24-1221 Johnson v. United States 1 United States Court of Appeals 2 for the Second Circuit 3 _________________ 5 August Term 2024 7 Argued: March 18, 2025 8 Decided: July 17, 2025 10 No. 24-1221 11 _________________ 13 MARK JOHNSON, 15 Petitioner-Appellant, 17 v. 19 UNITED STATES OF AMERICA, 21 Respondent-Appellee.

23 _________________ 25 On Appeal from the United States District Court 26 for the Eastern District of New York, Garaufis, J.

27 _________________ Before: CALABRESI, NATHAN, and KAHN, Circuit Judges.

31 Petitioner-Appellant Mark Johnson appeals from a judgment of the United States District Court for the Eastern District of New York (Garaufis, J.) dismissing his Petition for a writ of coram nobis. Johnson was convicted in a general verdict after the government presented two theories of fraud to the jury, one of which was 24-1221 Johnson v. United States the now legally invalid right-to-control theory. Below, the government opposed his Petition on the grounds that the jury, in effect, also convicted Johnson on the legally valid misappropriation theory, rendering harmless the erroneous presentation to the jury of the right-to-control theory. We find that the government’s case against Johnson under the misappropriation theory is comparatively weak and have grave doubt that the presentation to the jury of the right-to-control theory was harmless.

8 Accordingly, we REVERSE the district court judgment and REMAND for entry of an order granting the Petition.

10 _____________________________________ 12 ALEXANDRA A.E. SHAPIRO, (Jason A. Driscoll, on the brief), 13 Shapiro Arato Bach LLP, New York, NY for Petitioner- 14 Appellant.

16 ANDREW W. LAING, Appellate Counsel, Criminal 17 Division, Fraud Section (Lisa H. Miller, Deputy Assistant 18 Attorney General, on the brief), for Nicole M. Argentieri, 19 Principal Deputy Assistant Attorney General, 20 Department of Justice, Washington, DC for Respondent- 21 Appellee.

23 (Jacqueline Jamin Drohan, Vivian Rivera Drohan, 24 Drohan Lee LLP, New York, NY for Amicus Curiae ACI – 25 The Financial Markets Association, in support of 26 Petitioner-Appellant.)

28 _____________________________________ CALABRESI, Circuit Judge: 31 Petitioner-Appellant Mark Johnson served two years in prison for wire fraud and conspiracy to commit wire fraud. Johnson filed this Petition seeking a writ of coram nobis after the Supreme Court’s decision in Ciminelli v. United States, 24-1221 Johnson v. United States 598 U.S. 306 (2023), rendered legally invalid one of the two theories of fraud liability presented to his jury which returned a general verdict convicting him. The government argued that submitting the invalid theory was harmless because Johnson was also convicted under the valid, alternate theory of fraud— misappropriation. The government’s misappropriation case against Johnson was weak. And we think it very unlikely that it was an independent basis for Johnson’s conviction. We therefore REVERSE the district court decision and REMAND for the district court to GRANT the Petition.

9 BACKGROUND 10 Mark Johnson was convicted in 2017 by a jury in the Eastern District of New York of wire fraud and conspiracy to commit wire fraud. The charges centered on a transaction Johnson conducted in 2011 as global head of HSBC’s foreign exchange trading desk, in which HSBC converted U.S. Dollars into 2.25 billion British Pounds for the oil and gas company Cairn Energy. The details of the case have been discussed extensively before, both by this Court, United States v. Johnson, 945 F.3d 606, 608–12 (2d Cir. 2019), and by the district court, Johnson v. United States, No. 23-CV-5600 (NGG), 2024 WL 1740916 at *1–5 (E.D.N.Y. Apr. 23, 2024).

24-1221 Johnson v. United States We therefore restrict our discussion to the, still lengthy, facts most pertinent to the present decision.

3 The Foreign Exchange Market 4 The foreign exchange (FX) market is a decentralized market for the trading of currencies. Unlike the stock market, it does not have a closing price. Instead, banks and financial services companies publish a “fix”—a benchmark exchange rate for each pair of currencies being traded. The relevant actors here used World Market/Reuters (WM/Reuters), which publishes its fix rate hourly, based on the average price of actual trades executed in a one-minute window beginning 30 seconds before the hour and ending 30 seconds after the hour. Movements in exchange rates are measured in “pips,” 100 pips being equivalent to one cent.

12 FX dealers, usually banks, can act as agents or principals. When a dealer acts as a client’s agent, the dealer functions as a middleman and trades with a third party on behalf of the client. For example, if a client wants to purchase pounds, the dealer, for a fee, will find a seller with the best price and purchase pounds on behalf of the client.

17 In contrast, when a dealer acts as a principal, the dealer transacts directly with the client to sell currency from its own inventory. There are multiple ways to 24-1221 Johnson v. United States structure transactions in which dealers act as principals. In a full-risk transfer, the dealer and client agree upon a set exchange rate, and the dealer sells the currency to the client at that rate. The dealer therefore bears the risk that the market value of the currency will change after the deal is made. In other words, it risks purchasing currency at a higher price than the price at which it agreed to sell. To compensate for that risk, the dealer charges the client a fee for executing the exchange.

8 On the other hand, in a fix transaction, the dealer agrees to sell currency to the client at a rate determined by the “fix” at a specific time set in the future. For publicly held clients, this method offers transparency: a corporation can show its shareholders that it received the market rate based on the fix. The dealer usually does not charge a fee for a fix transaction. Instead, it attempts to profit by “beating the fix”—that is, by purchasing the currency at a price below the fix rate at which it will sell the currency to the client.

15 A dealer in a fix transaction must first buy sufficient currency to fulfill the order. Ideally this is done slowly, in the several hours leading up to the transaction.

17 This accumulation of the currency in anticipation of the fix is called “pre-hedging,” and it helps protect against last-minute spikes in the price of the currency. If the 24-1221 Johnson v. United States market catches on to the dealer’s intentions and realizes the dealer needs to purchase, for example, several billion pounds, the market price of pounds will rise.

3 If a dealer buys currency in an aggressive manner just before a fix, this may lead to an increase of the fix price, thereby inflating the price the client must pay and with it the dealer’s potential profits. Doing this intentionally is referred to as “ramping” the fix. In any case, the dealer looks to purchase the currency in a fix transaction at a price lower than what it predicts the eventual fix price to be.

8 A dealer may also purchase currency for its own proprietary accounts and aim to sell to third parties immediately after the fix transaction, anticipating that the fix transaction itself may cause a momentary spike in the price of the currency from which it can profit. This is called “trading ahead.” The line between pre- hedging and trading ahead is, however, necessarily blurry because dealers simply do not know how many pounds, for example, they must buy to fulfill the client’s order until they see the final fix price. If dealers purchase too much currency, they will seek to sell it at the high point after the transaction, and if too little, they may have to reach into their own proprietary accounts to fulfill the order.

17 In the retail context, where dealers purchase currency entirely on their client’s behalf as their agent and receive a commission, pre-hedging or trading 24-1221 Johnson v. United States ahead is called front-running and is illegal. Dealers making such transactions on behalf of a client are not allowed to use the knowledge that the transaction may move the market to make a profit by making parallel trades in their own proprietary accounts. In principal-to-principal transactions, where dealers do not receive a commission, such conduct is, instead, explicitly not regulated. Indeed, pre-hedging and trading ahead is how banks profit and manage their own risk in FX transactions where they do not charge a commission.

8 The Contract Negotiations 9 In October of 2011, Cairn Energy sent a Request for Proposal (RFP) to nine major banks. Cairn planned to sell its subsidiary in India for around $4 billion, convert those dollars to British pounds, and pay out the proceeds to its shareholders. Cairn had never before been involved in an FX transaction of that size, and it sought proposals from banks about how to conduct the trade. Cairn retained the London investment bank Rothschild & Co. as its financial advisor, and Rothschild ran the RFP process. Each bank invited to participate in the process was required to sign a Non-Disclosure Agreement (NDA) stating that “[c]onfidential [i]nformation” was being provided “solely for the purposes” of “assist[ing] the . . . banks in their analysis of the proposed currency exchange 24-1221 Johnson v. United States transaction” so that Cairn could “[o]btain feedback” and “select” a bank. App’x at 166, 169, 172. The NDA stated that the agreement would remain in effect for two years. Id. at 167.

4 That month, Dipak Khot of HSBC pitched Cairn on several methods of executing the FX transaction, including the full risk transfer and fix transaction described above, as well as some methods (an “at best” price) where HSBC would serve as a broker or agent for Cairn and be paid a commission. The parties disagree on the extent to which the pitch should be read as “puffery” or “salesman’s banter” as opposed to binding promises about how HSBC would handle the trade if it secured the contract. The presentation stated: 11 • “We would like to execute this in the best interest of [Cairn].” App’x 12 at 177.

13 • “HSBC would work with you to ensure best execution during the 14 day.” Id. at 180.

15 • “It is important to choose a reliable partner[] who has a track record 16 and ability to seamlessly execute a transaction of this magnitude 17 without creating excessive market volatility.” Id. at 175.

24-1221 Johnson v. United States 1 • “[I]t is in the interest of [Cairn] to manage the process jointly with 2 HSBC in case of undue market volatility.” Id. at 177.

3 But it also said: 4 • “[Cairn] should not rely on any information in the document.” Id. at 5 204.

6 • A fix would “fully expose[] [Cairn] to any adverse movements” and 7 is among “the riskiest of the strategies to consider.” Id. at 178.

8 • “Neither HSBC nor any of its affiliates are responsible for providing 9 [Cairn] with . . . specialist advice.” Id. at 204.

10 • “[Cairn] is solely responsible for making [its] own independent 11 appraisal of and investigation into the . . . transaction.” Id. 12 While still considering which bank to move forward with, the Rothschild partner advising Cairn spoke to Petitioner-Appellant Johnson about a fix transaction and how it would be conducted. During the conversation Johnson explained: 16 [L]et’s say it’s, for argument’s sake, it’s 11 o’clock. In a perfect world 17 we start at 9 and we, or 8 in the morning, and we gradually build it 18 up, build it up, build it up and then just try and control the market so 19 it doesn’t look too noisy because obviously, you know our aim is to 20 make a small amount of money out of this clearly because that’s our 21 business. But, you know, have a happy customer go away.

24-1221 Johnson v. United States 1 ...

2 So if we just accumulate the position and then just try and control the 3 market over the last 20 minutes, so that we give you a fair price and 4 then we have something that you know, we keep Dipak and the 5 bosses happy, . . . Then, then it’s a win-win. If you’ve told us, you 6 know, we need the fix in 30 minutes then we’ve got a lot to buy and 7 we’re gonna cause a lot of noise and we run the risk one of losing a 8 lot and two of upsetting the customer. And that’s just not in our 9 interest.

11 App’x at 275.

12 The Rothschild advisor asked whether, if HSBC was able to buy the pounds at a price significantly better than the fix, it would be willing to share some of that profit with Cairn. Johnson said no and warned him that banks who offer a discount on the fix will just compensate by trading in a way that pushes up the fix: 16 Now if you’re doing that clearly you’re doing it cause you’re 17 intending to ramp the fix. . . . If we use the fix then we say actually we 18 can do it 10 pips better than that fix[,] it then opens a whole other 19 question[] as to why did you do it 10 pips better than the fix, you 20 obviously ramped the fix. So where would it have been if you hadn’t 21 ramped the fix and it just opens up a whole load of stuff. The point of 22 doing a fix is you basically look to buy the amount on average at or 23 around that fix at that time, which, which minimizes the ability of you 24 to really beat the fix by a really large amount anyway.

26 Id. at 277—78.

27 The parties have drastically different interpretations of this conversation.

28 According to the government, Johnson represented that HSBC would not trade in 24-1221 Johnson v. United States a way that would ramp the fix. According to Johnson, he was informing Cairn and Rothschild that HSBC would trade ahead of the fix which would naturally risk causing upward pressure on the price of pounds, and that the shorter the window the bank was given before the fix, the steeper the pressure would be. Johnson further argues that he made clear that HSBC would seek to profit by beating the fix.

7 In late October, Cairn awarded the contract to HSBC. The parties signed a Mandate Letter in which they agreed that HSBC would sell pounds to Cairn on a principal-to-principal basis, but the agreement left open the exact method to be chosen. If a fix were chosen, the agreement required Cairn to inform HSBC of the full amount needed at least two hours before the chosen fix time.

12 The Mandate Letter also included a number of disclaimers. It stated: 13 HSBC is not responsible for providing the recipient with legal, tax, or 14 other specialist advice and the recipient should make its own 15 arrangements accordingly. The recipient is solely responsible for 16 making its own independent appraisal of and investigation into the 17 products, investments, and transactions referred to in this document 18 and should not rely on the information in this document as 19 constituting investment advice.

20 ...

21 The issue of this document shall not be regarded as creating any form 22 of advisory or other relationship, and HSBC may only be regarded as 23 acting on behalf of the recipient as financial advisor or otherwise 24-1221 Johnson v. United States 1 following the execution of an engagement letter on mutually 2 satisfactory terms.

4 Id. at 211.

5 The Mandate Letter also incorporated an International Swap Dealers Association Master Agreement (ISDA) previously signed by the parties. The ISDA stated in relevant part: 8 Each party will be deemed to represent to the other party on the date 9 on which it enters into a Transaction that . . . : 10 Non reliance. It is acting for its own account and it has made its own 11 independent decisions to enter into that Transaction and as to 12 whether that Transaction is appropriate or proper for it based upon 13 its own judgment and upon advice from such advisers as it has 14 deemed necessary.

15 ...

16 Status of the Parties. The other party is not acting as fiduciary for or as 17 an adviser to it in respect of that Transaction.

19 Id. at 261.

20 Cairn kept the exact timing and amount of the transaction close to its chest.

21 It did not inform even HSBC for fear of influencing the market. But the fact of the transaction—that Cairn was in the process of selling a major subsidiary, the deal would be conducted in dollars, Cairn would conduct a major foreign exchange transaction, and then it would distribute most of the proceeds to its shareholders in pounds—was public before the RFP process was even held, and the NDA did 24-1221 Johnson v. United States not purport to apply to this publicly available knowledge.

2 Execution of the Trade 3 After the Mandate Letter was signed but before the transaction date, Johnson purchased pounds in his proprietary trading book on behalf of HSBC. On December 7, 2011, at 1:56 PM, Cairn officially placed its order with HSBC for 1.2 billion pounds at the 3:00 PM fix rate. At 2:28 PM, Cairn increased its order to approximately 2.25 billion pounds. Johnson responded to the news of the increased order with, “fucking Christmas!” Johnson was in New York on December 7, so he delegated the responsibility of overseeing the trade to Stuart Scott, an HSBC trader in London. Scott supervised Frank Cahill, another HSBC trader in London who was tasked with buying the pounds to fill Cairn’s order.

12 Cahill testified that he didn’t receive any direct instruction from Johnson about how to trade, and he went about it as he normally would: some “aggressive” methods that would ramp the price, followed by pauses to let the market breathe and minimize upward movement.

16 At 2:54 PM, Johnson learned that HSBC was still 1.2 billion pounds short of the 2.25 billion needed. He spoke with Scott by phone, and again the parties have

24-1221 Johnson v. United States different interpretations of this conversation. The relevant portions of the call are as follows: 3 JOHNSON: Just uh, the email I sent to Dipak [Khot]. So we should 4 use that as reference rate, right? So to, I mean obviously, we’ve got a 5 bit of a way to go, right. But I don’t know what his average is but as 6 long as it is under that rate, they can’t really complain. Right?

7 SCOTT: Yeah, yeah.

8 JOHNSON: If it’s over 5730 they’re gonna squeal.

9 SCOTT: If it’s over what?

10 JOHNSON: If it’s over fifty—well if it’s—it was around 5630 when 11 they called us, right, and it spent most of the morning around 5620- 12 30 so we should probably make sure we don’t ramp it up through 13 there.

15 App’x at 236.

16 Johnson then told Scott that they could “afford to go short some,” that is, either sell some pounds to prevent the price from being ramped above 5730 or stop short of purchasing 2.25 billion pounds and fulfill the order from HSBC’s proprietary account. 1 According to the government, Johnson and Scott were strategizing about how much they could ramp up the price before Cairn would complain. According to Johnson, he was warning Scott against ramping up the price too high and specifically advising him to keep the price below that of a full risk transfer, which had been the other FX strategy Cairn had seriously considered.

1 “5730” in this context refers to an exchange rate of 1.5730 British Pounds to U.S. dollars.

24-1221 Johnson v. United States 1 Cahill purchased 1.2 billion pounds in the final six minutes before 3:00 PM.

2 The price of the pound increased and reached its highest point of the day at the 3:00 PM fix. Following the transaction, Cairn expressed concern about the spike in price. Cairn asked to speak with Johnson and Scott. Before the call, Khot warned Johnson “to be a bit careful if it’s us.” Supp. App’x at 135. On the phone, Scott told Cairn that such spikes were pretty normal and said that the Russian Central Bank had also been buying pounds at that time. Johnson chimed in saying that the Russians were “always selling dollars.” Supp. App’x at 147.

9 Before Cairn placed its order on December 7, Johnson, from New York, had sent the following message to Paul Clark, an HSBC trader in London: “I left my watch in the hotel, have to go and get it . . . .” Supp. App’x at 219. The government argues that Johnson used “watch” as a code word, in fact, as a tip to buy pounds.

13 In any event, Clark and several other HSBC traders began buying pounds shortly after Johnson’s message. Many, but not all, of these pounds were used to fulfill Cairn’s order. The remaining pounds were sold shortly after 3:00 PM to third parties, when the price of pounds was at its peak for the day, earning much money for HSBC’s proprietary books.

24-1221 Johnson v. United States 1 HSBC ultimately decided to give Cairn a discount of 2.5 pips. HSBC’s total profits from the transaction were around $7 million. Johnson asserts that the transaction cost Cairn significantly less than the full-risk transfer would have cost and that HSBC’s “commission” on the deal amounted to 0.2% of the $3.5 billion transaction.

6 Procedural Background 7 In 2016, Johnson was indicted on one count of conspiracy to commit wire fraud in violation of 18 U.S.C. § 1349 and 10 counts of wire fraud in violation of 18 9 U.S.C. § 1343. Johnson’s four-week trial began in late 2017.

10 At trial, the government advanced two theories of wire fraud. Under the misappropriation theory, the government argued that Johnson breached a duty to Cairn by using Cairn’s confidential information to make proprietary trades prior to the fix in a way that ramped the price for Cairn and made profit for HSBC.

14 Under the right-to-control theory, the government argued that Johnson, after misleadingly representing to Cairn that HSBC would not ramp up the price of the pound, intentionally ramped the fix, depriving Cairn of pricing information that went to the core of the deal.

24-1221 Johnson v. United States 1 The jury found Johnson guilty on one count of conspiracy and eight counts of wire fraud in a general verdict that did not specify upon which theory of liability it relied. The district court sentenced Johnson to 24 months’ imprisonment, three years of supervised release, and a $300,000 fine. On appeal, this Court upheld Johnson’s conviction on the right-to-control theory, expressly deciding “not [to] reach Johnson’s arguments as to the misappropriation theory.” Johnson, 945 F.3d 7 at 608.

8 As Johnson was completing his sentence, the Supreme Court invalidated the right-to-control theory of fraud. Ciminelli v. United States, 598 U.S. 306 (2023).

10 Shortly thereafter, Johnson filed this Petition for coram nobis with his sentencing court. The district court rejected his Petition, concluding that “the jury would have found that Johnson met each of the elements required under the misappropriation theory.” Johnson, 2024 WL 1740916 at *12. As a result, it found that the inclusion of the invalid right-to-control theory in the jury charge was harmless.

15 DISCUSSION 16 Coram nobis, the common law writ used to correct errors “of the most fundamental character” after a final conviction and exhaustion of all appeals, is an “extraordinary remedy” to be used only where “circumstances compel[] such 24-1221 Johnson v. United States action to achieve justice.” United States v. Morgan, 346 U.S. 502, 511–12 (1954) (citation and quotation marks omitted). In this Circuit, we have said that the writ should issue when a petitioner shows “1) there are circumstances compelling such action to achieve justice, 2) sound reasons exist for failure to seek appropriate earlier relief, and 3) the petitioner continues to suffer legal consequences from his conviction that may be remedied by granting of the writ.” Kovacs v. United States, 744 F.3d 44, 49 (2d Cir. 2014).

8 Nobody disputes that Johnson filed his Petition at his earliest possible opportunity. 2 And, as the district court found, Johnson continues to suffer legal consequences from his conviction in the form of restrictions on his ability to work in banking and to travel to this country. Johnson, 2024 WL 1740916, at *6 n.14. The only question before us on appeal is, therefore, the main one—whether justice compels issuance of the writ.

14 I. A Possible Yates Error 15 When a jury is presented with an illegal or unconstitutional basis for conviction and a legally valid basis, a so-called Yates error arises if the jury returns

2Johnson filed this Petition 75 days after the Supreme Court handed down its decision in Ciminelli and 72 days after he was released from prison.

24-1221 Johnson v. United States a general verdict and a court cannot determine upon which basis the jury convicted. Griffin v. United States, 502 U.S. 46, 56 (1991). If Johnson were still imprisoned, he would be able to seek habeas relief on the grounds that his jury was invited to convict him on a legally invalid basis, the right-to-control theory, alongside the legally valid misappropriation theory. 3 Similarly, if his case were still on direct appeal, Johnson could have argued that a Yates error here required reversal of his conviction. See United States v. Garcia, 992 F.2d 409, 416 (2d Cir. 1993). As a result, the possibility that Johnson’s jury convicted him for behavior “that simply is not illegal” is adequate to support his claim that justice requires the issuance of coram nobis. United States v. Mandanici, 205 F.3d 519, 525 (2d Cir. 2000) (quotation marks omitted).

12 But Yates errors are not structural and are subject to harmlessness review.

13 Hedgpeth v. Pulido, 555 U.S. 57, 58 (2008). The district court denied Johnson’s Petition because it determined “that including the invalid right-to-control theory

3 It is unimportant that Johnson was convicted before the Supreme Court’s determination in Ciminelli that the right-to-control theory is “invalid,” 598 U.S. at 317, because that decision created a new rule of substantive criminal law. “A new rule of substantive criminal law is presumptively retroactive because a defendant may have been punished for conduct that simply is not illegal.”

United States v. Mandanici, 205 F.3d 519, 525 (2d Cir. 2000) (quotation marks omitted).

24-1221 Johnson v. United States [did not affect] the verdict’s outcome” and therefore was harmless. Johnson, 2024 2 WL 1740916, at *7. It is to that question we now turn.

3 II. Harmlessness Review in Coram Nobis 4 This Court has never squarely stated which standard of harmlessness review applies in coram nobis. In recognition of the extraordinary nature of coram nobis relief and the interest of finality, the government urges us to apply the so- called Kotteakos standard used for collateral review of state court convictions.

8 Under that standard, the government must show the error did not have “substantial and injurious effect or influence in determining the jury’s verdict.”

10 Kotteakos v. United States, 328 U.S. 750, 776 (1946). Johnson instead urges us to apply the Chapman standard used for initial review of constitutional errors. Under Chapman, the government must demonstrate that the error “was harmless beyond a reasonable doubt.” Chapman v. California, 386 U.S. 18, 24 (1967).

14 We note that the concerns about comity, federalism, and state interests cited by the Supreme Court for preferring Kotteakos in collateral review are not present in this case, where the Petitioner asks us to overturn a federal conviction originating in this Circuit. Brecht v. Abrahamson, 507 U.S. 619, 635 (1993). We note also that under both Chapman and Kotteakos the government bears the burden of 24-1221 Johnson v. United States persuasion. Id. at 630 (“The State bears the burden of proving that an error passes muster under [the Chapman reasonable doubt] standard.”); O'Neal v. McAninch, 513 U.S. 432, 438–39 (1995) (stating the same for the Kotteakos substantial-influence standard).

5 In the end, however, because Johnson prevails under either standard, we decline to decide in this case which harmlessness test is more appropriate for coram nobis. And we hold that the government has not met its burden under the less burdensome Kotteakos standard.

9 III. Weaknesses of the Misappropriation Case against Johnson 10 The issue before us is whether the inclusion of the misappropriation theory in the jury instructions renders harmless the inclusion of the invalid right-to- control theory. We are unconvinced by the government’s argument that Johnson’s jury was not substantially influenced by the inclusion of the invalid right-to- control theory. Indeed, as to least two elements of the misappropriation theory, the government’s case was so weak that we find ourselves doubting that a jury would have convicted Johnson on that basis.

17 To convict on the misappropriation theory, the jury needed to find four elements beyond a reasonable doubt: “(1) the Defendant entered into a 24-1221 Johnson v. United States relationship of trust and confidence with Cairn; (2) Cairn provided the Defendant with confidential information in the course of such a relationship; (3) the defendant . . . secretly used that information for his own benefit, under circumstances where that use could or did result in a tangible harm to Cairn; and (4) the defendant acted with knowledge and fraudulent intent.” Johnson, 2024 WL 6 1740916 at *7 (citation and quotation marks omitted). On the first element, which we call the fiduciary relationship element, and the third, which we call the misappropriation element, the government’s evidence was decidedly weak.

9 A. The Fiduciary Relationship Element 10 A person can only commit fraud by misappropriation if they misappropriate confidential information from a person or entity with which they have a pre-existing fiduciary or quasi-fiduciary relationship. See United States v. Chestman, 947 F.2d 551, 570–71 (2d Cir. 1991) (en banc). A person acts as another’s fiduciary when “the business which he transacts, or the money or property which he handles, is not his own or for his own benefit, but for the benefit of” the other.

16 Id. at 568 (quoting Black's Law Dictionary 564 (5th ed. 1979)). This fiduciary relationship cannot be “lightly implied,” and a customer’s placing “great confidence and trust” in their broker is insufficient. United States v. Skelly, 442 F.3d 22 24-1221 Johnson v. United States 94, 98–99 (2d Cir. 2006). Rather, a fiduciary relationship requires “de facto control and dominance” by the agent over the principal’s affairs. Id. (citation and quotation omitted).

4 In his defense, Johnson pointed to the Mandate Letter and its incorporation of the ISDA disclaiming any fiduciary relationship between HSBC and Cairn. As a result he asserts that, as a matter of law, he had no fiduciary duty to Cairn.

7 During Johnson’s trial, the district court acknowledged that whether the fiduciary duty element was a question of law or an issue of fact was a “knotty issue.” The court ultimately decided that this element presented an issue of fact for the jury.

10 Where the dispute over the existence of a fiduciary relationship is based in factual questions, like who promised what to whom, it is of course the provenance of the jury to decide those facts. Here, the dispute was not so much what was promised as which promises were controlling.

14 We do not, however, conclude that the district court was incorrect to allow the jury to decide the fiduciary duty element. It is legally possible for a person to form a de facto fiduciary relationship notwithstanding a purported contractual disclaimer of such a relationship. But the evidence for such a fiduciary relationship would need to be quite strong and the jury charge quite clear. And the explicit 24-1221 Johnson v. United States disclaimer of fiduciary liability and the absence of any contractual breach carry quite a lot of weight.

3 A factfinder in such a situation must begin from the presumption that no fiduciary relationship exists, and only the strongest parol evidence that a defendant deliberately created a quasi-fiduciary relationship can suffice to override that presumption. Significantly, in the different but similar context of honest-services fraud, the Supreme Court recently found that this Court violated a defendant’s Due Process rights when we affirmed a jury instruction that allowed the jury to impute a fiduciary duty to the public from a finding that a defendant “dominated and controlled” government business and had a “special relationship” to the government even when that defendant had no official public position. Percoco v. United States, 598 U.S. 319, 322 (2023). The Supreme Court explicitly did not say that a formal public position is the only way to assume a fiduciary duty to the public, but the Supreme Court made clear that in the absence of such a position, a fiduciary relationship should not be lightly implied from vague principles. The same also must be true for a defendant with no explicit,

24-1221 Johnson v. United States contractually binding fiduciary relationship with his purported victim. 4 2 All in all, we find it unlikely that a reasonable jury would have reached unanimous agreement on the fiduciary duty element here.

4 B. The Misappropriation of Confidential Information Element 5 To create fraud liability, it is not enough for a defendant, without full disclosure, to appropriate another’s confidential information. A defendant must misappropriate that information—that is, use it in a way that was not permitted— for their own benefit and to the detriment of the other. The case against Johnson on this element is far from straightforward.

10 The district court was correct that it is not theoretically impossible to prove misappropriation fraud, which is typically used to prosecute securities fraud, also

4The government appears at times to argue that the NDA counteracted the contractual disclaimer in some way and created an independent fiduciary duty. But “[a] mutual nondisclosure agreement does not per se create a fiduciary relationship.” Converged Compliance Sols., Inc. v. XOP Networks, Inc., No. 21-CV-5482 (PGG) (OTW), 2024 WL 4665114, at *18 (S.D.N.Y. Sept. 25, 2024) (citing Gate Technologies, LLC v. Delphix Capital Mkts., LLC, 12-CV-7075 (JPO), 2013 WL 3455484, at *8 (S.D.N.Y. July 9, 2013)). And both this Court and the District Court implicitly found that Johnson had committed no contractual breach. Johnson, 945 F.3d at 613 (“[The criminal fraud statute here] applies even if the parties’ contract was never breached.”); Johnson, 2024 WL 1740916, at *9 (“[A] defendant may commit wire fraud even absent a breach of contract when the defendant had an intent to defraud.”).

As a result, we analyze the NDA as yet another piece of parol evidence pointed to by the government that Johnson intended to create a quasi-fiduciary relationship with Cairn notwithstanding the contract between the parties.

24-1221 Johnson v. United States in the foreign transactions market. But when this form of liability is applied in this new, foreign exchange context, the test must take into account the differences in the markets. The government does not dispute that, unlike traders in stock, foreign exchange dealers cannot simply abstain from the market when they receive “inside” information. A foreign exchange dealer must buy and sell many millions of pounds and dollars every day, even if it is a party to an upcoming fix transaction. Indeed, a foreign exchange dealer engaged in a large fix transaction must trade ahead of the fix to execute the deal and can further trade ahead, to some disputable degree, to hedge against the risk of an unfavorable fix price. Insofar as either the fact of the fix transaction or its timing is confidential, the foreign exchange dealer uses confidential information in both cases.

12 The government never convincingly made out its theory that Johnson’s use of Cairn’s information was misappropriative. Its evidence did not show Johnson’s behavior to be extra-ordinary, and thus mis-appropriative, in any of the above respects. Indeed, the government’s evidence showed that Cahill, the trader in charge of executing the fix, conducted himself as he “normally” would. App’x at 133. This was important evidence for the government’s right-to-control case, in which it accused Johnson of misleading Cairn about the extent to which HSBC 24-1221 Johnson v. United States would treat the deal as a normal fix transaction. But it undermines the government’s argument that Johnson used confidential information about the fix in an abnormal and knowingly impermissible way.

4 As a result, we doubt that a properly instructed jury would have found the government satisfied its burden of showing that Johnson misappropriated Cairn’s information.

7 IV. The Case for Grave Doubt 8 With these concerns in mind, we must determine whether the presentation of the erroneous right-to-control theory was harmless. The question before us is not whether a jury could have or even would have convicted Johnson if presented only with a misappropriation theory of fraud. It is, rather, whether Johnson’s jury did convict him on that basis.

13 The right-to-control case against Johnson was straightforward and that was the only basis upon which his conviction was previously upheld by this Court.

15 The government’s misappropriation theory, instead, was substantially more complicated and had serious shortcomings. Given those circumstances, we find it impossible to avoid grave doubt that the jury was “substantially swayed by” the presentation of the invalid right-to-control theory alongside the misappropriation 24-1221 Johnson v. United States theory. 328 U.S. at 765.

2 Much in this case was confusing or disputed, but the evidence was clear that Johnson, immediately after the transaction, misled Cairn about why the 3:00 PM fix price rose so much. From this, the jury was invited to infer that Johnson intentionally ramped the 3:00 PM fix price, and we think that inference is strong.

6 It is also indisputable that Johnson led Cairn to believe that HSBC, unlike other foreign exchange dealers the company was considering in the RFP process, would not intentionally ramp the fix price. These facts created a textbook case of right-to- control fraud, and the jury was invited to convict Johnson on that basis.

10 On the other hand, to convict on the misappropriation theory the government needed to prove (1) that Johnson owed a fiduciary duty to Cairn, despite the basic principal-to-principal structure of the deal and the contractual disclaimers of such a duty, because Johnson had made statements inducing Cairn to believe Johnson had assumed that duty, (2) that because of this fiduciary duty, Cairn shared confidential information with Johnson, (3) that Johnson misused this confidential information, intentionally buying more pounds than he could reasonably need to fulfill the deal or hedge against risk for HSBC and then intentionally ramped the price of pounds to Cairn’s detriment, and (4) that 24-1221 Johnson v. United States Johnson intended to defraud Cairn when doing all of the above.

2 Despite our doubts expressed earlier, it is not impossible that a reasonable jury could find that the government proved all four of these elements beyond a reasonable doubt. But that is not the test. And we find it highly unlikely that a reasonable jury would have reached unanimous agreement on the more complicated and contestable misappropriation theory when it had the right-to- control theory as an available alternative. Indeed, we find ourselves—at the very least—in “virtual equipoise” as to whether any jury, presented only with the misappropriation theory, would convict Johnson. O'Neal, 513 U.S. at 435. That is more than enough to leave us with grave doubt.

11 CONCLUSION 12 We therefore REVERSE the judgment of the district court and REMAND for entry of an order granting the Petition.