The exercise of market dominance in a commodities market is often seen as a paradigmatic form
of price manipulation. See, e.g., Pirrong, supra note 88 at 947 (“The most important form of
manipulation consists of the exercise of market power in a commodity market”).
difficult to make out given its higher requirement of intent to cause an artificial
price. But Rule 180.1’s lower standard of recklessness toward the creation of
prices not reflecting only the forces of supply and demand, as involved in
180.1(a)(3) manipulative acts, may offer a more achievable route to imposing
liability for exploiting a position of market dominance obtained through access
to POF in the manner just described. While price manipulation liability is more
plausible in this context relative to the context of sandwiching public
information, it is still relatively unlikely given the prosecutorial burdens
involved in establishing price manipulation.
Recall from our discussion in the previous section that an Ethereum user,
submitting transactions otherwise than to a privacy RPC, may think that their
pending transaction is public while, in reality, it is not meaningfully so. In
general, there are two ways in which this could happen. First, the default RPC
used by the wallet software could treat such transactions as private, not re-
broadcasting them in the Ethereum peer-to-peer network, instead forwarding
them, e.g., only to selected, cooperating block-builders. Second, the default
RPC, or even the wallet software provider herself, could delay forwarding
transactions long enough to have a sufficient advantage in extracting MEV from
those transactions.
In traditional finance, payment for order flow (PFOF) refers to an
arrangement in which an “internalizer” (e.g., Citadel) pays a broker (e.g.,
Robinhood) in exchange for the broker routing her retail clients’ orders to the
internalizer, who then executes trades against those orders.271 In an analogous
fashion, service providers like wallet software providers and RPC operators can,
in a sense, sell the order flow of their customers to MEV extractors, acting as
internalizers.272 A wallet software provider could do this by setting an
internalizer-operated RPC as the default option, which could – but in practice
may be unlikely to be – changed by users. But this could also be done by not
giving users the choice as to which RPC the wallet software will use. Finally, a
wallet software provider could delay sending transactions to a public RPC, while
sending the information about new transactions to paying internalizers first, thus
guaranteeing them an advantage in extracting MEV. A popular RPC operator
may be able to use similar methods (exclusive transaction forwarding, temporal
priority in forwarding).
Both in the standard case of non-explicit POF, and in its special case –
PFOF arrangements, user transactions constitute, at least temporarily, non-
motivated by a desire to obtain compensation rather than by a desire to affect a market price, if the
trader recklessly effected the manipulative trades, he will be held liable”).
arrangements.278 While not directly applicable in this DeFi context, this rule
plausibly indicates a policy judgment that some fiduciary or other trust-based
obligation exists requiring actors with privileged access to transaction
information to act honestly and transparently with respect to the actual
originators of that transaction information. It would not be unreasonable to
assume a similar policy judgment might be implemented for DeFi as CFTC and
SEC begin to flex their regulatory muscles in the crypto space more fully in the
near term.
consequences for trades that create an arbitrage opportunity, as well as whether the competition for
exploiting arbitrage and liquidation opportunities can be said to be fair (non-manipulative) when it
entails leveraging the privileged position of a block producer.
Though this may change, currently the paradigmatic case of the use of
oracle manipulation is to attack on-chain lending systems; notably, but not only,
to create artificial loan liquidation opportunities.284 For on-chain liquidations to
work, lending systems require external information about prices of the assets
involved. Technically, the smart contracts that are the lending systems rely on
other “oracle” smart contracts. The oracle smart contracts are meant to faithfully
report prices (or other information), e.g., based on what happens on some other
on-chain (DEX) or off-chain (CEX) markets.
Hence, a liquidation opportunity can be created artificially in two
scenarios: i) if an oracle can be made to report an incorrect (low) price of the
asset used as collateral or ii) if the benchmark markets from which an oracle
collects price information can be manipulated. Here, we only focus on the
second strategy, as it is a more central case of market manipulation, without the
complicating factor of hacking that would likely be needed for the first strategy.
Strictly speaking, the second strategy does not manipulate the oracle itself,
because the oracle functions as intended: it is just that the prices that it reports
may result from manipulation. It may thus be more precise to refer, e.g., to “loan
liquidation price benchmark manipulation,” but the label of “oracle
manipulation” is already established.285
Not all oracle manipulations involve MEV extraction. Yet, as Mackinga et
al. show, the profitability of a self-created liquidation opportunity is often
contingent on control over transaction ordering.286 That is, where a liquidation
opportunity is triggered by an instance of oracle manipulation, it is publicly
visible and able to be captured not only by the oracle manipulator but also by
any other strategic actors paying attention. Because oracle manipulation
generally requires a significant initial investment to create a sufficiently large
change in the benchmark price, a rational manipulator will only expend that
capital if they harbor a degree of certainty that i) they will be the first to execute
the profit opportunity created by their oracle manipulation, and ii) that they will
be able to reverse (de-manipulate) their oracle manipulation transaction in order
to redeem the upfront capital expended.287
284 For a general description of loan liquidations, see supra Section II.C. For in-depth discussions
regarding the mechanics of how price oracles can be manipulated in decentralized finance, see
Austin Adams, Xin Wan, and Noah Zinsmeister, Uniswap v3 Price Oracles in Proof of Stake,
ᴜɴɪsᴡᴀᴘ ʙʟᴏɢ (Oct. 27, 2022), https://uniswap.org/blog/uniswap-v3-oracles; Torgin Mackinga,
Tejaswi Nadahalli & Roger Wattenhofer, TWAP Oracle Attacks: Easier Done than Said?, in 2022
IEEE INTERNATIONAL CONFERENCE ON BLOCKCHAIN AND CRYPTOCURRENCY (ICBC) 1 (2022).
285 See also Barczentewicz, supra note 18 at 12–14.
286 See generally Mackinga, Nadahalli, and Wattenhofer, supra note 284.
287 Id. at 2. (“the attack’s profitability ... rests on whether the [manipulator] can de-manipulate the
price [of the collateralized crypto asset] back to market price without other users front-running the
[manipulator]”).
reliant on a price oracle affected by the DEX price, that trader has engaged in a
form of covered open-market manipulation.
Notably, oracle manipulation is one of few DeFi market practices which
has been directly challenged by the CFTC and the SEC.292 While Avi
Eisenberg’s manipulation of Mango Markets was not a direct result of MEV
extraction, the instance of multi-block MEV described above may produce
analogous results. As such, there is a high likelihood that oracle manipulation
strategies triggering liquidations and other profitable, publicly viewable
opportunities -- thus implicating MEV extraction -- would be considered
impermissible manipulative trading practices.
substantively inquired into the efficiency effects of sandwich trades has done so
through the lens of algorithmic game theory, rather than empirical analysis.294
Notably, this research found that sandwich trades, under some circumstances,
can actually increase market efficiency and social welfare at the network level
by causing more effective transaction routing patterns295 and stronger incentives
for validators to stake (i.e. participate in block construction).296 This research
indicates that limiting one’s analysis of the market efficiency implications of
sandwich trades and other MEV extraction practices to their purely local
implications on individual trades is insufficient, as the individual harm to a
sandwiched user may be outweighed by the market-wide efficiency gains
produced by sandwich trades considered in the aggregate. Yet, because this
research is theoretical in nature, we cannot currently predict the extent to which
the social welfare and market efficiency gains it proposes does in fact affect the
market. Further research on this point is crucial for the instant policy question.
Complicating the efficiency calculus further are the differences in
underlying settlement technologies and market structures employed by DeFi and
traditional financial systems. For instance, the security and strength of a
blockchain’s distributed consensus mechanism – i.e. for Ethereum, the network
of independent and honest validators – can have significant implications for the
efficiency of a DEX which is built on top of that blockchain. Meanwhile, there
is no direct analogue for such a consideration in traditional markets.297
Additionally, the scope of a “market” for the purposes of a market efficiency
analysis is broader in the context of MEV extraction than an examination of this
concept in traditional finance may suggest. 298 In traditional finance, one thinks
of a “market” where manipulation can occur as any locus of the buying, selling,
and lending of a financial instrument. Yet, MEV can be extracted in blockchain-
based markets beyond the realm of DeFi – MEV extraction can arise in the
context of non-fungible token (NFT) sales,299 and other decentralized
application functionalities. As such, it may prove advisable to broaden the scope
of a market efficiency analysis for MEV extraction beyond what would have
been standard in traditional finance.
294 See generally Kulkarni, Diamandis, and Chitra, supra note 12 at 22–24.
295 Id.
296 See generally Tarun Chitra & Kshitij Kulkarni, Improving Proof of Stake Economic Security via
market’s efficiency still exists, but is more indirect and fragmented than this connection is in
decentralized crypt oasset markets. In traditional finance, market structure factors like supply chain,
clearing/settlement processes, and trading rules implicate market efficiency considerations like a
market’s competition, liquidity, transparency, and price formation. Whereas, in the context of
decentralized crypto asset markets, there is a more direct and coherent link between network/
consensus security and market efficiency.
298 Barczentewicz, supra note 18 at 17.
299 See supra note 75.
50% of the nodes in a blockchain network altering the blockchain record. While PoS blockchains
like Ethereum are often considered resistant to 51% attacks, it is still possible for a malicious
validator who acquires a large proportion of Ethereum’s total stake to engage in a profitable 51%
attack. For instance, this can be done through short-selling. See generally Suhyeon Lee and
Seungjoo Kim, Short Selling Attack: A Self-Destructive But Profitable 51% Attack On PoS
Blockchains, CIST (2020), https://eprint.iacr.org/2020/019.pdf.
303Double-spending occurs on blockchain networks where the same crypto asset is spent more than once;
see Nakamoto, supra note 22.
304 Ethereum Market Cap, YCHARTS,
https://ycharts.com/indicators/ethereum_market_cap#:~:text=Ethereum%20Market%20Cap%20is
%20at,36.70%25%20from%20one%20year%20ago ; https://coinmarketcap.com/ (accessed
January 29 , 2023).
borderless in nature, a U.S. ban on sandwich trades would still leave U.S. users
vulnerable to the activity of sandwichers from other jurisdictions – particularly
where the enforcement of U.S. laws are difficult (either due to restrictions on the
extraterritorial application of securities laws or the difficulty of enforcing a
judgment from U.S. courts in certain foreign jurisdictions).305 Moreover, there
are well-known difficulties with enforcing securities laws within crypto markets
in particular, given the heightened levels of anonymity these markets currently
involve. Together these enforcement difficulties likely limit the benefit to
consumer protection that a broad ban on certain forms of MEV extraction would
have in practice.
Accordingly, rather than jumping the regulatory gun with a sweeping ban,
it is crucial to explore whether consumer protection concerns surrounding
sandwich attacks might not be better assuaged through technical solutions
tailored to global blockchain networks, as opposed to regulatory measures bound
by jurisdictional limits.306
305 See, e.g., Rebecca Cloeter, The Extraterritorial Application of the Commodity Exchange Act, 41
externalities of MEV extraction, see Sen Yang, Fan Zhang, Ken Huang, Xi Chen, Youwei Yang,
and Feng Zhu, SoK: MEV Countermeasures: Theory and Practice (Dec. 2022),
https://arxiv.org/pdf/2212.05111.pdf.
307 See supra Section I, Section IV.A.
saw questions remain about the benefits to consumer protection that such
prohibitions can produce. More empirical research on both questions is
imperative.
Where does this leave the debates both about permissibility under existing
law of MEV extraction and the proper policy response thereto? To guide future
conversations, we close with four general recommendations:
1. Focusing regulatory attention on MEV extraction practices that cause
harm: The phenomenon of MEV extraction presents far more nuance than
initially meets the eye. As discussed in Section IV, the publicness of transaction
information, the atomicity of a MEV extraction opportunity, and the trust
relationship between a user and MEV extractor are all crucial factors for
determining whether a particular instance of MEV extraction is manipulative.
Moreover, as we saw, there are myriad varieties of MEV extraction.
Accordingly, given all the technical and legal nuance that this presents, we
warn against viewing “MEV extraction” as a singular phenomenon, and
recommend an approach that recognizes that there may be as many different
varieties of MEV extraction as there are opportunistic trading practices in
traditional finance. Thus, it is not really “MEV extraction” that should be the
subject of regulatory attention, but rather specific MEV practices which harm
ordinary traders or the market at large. For instance, while DEX arbitrage – akin
to cross-exchange arbitrage in traditional markets – usually supports price
discovery without harming consumers, intentional self-created liquidations like
those involving oracle manipulation strike a close resemblance to covered open-
market manipulation schemes like “banging the close.”308 Likewise, the
presence of express fraud (i.e., a node operator, wallet operator, or block-builder
promising not to front-run or sandwich a user, and subsequently proceeding to
do so), or the existence of a relationship of trust that gives rise to legitimate
expectations that then are contravened, are also indicators of manipulative
conduct worthy of legal action. Accordingly, a regulator would understandably
opt to expend limited resources pursuing the more obviously harmful MEV
practices than to launch a broadside against MEV extraction in general, given
its multiplicity and pervasiveness.
2. Cross-Disciplinary Research: Knowing which MEV practices rise to the
level of harm that might make them a regulatory priority, however, requires
more empirical research, as seen in Section V. Thus, we recommend that
economists, technical researchers, and legal scholars engage in cooperative
research to better understand the market efficiency and social welfare effects –
on an individual and system-wide scale – of the many different strategies of
MEV extraction. Such research, particularly when focused on individually
harmful MEV extraction practices like sandwich trades, is the only way to shed
light on the metaphorical dark forest of MEV.
309 See, e.g., Pirrong, supra note 88; Janet Austen, What Exactly is Market Integrity? An Analysis
of One of the Core Objectives of Securities Regulation, 8 Wm. & Mary Bus. L. Rev. 215 (2017).
310 See, e.g., Piet, supra note 12 at 5; Kulkarni, supra note 12 at 4; What is Maximal Extractable