Frederick Grede v. FCStone LLC

In the United States Court of Appeals For the Seventh Circuit ____________________ No. 16-1896 & 16-1916 FREDERICK J. GREDE, not individually but as Liquidation Trus- tee of the Sentinel Liquidation Trust, Plaintiff-Appellant / Cross-Appellee, v. FCSTONE, LLC, Defendant-Appellee / Cross-Appellant. ____________________ Appeals from the United States District Court for the Northern District of Illinois, Eastern Division. No. 09 C 136 — James B. Zagel, Judge. ____________________ ARGUED JUNE 7, 2017 — DECIDED AUGUST 14, 2017 ____________________ Before RIPPLE, ROVNER, and HAMILTON, Circuit Judges. HAMILTON, Circuit Judge. The 2007 bankruptcy of Sentinel Management Group, Inc. has echoed through the courts for ten years now. This is our fifth appeal dealing with Sentinel. In a pair of cases decided in 2013 and 2016, we addressed the priority of a claim against the bankruptcy estate by the Bank of New York, Sentinel’s largest (but no longer secured) credi- tor. In re Sentinel Management Group, Inc., 728 F.3d 660 (7th Cir. 2 Nos. 16-1896 & 16-1916 2013); Grede v. Bank of New York Mellon Corp. (In re Sentinel Management Group, Inc.), 809 F.3d 958 (7th Cir. 2016). Earlier this year, we affirmed the convictions and sentence of Senti- nel’s former president and CEO, who was prosecuted for wire fraud and investment adviser fraud. United States v. Bloom, 846 F.3d 243 (7th Cir. 2017). In Grede v. FCStone, LLC, 746 F.3d 244 (7th Cir. 2014) (FCStone I), the direct predecessor to this appeal, we consid- ered among other issues a distribution of $297 million to a group of Sentinel customers a few days after Sentinel filed for bankruptcy protection in August 2007. Following a bench trial, the district court had allowed the trustee in bankruptcy to avoid this post-petition transfer under 11 U.S.C. § 549. We reversed, holding that relief under § 549 was unavailable to the trustee because the bankruptcy court had authorized the transfer. We rejected the trustee’s reliance on an October 2008 “clarification” through which the bankruptcy judge indicated that he had not intended to foreclose a § 549 avoidance action. Later statements by the judge about his subjective intentions could not, we concluded, defeat the plain language of the or- der authorizing the transfer. We remanded for further pro- ceedings, which led to these new appeals. Despite our holding in FCStone I that “the bankruptcy court authorized the post-petition transfer” and that the trus- tee “therefore cannot avoid the transfer,” 746 F.3d at 258, the trustee argued on remand that the bankruptcy judge’s Octo- ber 2008 “clarification” was entitled to preclusive effect. Since FCStone did not appeal that “clarification” when it was made, the trustee argued, FCStone should be bound by it and collat- erally estopped from arguing that the post-petition transfer Nos. 16-1896 & 16-1916 3 was authorized. The district court rejected the trustee’s argu- ment on this point, and we affirm on two independent grounds. First, pursuant to the mandate rule and the law-of- the-case doctrine, the collateral estoppel theory was unavail- able to the trustee on remand. Second, even if the theory were available despite our unambiguous holding in FCStone I, the bankruptcy judge’s “clarification” was not the sort of final rul- ing that could be entitled to preclusive effect. On cross-appeal, FCStone raises an issue that lingered in the background but was not squarely presented during the previous appeal. The question concerns the proper distribu- tion of nearly $25 million held in reserve under the confirmed bankruptcy plan. FCStone argues that these funds are trust property belonging to it and other creditors in its customer class who are protected by statutory trusts under the Com- modity Exchange Act. The district court disagreed, treating the funds instead as property of the bankruptcy estate subject to pro rata distribution among all Sentinel customers and other unsecured creditors. On this cross-appeal, we reverse. Under the Bankruptcy Code, property held by the debtor in trust for others is by definition not property of the bankruptcy estate. Pursuant to the confirmed bankruptcy plan, FCStone and similarly situated customers preserved their right to re- cover their trust property. These creditors are entitled to the benefit of reasonable tracing conventions. Moreover, FCStone introduced essentially unrebutted evidence at trial showing that it can trace a portion of the reserve funds back to its in- vestment. FCStone is entitled to recover its proportionate share of the reserve funds. The reserve funds should be dis- tributed pro rata among FCStone and other members of its customer class. 4 Nos. 16-1896 & 16-1916 I. Factual Overview and Procedural History A. Sentinel’s Collapse We review the most salient facts of the case, drawing from the district court’s findings after the earlier bench trial. More complete discussions of Sentinel’s downfall and the criminal misconduct of senior executives are included in the district court’s earlier opinion, Grede v. FCStone, LLC, 485 B.R. 854, 859–67 (N.D. Ill. 2013), and in our opinions in Bloom, 846 F.3d at 246–50, and FCStone I, 746 F.3d at 247–51. In brief, Sentinel managed investments for futures com- mission merchants (FCMs) like FCStone, as well as for other classes of investors. FCMs act as financial intermediaries be- tween investors and futures markets. They are regulated un- der the Commodity Exchange Act. Sentinel itself was an FCM and so was regulated under the Act. Sentinel organized its customers in different tranches known as segments or “SEGs.” Within each SEG, customers were further divided into investment groups based on their risk appetites and financial goals. As relevant to this appeal, FCM customer assets were held in SEG 1, with FCStone’s cus- tomer assets placed in Group 7. SEG 3 contained assets be- longing to hedge funds and other sophisticated investors, as well as FCM proprietary or “house” funds. When customers invested funds with Sentinel, those funds were exchanged for securities and interest-bearing cash through a process that Sentinel called “allocation.” Customers did not own securities outright but instead held indirect pro rata interests in the securities allocated to their group portfo- lios, as determined by their level of investment. Nos. 16-1896 & 16-1916 5 Both the SEG 1 and SEG 3 customers were entitled to spe- cial protections under federal law. FCM customers who in- vested their own clients’ funds in SEG 1 were protected by the Commodity Exchange Act and regulations promulgated by the Commodity Futures Trading Commission (CFTC). SEG 1 and SEG 3 customers alike were protected by the Investment Advisers Act of 1940 and regulations promulgated by the Se- curities and Exchange Commission (SEC). Both regulatory re- gimes required Sentinel to hold customer funds in segrega- tion, i.e., separate from funds belonging to other customer classes and separate from Sentinel’s own or “house” funds. Both regimes also created statutory trusts in the customers’ favor to protect their property from Sentinel and its other creditors. Sentinel, unfortunately, did not honor the statutory trusts and comply with the segregation rules under the Commodity Exchange Act and the Investment Advisers Act. Instead, as the district court found, Sentinel routinely used hundreds of millions of dollars in securities it had allocated to customers as collateral to support Sentinel’s own borrowing to pursue its leveraged trading strategy for its own benefit. It moved those securities out of segregation and into a lienable account at the Bank of New York, its main lender, putting customer property at risk for Sentinel’s benefit. As Sentinel’s leveraged trading increased, its outstanding debt ballooned, and it drew more and more on its customers’ assets to support its borrowing habit. During the summer of 2007, Sentinel’s investment scheme collapsed. As credit markets tightened and liquidity dried up on Wall Street (this was the beginning of what would become the financial crisis of the late 2000s), the market value of many 6 Nos. 16-1896 & 16-1916 Sentinel assets dropped. Sentinel’s trading partners began making demands that forced it to borrow more heavily and in turn to provide more collateral—which it did by using cus- tomers’ property as collateral. In late June and July 2007, Sen- tinel moved $250 million worth of securities allocated to SEG 1 to the lienable Bank of New York account. Then, in late July, Sentinel swapped these securities with securities allocated to SEG 3 customers, resulting in a “massive shift of loss expo- sure” from SEG 1 to SEG 3. See Grede, 485 B.R. at 866. That final manipulation proved fateful for SEG 3 custom- ers in the looming bankruptcy. Sentinel’s wheeling and deal- ing had bought it some time, but in the end the firm could not keep up with redemption requests and demands from the Bank of New York. On Monday, August 13, 2007, Sentinel ad- vised its customers that it was halting all redemptions (i.e., payments to them from their accounts). On Thursday, August 16, Sentinel sold a large portfolio of securities then allocated to SEG 1 to a firm called Citadel, depositing the proceeds in a SEG 1 cash account at the Bank of New York. The next day, Sentinel filed for Chapter 11 bankruptcy protection. B. Chapter 11 Proceedings 1. Early Litigation On Monday, August 20, 2007, the first business day after it filed for bankruptcy, Sentinel (still under the control of its in- siders) filed an emergency motion in the bankruptcy court seeking an order approving payment of the Citadel sale pro- ceeds to SEG 1 customers. After emergency hearings, the bankruptcy court issued an order authorizing the Bank of New York to disburse the funds, less an approximately five percent holdback. The bank did so, and the SEG 1 customers Nos. 16-1896 & 16-1916 7 received $297 million in what the parties describe as the “post-petition transfer,” with FCStone receiving a little shy of $15 million. Frederick Grede was appointed as Chapter 11 trustee. The bankruptcy court approved his appointment on August 29, 2007, within the fourteen-day window for appealing the order authorizing the post-petition transfer. See Fed. R. Bankr. P. 8002. The trustee did not appeal. A year later, however, he filed a “Motion to Clarify or in the Alternative to Vacate or Modify the Court’s August 20, 2007 Order.” In essence, the trustee argued that he should be permitted to bring avoidance actions against FCStone and the other SEG 1 customers who received, in the trustee’s view, a disproportionate payout through the post-petition transfer. A group of SEG 1 customers including FCStone opposed the trustee’s motion. At the conclusion of a hearing on the mo- tion, the bankruptcy judge declined to vacate or modify the prior order. The judge said, however, that he was clarifying the order in that he “did not decide on August 20 and … am not deciding today whether or not any of the proceeds that were the subject of that order are property of the estate … or whether … they were trust funds.” The judge said that in his August 20 order, he “did not intend to foreclose the trustee or any party from any avoidance action whatsoever.” 2. FCStone I The bankruptcy court entered an order confirming the Fourth Amended Chapter 11 Plan of Liquidation in late 2008. (We discuss several provisions of the confirmed Chapter 11 plan in Part II-B.) Around that same time, the trustee com- menced adversary actions against FCStone and other SEG 1 8 Nos. 16-1896 & 16-1916 customers who had received distributions from the Citadel security sale back in August 2007. The trustee sought to avoid the post-petition transfers and to recover the Citadel sale pro- ceeds (Count I), and he requested a declaration that funds held in reserve are property of the bankruptcy estate (Count III). 1 The district court withdrew the reference to the bank- ruptcy court, see 28 U.S.C. § 157(d), and presided over the case against FCStone as a test case. Following a bench trial, the district court ruled in favor of the trustee on Counts I and III. Grede, 485 B.R. at 889–90. The court concluded that the Citadel sale proceeds should be treated as property of the bankruptcy estate. Id. at 880. The court reasoned that (1) both SEG 1 and SEG 3 customers were protected by statutory trusts; (2) because the two classes of customers were similarly situated and because there were in- sufficient funds to satisfy all their claims, tracing fictions or conventions were inappropriate; and (3) FCStone and other SEG 1 customers could not trace the Citadel sale proceeds back to their original investments given Sentinel’s coming- ling and misappropriation of customer assets that should have been segregated in trust for the customers. Id. at 873, 878, 880. The district court added that, for purposes of 11 U.S.C. § 549, and in light of the bankruptcy court’s 2008 “clarifica- tion,” the 2007 post-petition transfer had not been “author- ized” by the bankruptcy court. Id. at 881. 1 The trustee also sought to avoid a pre-petition transfer (Count II) and to recover on a theory of unjust enrichment (Count IV). Those claims are no longer at issue. The trustee also brought a disallowance claim under 11 U.S.C. § 502(d), but that claim turns on the outcome of Count I and re- quires no separate consideration in this appeal. Nos. 16-1896 & 16-1916 9 We reversed in FCStone I, 746 F.3d at 260. We explained that the bankruptcy court’s after-the-fact “clarification” of its subjective intentions concerning the post-petition authoriza- tion order ran contrary to the plain language of the order and amounted to an abuse of discretion. Id. at 255. “Whether the property belonged to the estate or not,” we reasoned, “in the absence of reversal, the authorization order ended any discus- sion about its original ownership, and the disputed property cannot later be clawed back by the trustee.” Id. Because the parties had focused on the post-petition trans- fer, we did not specifically address the status of the funds held in reserve. 2 For that matter, we declined to decide “whether the funds at issue were, in fact, property of the estate,” id. at 258, though we agreed with the district court that there was 2 In the prior appeal, FCStone’s opening brief noted in passing the five percent holdback from the post-petition transfer, and it cited reservation- of-rights language appearing in the transfer authorization order. FCStone also devoted a few sentences to the reserves created under the confirmed Chapter 11 plan, but it did not discuss the composition of those reserves or explain why a favorable ruling on the post-petition transfer issue would not necessarily resolve the parties’ disagreement over the allocation of the reserve funds. The trustee barely acknowledged the five percent holdback, and he made no mention of the reserves. While the parties chide us more or less gently for assuming that our resolution of the post-petition transfer issue also resolved the trustee’s declaratory judgment request in Count III, we believe the onus was on the parties to identify clearly each source of disputed funds and the arguments relevant to the disposition of those funds. E.g., Montgomery v. Amoco Oil Co., 804 F.2d 1000, 1004 n.8 (7th Cir. 1986) (issue raised in a “two-sentence reference in the statement of facts” and through a citation to a statute was “hardly properly presented”); Pru- dential Ins. Co. of America v. Sipula, 776 F.2d 157, 161 n.1 (7th Cir. 1985) (lit- igant who attempted to incorporate by reference into his appellate brief arguments he had raised in the district court assumed the risk that appel- late court might overlook issues not clearly presented). 10 Nos. 16-1896 & 16-1916 no generally applicable legal basis for placing one statutory trust ahead of another. We tentatively approved the district court’s requirement that would-be trust claimants (such as FCStone) must actually trace their investment property to the disputed funds. Id. at 259. We remanded for further proceed- ings. 3. The Decisions on Remand On remand, the trustee moved for judgment on Counts I and III, while FCStone sought judgment on Count I and sum- mary judgment on Count III. With respect to Count I, the trus- tee argued that FCStone should be collaterally estopped from asserting that the post-petition transfer was authorized. In the trustee’s view, the bankruptcy judge’s October 2008 “clarifica- tion” was entitled to preclusive effect. The district judge disa- greed, writing that our decision in FCStone I—holding that the “clarification” was an abuse of discretion—stripped that rul- ing of “any force and effect.” Grede v. FC Stone, LLC, 556 B.R. 357, 362 (N.D. Ill. 2016). The court entered judgment for FCStone on Count I. The trustee appeals that decision. The district judge then considered the status of the dis- puted reserves. The judge reiterated his view that equity pre- vented him from “favoring one statutory trust claim over an- other” and that actual tracing is difficult if not impossible given Sentinel’s egregious pattern of commingling. Id. at 365. The judge concluded that the reserve funds should be treated as property of the estate, subject to pro rata distribution ac- cording to the confirmed Chapter 11 plan. Id. at 366. The court entered judgment for the trustee on Count III. FCStone cross- appeals that decision. Nos. 16-1896 & 16-1916 11 II. Analysis A. The Trustee’s Appeal — Collateral Estoppel On the issue of collateral estoppel (also known as issue preclusion) presented by the trustee’s appeal, no facts are dis- puted, so we review de novo the district court’s decision on this question of law. Adams v. Adams, 738 F.3d 861, 864 (7th Cir. 2013); Bernstein v. Bankert, 733 F.3d 190, 225 (7th Cir. 2013). 1. Mandate Rule and Law of the Case The trustee’s collateral estoppel argument is straightfor- ward, if improbable. In the trustee’s view, FCStone’s failure to appeal the bankruptcy court’s oral “clarification” of its prior written order means that FCStone should be bound by that “clarification” rather than being able to rely on the underlying order. As a preliminary matter, we conclude that the collateral estoppel argument was not even available to the trustee on remand following our decision in FCStone I. We did not spe- cifically address collateral estoppel in our prior opinion be- cause the trustee did not raise the issue, even though he had presented it earlier to the district court as an alternative argu- ment in support of his § 549 avoidance action. But our broader discussion of the post-petition transfer left nothing to the im- agination on this point. We said that the transfer was author- ized and that it therefore “cannot be avoided under the ex- press terms of 11 U.S.C. § 549.” FCStone I, 746 F.3d at 247. We repeated that the transfer was “clearly authorized” and that, regardless whether the transferred property was part of the bankruptcy estate, “in the absence of reversal, the authoriza- tion order ended any discussion about its original ownership, 12 Nos. 16-1896 & 16-1916 and the disputed property cannot later be clawed back by the trustee.” Id. at 255. Given our unambiguous resolution of the dispute over the post-petition transfer, two closely related doctrines—the mandate rule and the law-of-the-case doctrine—should have precluded the trustee from resurrecting his collateral estoppel theory in the district court and getting a second bite at the § 549 apple. Compare EEOC v. Sears, Roebuck & Co., 417 F.3d 789, 796 (7th Cir. 2005) (“In general, any issue conclusively de- cided by this Court on appeal may not be reconsidered by the district court on remand.”), and United States v. Polland, 56 F.3d 776, 777 (7th Cir. 1995) (“The mandate rule requires a lower court to adhere to the commands of a higher court on remand.”), with United States v. Adams, 746 F.3d 734, 744 (7th Cir. 2014) (“The law of the case doctrine is a corollary to the mandate rule and prohibits a lower court from reconsidering on remand an issue expressly or impliedly decided by a higher court absent certain circumstances.”) (emphasis added) (cita- tion omitted). The trustee argues that, as appellee in FCStone I, he was “not required to advance every possible ground for affir- mance.” See Door Systems, Inc. v. Pro-Line Door Systems, Inc., 83 F.3d 169, 174 (7th Cir. 1996); see also Frank v. Walker, 819 F.3d 384, 387 (7th Cir. 2016). That is true as a general principle, but it is difficult to understand why—when the legal weight of the bankruptcy judge’s “clarification” was squarely at is- sue—the trustee did not cover his bases by arguing in the al- ternative that the ruling (even if erroneous) was entitled to preclusive effect. In any event, the general privilege of the ap- pellee to renew on remand arguments preserved in the dis- trict court gives way to the mandate rule and the law of the Nos. 16-1896 & 16-1916 13 case where the argument the appellee would raise is flatly in- compatible with a prior mandate of this court. 3 While we need not go so far as to say, as the district court did, that FCStone I rendered the bankruptcy judge’s “clarifica- tion” a “legal nullity,” Grede, 556 B.R. at 362, we conclude that the district court acted appropriately in declining to reach the merits of an argument that, if decided in the trustee’s favor, would have eviscerated our prior holding. 4 3 At oral argument, we asked the trustee whether any appellate court has ever applied collateral estoppel to revive a trial court ruling it had previously disapproved on the merits. The trustee cited Federated Depart- ment Stores, Inc. v. Moitie, 452 U.S. 394 (1981), but that case is readily dis- tinguishable. In Moitie, two plaintiffs in a consolidated case declined to appeal the district court’s dismissal of their claims, while other plaintiffs appealed and obtained relief in the Ninth Circuit after the Supreme Court decided a case raising similar issues. The non-appealing plaintiffs re-filed in state court; the defendants removed the actions, and the federal district court dismissed the claims on res judicata grounds. In a subsequent ruling, the Ninth Circuit held that the non-appealing plaintiffs could “benefit from a reversal” since their litigating position was “closely interwoven” with that of the appealing parties. Id. at 398 (citation omitted). The Su- preme Court reversed that decision, declining to “countenance[] an excep- tion to the finality of a party’s failure to appeal merely because his rights are ‘closely interwoven’ with those of another party.” Id. at 400. Moitie is not responsive to the question we asked at oral argument. It does not in- volve a situation in which an appellate court applied preclusive effect to a trial court ruling that the same appellate court previously rejected on its mer- its. Moitie also fails to support the trustee’s position because, as discussed below, FCStone could not have appealed the bankruptcy judge’s “clarifi- cation” even if it had wanted to. 4 As a matter of appellate advocacy, it would ordinarily be prudent for an appellee who deliberately chooses not to argue alternative grounds for affirmance to alert the appellate court to the existence of those alterna- tive grounds. The first time the trustee alerted us to his collateral estoppel 14 Nos. 16-1896 & 16-1916 2. Collateral Estoppel and the Requirement of Finality Even if the trustee could have pursued his collateral estop- pel theory on remand without running afoul of our mandate in FCStone I, the theory would fail on the merits. A party is constrained by collateral estoppel as a matter of federal law only where four criteria are satisfied: “(1) the issue sought to be precluded must be the same as that involved in the prior litigation, (2) the issue must have been actually litigated, (3) the determination of the issue must have been essential to the final judgment, and (4) the party against whom estoppel is in- voked must [have been] fully represented in the prior action.” Matrix IV, Inc. v. American Nat’l Bank & Trust Co. of Chicago, 649 F.3d 539, 547 (7th Cir. 2011) (citation omitted). The key criterion in this case is finality. Collateral estoppel does not attach to tentative orders. See Loera v. United States, 714 F.3d 1025, 1028 (7th Cir. 2013) (“The doctrine of collateral estoppel … teaches that a judge’s ruling on an issue of law or fact in one proceeding binds in a subsequent proceeding the party against whom the judge had ruled, provided that the ruling could have been … challenged on appeal, or if not that at least it was solid, reliable, and final … .”) (emphasis added); Amcast Industrial Corp. v. Detrex Corp., 45 F.3d 155, 158 (7th Cir. 1995) (“[W]hether a judgment, not ‘final’ in the sense of 28 U.S.C. § 1291, ought nevertheless be considered ‘final’ in the sense of precluding further litigation of the same issue, turns upon such factors as the nature of the decision (i.e., that it was not avowedly tentative), the adequacy of the hearing, and the theory was in his petition for rehearing following our decision in FCStone I. Even at that point, the trustee’s theory was quite underdeveloped. He did not flesh out this theory until the remand proceedings. Nos. 16-1896 & 16-1916 15 opportunity for review.”), quoting Lummus Co. v. Common- wealth Oil Refining Co., 297 F.2d 80, 89 (2d Cir. 1961). The trustee’s collateral estoppel argument fails because the bankruptcy judge’s oral “clarification” was not the kind of “solid, reliable, and final” order entitled to preclusive effect. See Loera, 714 F.3d at 1028. The judge was ambivalent about whether the post-petition transfer involved funds that should be characterized as property of the estate. He said that he “did not decide” the issue in 2007 and was “not deciding today” in October 2008. By contrast, as we held in FCStone I, the August 2007 order itself unambiguously authorized the post-petition transfer and “ended any discussion about … original owner- ship.” 746 F.3d at 255–56. If any order was entitled to preclu- sive effect, it was the August 2007 order authorizing the post- petition transfer, not the bankruptcy judge’s later comments about his subjective intentions. The trustee points out that he brought his motion in the bankruptcy court under Federal Rule of Civil Procedure 60(b) and that the denial of Rule 60(b) relief is “appealable as a sep- arate final order.” See Stone v. INS, 514 U.S. 386, 401 (1995). In the trustee’s view, FCStone should have appealed from the bankruptcy court’s ruling on the Rule 60(b) motion, and its failure to do so should stop it from challenging the substance of that ruling. But Rule 60(b) does not authorize a judge to “clarify” the meaning of a prior order. The rule does authorize a judge to “relieve a party … from a final judgment, order, or proceeding” for good cause, and indeed the trustee asked the bankruptcy judge in the alternative to vacate or modify the August 2007 order. But the judge denied the motion. “For the rec- ord,” he said, “I’m denying the Rule 60(b) motion. I am not going to vacate or modify my order. It stands. I don’t think I 16 Nos. 16-1896 & 16-1916 made any mistake … .” In other words, though the trustee did indeed bring a Rule 60(b) motion, he lost his motion. The court took no firm action adverse to FCStone. It would have been extraordinary for FCStone to have tried to appeal a decision in its favor. We acknowledge that the clarification aspect of the bank- ruptcy judge’s oral ruling arguably was contrary to the posi- tion argued by FCStone, but the clarification had no actual ef- fect at that time. It was at best a helpful signal for the avoid- ance claims the trustee then asserted against FCStone and other SEG 1 customers seeking actual return of the money. It was the district court’s later decision on the avoidance claim against FCStone that would have moved money from one party to another (and that we reversed on a timely appeal in FCStone I). The trustee cites no authority for the proposition that a “clarification” that, as we previously held, runs “contrary to the plain language” of the underlying order,” FCStone I, 746 F.3d at 255, is itself an appealable final order. It is not even clear on what procedural basis the trustee brought his motion to clarify. See Barton v. Uniserv Corp., No. 15 CV 4149, 2016 WL 4577033, at *3 (N.D. Ill. Aug. 30, 2016) (Federal Rules of Civil Procedure do not expressly authorize “motion for clarifica- tion”); Lou v. Ma Laboratories, Inc., No. 12-cv-05409 WHA (NC), 2013 WL 1615785, at *1 (N.D. Cal. Apr. 15, 2013) (same); United States v. Philip Morris USA, Inc., 793 F. Supp. 2d 164, 168 (D.D.C. 2011) (same). While the trustee cites a handful of cases illustrating the breadth of potentially appealable orders, none involved a rul- ing as vague, open-ended, and inconclusive as the “clarifica- tion” here. E.g., Matrix IV, 649 F.3d at 549 (bankruptcy orders Nos. 16-1896 & 16-1916 17 confirming asset sale, denying creditor’s Rule 60(b) challenge to asset sale, and resolving lien priority after full trial, were final and appealable); Tidwell v. Smith (In re Smith), 582 F.3d 767, 776 (7th Cir. 2009) (bankruptcy order modifying dis- charge injunction was final and appealable); Chase Manhattan Mortgage Corp. v. Moore, 446 F.3d 725, 728 (7th Cir. 2006) (dis- trict court order entering summary judgment in bank’s favor but failing to order foreclosure was nevertheless final and ap- pealable by mortgagor); In re UAL Corp., 411 F.3d 818, 822 (7th Cir. 2005) (bankruptcy order vacating authorization allowing airline to retain leases was final and appealable). As a fallback, the trustee argues that “an order does not need to be final to be preclusive, provided it is sufficiently firm and not tentative and the parties were fully heard.” That’s not quite right. Finality is critical to the application of preclusion doctrines, though as we recognized in Gilldorn Sav- ings Ass’n v. Commerce Savings Ass’n, 804 F.2d 390, 393 (7th Cir. 1986), “finality for collateral estoppel is not the same as that required to appeal under 28 U.S.C. § 1291.” In Gilldorn Sav- ings, however, we added that appealability is still an im- portant factor in considering “whether a decision is ‘final’ for collateral estoppel purposes.” Id. at 393, citing Miller Brewing Co. v. Joseph Schlitz Brewing Co., 605 F.2d 990, 996 (7th Cir. 1979). Even if FCStone’s inability to appeal the oral “clarifica- tion” should not be treated as dispositive, nothing about that ruling was firm or definite. The ruling did not even make sense. As noted, the bankruptcy judge denied the trustee’s Rule 60(b) motion, declared that the August 2007 order “stands,” and then asserted that he was unprepared to decide the prop- erty-of-the-estate question. 18 Nos. 16-1896 & 16-1916 We are sympathetic to the bankruptcy judge’s situation back in August 2007. He faced extraordinary pressure to make a $300 million decision within a few hours. He was be- ing told by numerous parties that failure to authorize the transfer would force many of the SEG 1 FCMs into bank- ruptcy themselves, with ripple effects throughout commodity markets and other financial markets. The judge said on the record that he “didn’t have the foggiest idea what was going on in this very complicated case.” Still, in October 2008, the judge could not undo his prior order (on which financial firms and investors with significant exposure had relied) simply be- cause, in retrospect, the order had consequences he may not have fully appreciated at the time. In essence, the trustee urges us to adopt a rule providing that a trial court’s after-the-fact comments, however inconclu- sive, should be entitled to preclusive effect so as to supersede a prior binding order. That rule would be unrealistic, and we decline to adopt it. The rule would create new ambiguities that would set traps for parties who rely on court orders, and it would create both incentives and opportunities for their op- ponents to multiply litigation. Not only was the bankruptcy court’s October 2008 ruling unappealable by FCStone and other SEG 1 customers, it was indefinite and internally inconsistent, and it is entitled to no preclusive effect. Even if the trustee’s collateral estoppel argu- ment were not barred by the mandate rule and the law of the case, the argument would fail on its merits. We affirm Judge Zagel’s decision rejecting the trustee’s collateral attempt to re- vive his avoidance action against FCStone. Nos. 16-1896 & 16-1916 19 B. FCStone’s Cross-Appeal — The SEG 1 Reserve We now turn to FCStone’s cross-appeal. In authorizing the post-petition transfer following the Citadel sale, the bank- ruptcy court required the Bank of New York to hold back $15.6 million (about five percent of the sale proceeds). That amount, along with $4.9 million in proceeds from a late-set- tling security and certain proceeds of subsequent liquida- tions, remained in reserve in a SEG 1 account at the bank. As of September 30, 2014, that SEG 1 reserve account had a bal- ance of $24,551,622. In Count III of his operative complaint, the trustee seeks a declaration as to the ownership of these funds. The trustee ar- gues, and the district court concluded, that the funds should be treated as property of the estate that should be distributed pro rata among all Sentinel customers and other unsecured creditors (including the Bank of New York). Grede, 556 B.R. at 366. FCStone counters that the funds belong to it and other SEG 1 customers (the “SEG 1 Objectors”) who opposed early drafts of the Chapter 11 plan that would have treated all cus- tomers uniformly as unsecured creditors. FCStone argues that the funds are protected by a statutory trust and that it would be improper to disburse that trust property to other claim- ants—particularly the Bank of New York, which as we previ- ously determined was on “inquiry notice of Sentinel’s fraud.” In re Sentinel Management Group, Inc., 809 F.3d at 964. This issue arose on cross-motions for judgment and sum- mary judgment following a bench trial and subsequent re- mand. We review for clear error the district judge’s factual findings, but we review de novo his legal conclusions. Muham- mad-Ali v. Final Call, Inc., 832 F.3d 755, 760 (7th Cir. 2016). We hold that the funds in the SEG 1 reserve account are trust 20 Nos. 16-1896 & 16-1916 property belonging to FCStone and other SEG 1 Objectors. These claimants preserved their statutory trust rights and are entitled to the benefit of tracing conventions. Although there are understandable reasons for wanting to treat SEG 3 cus- tomers similarly based on their similar statutory protections, we conclude that the SEG 3 customers surrendered those pro- tections by agreeing to be treated as unsecured creditors un- der the confirmed Chapter 11 plan. Further, FCStone has shown that it can actually trace a portion of the reserve funds back to its initial investment, strengthening its claim to those funds. The district court’s reasons for rejecting the tracing would also, we believe, undermine important statutory pro- tections for FCM customers under the Commodity Exchange Act in any future FCM bankruptcies. We reverse the district court’s judgment on Count III and remand for entry of judg- ment in FCStone’s favor. 1. Statutory Trusts and Tracing Conventions We begin with a simple statutory proposition. Under the Bankruptcy Code, property held in trust by the debtor for a third party is not property of the debtor’s bankruptcy estate. See 11 U.S.C. § 541(d) (“Property in which the debtor holds, as of the commencement of the case, only legal title and not an equitable interest … becomes property of the estate … only to the extent of the debtor’s legal title to such property, but not to the extent of any equitable interest in such property that the debtor does not hold.”); Begier v. IRS, 496 U.S. 53, 59 (1990) (“Because the debtor does not own an equitable interest in property he holds in trust for another, that interest is not ‘property of the estate.’”); Marrs-Winn Co. v. Giberson Electric, Inc. (In re Marrs-Winn Co.), 103 F.3d 584, 589 (7th Cir. 1996) Nos. 16-1896 & 16-1916 21 (“From the plain language of the Code, one can easily con- clude that the debtor’s bankruptcy estate does not include property held in trust for another.”). Trust property does not lose its trust character simply be- cause, as in this case, the debtor misappropriated it or com- mingled it with the debtor’s own property. E.g., Old Republic Nat’l Title Ins. Co. v. Tyler (In re Dameron), 155 F.3d 718, 723–24 (4th Cir. 1998) (“courts have consistently rejected the notion that commingling of trust property, without more, is suffi- cient to defeat tracing”); Connecticut General Life Ins. Co. v. Uni- versal Ins. Co., 838 F.2d 612, 619 (1st Cir. 1988) (“mere coming- ling of the trust property with other property of the bankrupt corporation … does not defeat [trust beneficiary’s] claim”); Turley v. Mahan & Rowsey, Inc. (In re Mahan & Rowsey, Inc.), 817 F.2d 682, 684 (10th Cir. 1987) (the “trust pursuit will even al- low tracing of trust funds into a commingled mass”); accord, Universitas Education, LLC v. Nova Group, Inc., Nos. 11 Civ. 1590(LTS)(HBP) & 11 Civ. 8726(LTS)(HBP), 2013 WL 6123104, at *12 (S.D.N.Y. Nov. 20, 2013); Appalachian Oil Co. v. Kentucky Lottery Corp. (In re Appalachian Oil Co.), Bankr. No. 09-50259, 2012 WL 1067731, at *7 (Bankr. E.D. Tenn. Mar. 23, 2012); Han- ley v. Notinger (In re Charlie’s Quality Carpentry, LLC), No. 02- 11983-JMD, 2003 WL 22056647, at *5 (Bankr. D.N.H. Aug. 25, 2003); Carlson Orchards, Inc. v. Linsey (In re Linsey), 296 B.R. 582, 586 (Bankr. D. Mass. 2003). The trustee correctly acknowledges that FCStone and the other SEG 1 customers were the beneficiaries of a statutory trust. Under the Commodity Exchange Act, it is “unlawful for any person to be a futures commission merchant unless … such person shall … treat and deal with all … property re- ceived by such person to margin, guarantee, or secure the 22 Nos. 16-1896 & 16-1916 trades or contracts of any customer of such person, or accru- ing to such customer as the result of such trades or contracts, as belonging to such customer.” 7 U.S.C. § 6d(a) (emphasis added); see also 17 C.F.R. § 1.20(f)(3) (“No person … that has received futures customer funds for deposit in a segregated account … may hold, dispose of, or use any such funds as be- longing to any person other than the futures customers of the futures commission merchant which deposited such funds.”); Marchese v. Shearson Hayden Stone, Inc., 822 F.2d 876, 878 (9th Cir. 1987) (“The legislative history of the Act makes it clear that [§ 6d] establishes a specific statutory trust … and this fact has long been recognized.”). There are powerful policy reasons for according robust protection to investors whose trust property is covered by the Commodity Exchange Act. The CFTC, the regulator responsi- ble for enforcing the Act (and amicus in earlier proceedings), explained: The ability of participants in futures markets to rely on the protections provided by section 6d when an FCM becomes insolvent is critical to the functioning of these markets. … In futures transactions there is no equivalent of federal de- posit insurance for bank depositors or the Secu- rities Investor Protection Corporation Fund to protect securities investors … . Instead, the re- quirements of section 6d are the principal legal protection for commodity customer funds against wrongdoing or insolvency by FCMs and their depositories. … Participants in the futures market rely on this protection … and customers’ ability to rely on this protection when an FCM Nos. 16-1896 & 16-1916 23 faces insolvency contributes to the stability of markets in times of stress. Supplemental Amicus Curiae Memorandum of CFTC at 7–8, Grede v. FCStone, LLC, 485 B.R. 854 (N.D. Ill. 2013) (No. 09 C 136, Dkt. Entry 87), reprinted in FCStone App. at 839–40. The Futures Industry Association, Inc., a trade organiza- tion and amicus curiae in the current appeals, has made a sim- ilar point, describing the district court’s decision on remand as opening the door for “non-futures claimants in future FCM bankruptcies to litigate rights to futures margin account prop- erty, creating perilous delay and rendering unpredictable the return of futures customers’ assets.” Brief of Amicus Curiae, Futures Industry Ass’n, at 19. The statutory trust assures futures customers that their funds will be protected from an FCM’s general creditors and other customer classes. The trust also assures futures custom- ers that, in the event of an FCM failure, “funds and property can be immediately transferred to the segregated customer accounts of solvent FCMs to … support the ongoing obliga- tions of open trades” and thus can prevent the bankruptcy of a single FCM from starting a domino effect that overwhelms other firms and the larger market. Id. at 20. Cf. In re JPMorgan Chase Bank, NA, Comm. Fut. L. Rep. (CCH) ¶ 32,156, 2012 WL 1143791, at *5 (C.F.T.C. 2012) (“Without immediate access to customer funds, the FCM is hindered in its ability to satisfy margin requirements. In times where there is a market dis- ruption, any impediment or restriction upon the ability to im- mediately withdraw funds ‘could magnify the impact of any market disruption and cause additional repercussions.’”) (ci- tation omitted). We agree that the Commodity Exchange Act 24 Nos. 16-1896 & 16-1916 and Bankruptcy Code should be interpreted and applied with these concerns in mind. Though the trustee concedes that the SEG 1 customers were entitled to statutory trust protection, he argues that (1) the SEG 3 customers were likewise the beneficiaries of a stat- utory trust and (2) the two customer classes are similarly sit- uated under the confirmed Chapter 11 plan. The trustee is cor- rect as to the first point: the Investment Advisers Act, like the Commodity Exchange Act, creates a trust in favor of inves- tors.5 With respect, however, we believe the trustee is mis- taken as to the second point, and that undermines his argu- ment. This issue concerning the parallel statutory trusts presents the most difficult equitable problem we have confronted in these protracted bankruptcy proceedings. It was understand- ably the focus of the district court’s thinking. As we suggested in FCStone I, the default rule should be to treat customers pro- tected by statutory trusts under the Commodity Exchange 5 See 17 C.F.R. § 275.206(4)–2(a) (“If you are an investment adviser … it is a fraudulent, deceptive, or manipulative act … for you to have custody of client funds or securities unless: (1) … A qualified custodian maintains those funds and securities: (i) In a separate account for each client under that client’s name; or (ii) In accounts that contain only your clients’ funds and securities, under your name as agent or trustee for the clients.”). While this trust language appears in regulations promulgated by the SEC (whereas the Commodity Exchange Act trust is established by the Act’s statutory language itself), we recognized in FCStone I that there is “no le- gal basis for placing one trust ahead of the other,” 746 F.3d at 259; see also Chrysler Corp. v. Brown, 441 U.S. 281, 295 (1979) (“It has been established in a variety of contexts that properly promulgated, substantive agency regulations have the ‘force and effect of law.’”) (footnote omitted). Nos. 16-1896 & 16-1916 25 Act and those protected by statutory trusts under the Invest- ment Advisers Act as similarly situated. In this case, the SEG 1 Objectors, including FCStone, steadfastly asserted their rights under the Commodity Exchange Act. The SEG 3 cus- tomers, however, agreed to be treated as unsecured creditors without any carve-out or exception for any statutory trust claim that they might otherwise have brought. That agreement has consequences that we cannot over- look. See Ernst & Young LLP v. Baker O’Neal Holdings, Inc., 304 F.3d 753, 755 (7th Cir. 2002) (“A confirmed plan of reorganiza- tion is in effect a contract between the parties and the terms of the plan describe their rights and obligations.”), citing In re Chicago, Milwaukee, St. Paul & Pacific R.R. Co., 891 F.2d 159, 161 (7th Cir. 1989); see also 11 U.S.C. § 1141(a) (“[T]he provisions of a confirmed plan bind … any creditor … whether or not the claim or interest of such creditor … is impaired under the plan and whether or not such creditor … has accepted the plan.”); In re Harvey, 213 F.3d 318, 321 (7th Cir. 2000) (“[A] confirmed plan acts more or less like a court-approved contract or con- sent decree that binds both the debtor and all the creditors.”). 6 The plan’s language confirms that the SEG 1 Objectors and SEG 3 customers are situated differently under the plan itself. 6 To the extent we suggested in dicta in FCStone I that the two cus- tomer classes are similarly situated, that dictum is not controlling here. We made clear that we were not deciding the property-of-the-estate issue in our earlier opinion. We decided only that the post-petition transfer was authorized by the bankruptcy court. While we believe our prior dictum is correct as a general proposition, our earlier decision did not require us to scrutinize the provisions of the confirmed Chapter 11 plan that, we hold in this cross-appeal, treat the SEG 1 Objectors differently from the SEG 3 customers. 26 Nos. 16-1896 & 16-1916 Under Section 4.4(a), customer claimants are entitled to dis- tributions as set forth in Section 4.5(a). That section in turn provides that customer claimants and unsecured creditors are entitled to pro rata distributions of “Cash and Cash proceeds of all Property, including Customer Property, not allocated for payment of Allowed Claims in other Classes.” Section 7.20(a) then creates an exception: “Pending a determination by the Court whether the assets held in the SEG 1 Property Of The Estate Reserve … are property of the Estate, the Trustee shall continue to maintain the Property of the Estate Reserve[].” Section 7.20(a)(i) prescribes the “Seg 1 Property Of The Estate Reserve” as follows: On the Effective Date, the Liquidation Trustee shall establish a reserve equal to the amount of all funds held in any bank account denominated as a SEG 1 account, multiplied by a fraction, the numerator of which is the amount of Citadel Beneficiary … Customer Claims attributable to SEG 1 accounts … which voted against the Plan and/or lodged objections thereto, and the de- nominator of which is the total aggregate amount of … Customer Claims attributable to SEG 1 accounts. As the trustee himself explained prior to plan confirma- tion, “the Plan Proponents have established reserves to ad- dress the Seg 1 Objectors’ contention that certain funds are not property of the estate, see Plan § 7.20, and, as such, Customers will share pro rata with Holders of General Unsecured Claims only in property that the Court determines is property of the estate.” There are no similar provisions for SEG 3 customers. Nos. 16-1896 & 16-1916 27 Section 7.20(c)(i) explains how the disputed funds should be disbursed if the court “determines that the property … is not property of the estate.” In that event, “Sections 4.4 and 4.5 of the Plan shall be deemed modified to provide that Cus- tomer Property shall be distributed to the rightful owners of such property or to the Estate, as determined by the Court.” Reading all these provisions together, as we must, we find that while the confirmed plan treats SEG 1 and SEG 3 custom- ers by default as unsecured creditors, the SEG 1 Objectors alone preserved their right to recover trust property held in reserve, and the plan specifically contemplates that such property may be restored to those customers. SEG 3 custom- ers simply did not preserve a comparable right. The trustee argues that the final phrase of Section 7.20(c)(i)—“as determined by the Court”—authorizes the court (1) to find that the reserve funds are trust property be- longing to the SEG 1 Objectors, yet nevertheless (2) to distrib- ute the property pro rata to all customers and unsecured cred- itors. Though the text in a vacuum offers some support for the trustee’s reading, that reading would lead to a nonsensical re- sult. If the reserve funds belong to the SEG 1 Objectors, the court cannot simply disregard that fact and split the funds among differently situated creditors. For that matter, if the court were vested with such unfettered discretion, what would be the point of the extended adversarial proceedings that have already taken place? We decline to interpret the confirmed Chapter 11 plan as authorizing an absurd outcome or inviting futile litigation. E.g., BKCAP, LLC v. Captec Franchise Trust 2000-1, 572 F.3d 353, 360 (7th Cir. 2009) (where literal application of text would “lead to absurd results” and “thwart the obvious intentions of 28 Nos. 16-1896 & 16-1916 its drafters,” we cannot rely solely on plain language) (cita- tions and internal quotation marks omitted). In context, the phrase “as determined by the Court” refers to the threshold determination whether property is correctly characterized as trust property or property of the estate. Only property of the estate could and should be distributed pro rata to creditors. Property belonging to others must be returned to them. Since only the SEG 1 Objectors preserved their status as trust claimants with respect to the SEG 1 reserve funds, the problem of co-equal trust claimants addressed in Cunningham v. Brown, 265 U.S. 1 (1924), the key case on which the district court relied, is absent here. In its original opinion, the district court indicated that, if it were dealing with a single class of trust claimants, it would “apply every reasonable tracing fic- tion available to preserve the … trust.” Grede, 485 B.R. at 878. The district court’s original view is consistent with our dictum in FCStone I: we emphasized the “national interest” in protect- ing statutory trust claimants, and we suggested that, in a case involving competing trusts, claimants who cannot actually trace might still be entitled to priority over at least unsecured creditors. 746 F.3d at 259. Because this case no longer involves competing trust claims by SEG 1 customers under the Commodity Exchange Act and SEG 3 customers under the Investment Advisers Act, we need not rely on our proposed rule of priority. Instead, to ensure that the goals underlying the Commodity Exchange Act are honored, we should accord FCStone and the other SEG 1 Objectors every reasonable opportunity to recover their trust property. If such a customer can trace its initial invest- ment to funds remaining under the control of the Sentinel Li- Nos. 16-1896 & 16-1916 29 quidation Trust, that customer should be entitled to its pro- portionate share of those funds. And if the customer cannot actually trace, it should nevertheless be entitled to rely on rea- sonable tracing conventions (or “fictions”). Though a variety of tracing conventions might be helpful in a case like this, the CFTC proposed that “assets in the Sentinel Seg 1 accounts and portfolios at the time of the Sentinel bankruptcy must be con- sidered to have been held in a trust under the [Act] independ- ent of any requirement on the part of customers to trace par- ticular assets.” Supplemental Amicus Curiae Memorandum of CFTC at 7–8, Grede v. FCStone, LLC, 485 B.R. 854 (N.D. Ill. 2013) (No. 09 C 136, Dkt. Entry 87), reprinted in FCStone App. at 842. We agree with the CFTC’s view, though that logic would have extended to customers protected by the Investment Ad- visers Act as well, at least if they had not agreed to relinquish any such rights. 7 The securities Sentinel sold to Citadel in August 2007 were (with limited exceptions) segregated for the benefit of SEG 1 prior to the sale; the proceeds of that sale were deposited in segregation for SEG 1; and funds attributable to the five per- cent holdback and subsequent liquidations were likewise 7 We recognize that, unlike SEG 1, SEG 3 customers did not have hun- dreds of millions of dollars in securities or proceeds in their segregated accounts when Sentinel filed for bankruptcy protection. As the last cus- tomer class Sentinel raided before going belly-up, SEG 3 was in an inher- ently weaker negotiating position than SEG 1. But by agreeing to a plan that treated them as unsecured creditors, SEG 3 customers gave up what- ever trust claims they might otherwise have asserted. As a practical mat- ter, these customers may ultimately recover less than some other creditor classes. We cannot cover their losses with funds rightfully belonging to SEG 1 customers. 30 Nos. 16-1896 & 16-1916 kept in segregation. Pursuant to the tracing convention pro- posed by the CFTC and urged by the Futures Industry Asso- ciation and FCStone, the funds should be disbursed pro rata among the SEG 1 Objectors. 8 2. Actual Tracing As set forth above, FCStone and other SEG 1 Objectors are entitled to the full benefit of reasonable tracing conventions to recover their trust property. But even apart from those con- ventions or presumptions, FCStone has shown an independ- ent basis for its claim to a share of the SEG 1 reserves. It can actually trace its initial investment to the proceeds of the Cit- adel security sale (both those proceeds disbursed in the Au- gust 2007 post-petition transfer and those remaining in re- serve). FCStone has done so through the essentially unrebut- ted report and testimony of its key expert, Frances McCloskey, a certified public accountant with extensive experience in fo- rensic accounting. For this additional reason, FCStone is enti- tled to judgment on Count III of the trustee’s operative com- plaint. a. Trial Testimony McCloskey testified at trial that she had traced (1) all cash and securities within all of Sentinel’s records (not only those 8 At oral argument, FCStone contended that beneficiaries of a statu- tory trust should enjoy an “irrebuttable” presumption of entitlement to assets held in a segregated account “no matter how they got there.” We need not and do not endorse this view. Suppose, for example, that a seg- regated trust account were shown to contain at the time of a bankruptcy filing a balance exceeding its beneficiaries’ contributions and earnings. In such a scenario, an irrebuttable presumption of entitlement obviously would be unjustified. Nos. 16-1896 & 16-1916 31 assets allocated to SEG 1) for 2007, the year Sentinel failed; (2) all customer deposits during that same year to the securities allocated on customers’ statements; and (3) the allocation of all securities included in the Citadel sale back to their original dates of purchase (as early as 2004). She then explained how she accomplished this task. Securities are identified by Com- mittee on Uniform Security Identification Procedures (CUSIP) numbers, which are similar to serial numbers. Using these CUSIP numbers, McCloskey was able to test and prove the accuracy of two different accounting ledgers that Sentinel maintained on a daily basis. One of these two ledgers, the customer ledger, reflected all securities that had been allocated to a particular customer group. The securities allocated to—meaning owned by—a group would change to reflect the value of clients’ cash depos- its or redemptions. In other words, Sentinel “would exchange customer balances for an exact interest in the market value of a security.” Sentinel frequently participated in repurchase or “repo” transactions, that is, transactions where “one party … sells a security to a counterparty … with an agreement to re- purchase the security later with interest. “ Bloom, 846 F.3d at 248. McCloskey found that the customer ledger never re- flected “repo” transactions in which securities that Sentinel had loaned to counterparties. Nor did it include unallocated, house-owned securities. Sentinel’s accounting system also included a securities in- ventory, which, in contrast to the customer ledger, listed by CUSIP number every security controlled by Sentinel, includ- ing “repo” securities out on loan and Sentinel’s own ever- 32 Nos. 16-1896 & 16-1916 expanding pool of riskier securities not allocated to custom- ers. The securities inventory contained no information rele- vant to customer transactions. According to McCloskey, how- ever, the customer ledger accurately “reflected every cus- tomer and every group and each security’s allocation or each customer’s interest in a security.” McCloskey testified that she performed extensive testing to verify the accuracy of these ledgers and could trace, “to a reasonable degree of accounting certainty, the value that cus- tomers deposited at Sentinel, the balances that Sentinel held for their … benefit, and the securities that Sentinel held each day.” She found no instances of missing, fictitious, or double- allocated securities. To the contrary, Sentinel’s customer state- ments reflected to the penny the “economic exchange that oc- curred between the customers and Sentinel for an interest in a pool of securities valued at market for that day” and repre- sented a “true economic transaction.” McCloskey acknowledged, of course, that Sentinel had improperly commingled and illegally pledged customer- owned securities as collateral for its own loan. Even so, McCloskey did not find, after validating the accounting for all of Sentinel’s securities inventory for more than 200 business days, any misappropriation of customer assets apart from those segregation violations. Sentinel’s misuse of customer as- sets, although illegal, did not impede McCloskey’s ability to trace the assets. The assets still belonged to the respective cus- tomers despite having been placed, illegally and temporarily, at risk. The securities Sentinel sold to Citadel, with few excep- tions, had been out of segregation for no more than two brief periods—but long or short, those periods of unlawful conduct Nos. 16-1896 & 16-1916 33 did not affect the tracing analysis. McCloskey explained that the potential risks associated with Sentinel’s improper use of customer assets did not “have anything to do with what the customer accounting and customer records reflect in terms of the allocation of securities and the exchange of customer bal- ances for interests [in] pools of securities.” For his part, the trustee argued that the customer records on which McCloskey relied should be disregarded in their en- tirety because Sentinel’s segregation violations rendered its record-keeping “a complete fraud” that “can’t be the starting point for a tracing analysis.” The trustee grounded this asser- tion in expert testimony by James Feltman, also a CPA. Felt- man conceded that Sentinel’s customer ledger consistently matched the allocation of securities on customers’ statements, but he rejected that data as meaningless because “billions of dollars of [house-owned] securities that are in Sentinel’s in- ventory … don’t show up on customer statements,” and be- cause Sentinel kept customers in the dark about the risk in- herent in its improper pledging of customer assets to secure the Bank of New York loan. Feltman opined that tracing was inappropriate even for securities that had never been coming- led or subject to a lien in the bank’s favor because the securi- ties Sentinel allocated to customer groups typically had been purchased years earlier. In Feltman’s view, the allocation pro- cess was “arbitrary and didn’t have real economic meaning.” McCloskey countered (without contradiction) that Felt- man had analyzed only the omnibus custodial account rec- ords at the Bank of New York and the Sentinel securities in- ventory, which, McCloskey said, by their “very nature” are not designed to reflect balances or transactions on a customer- 34 Nos. 16-1896 & 16-1916 by-customer basis. She contended that Feltman had disre- garded the “fundamental concept of book-entry customer ac- counting” and “look[ed] in the wrong place” by simply ignor- ing Sentinel’s detailed customer ledgers. McCloskey added that Feltman’s conclusions were “inaccurate and misleading” because he had focused on the risks from collateralizing cli- ent-owned securities, which had no bearing on the ownership of those securities. She opined that his “notion that the cus- tomer ledger is somehow fiction or it doesn’t represent reality is just wrong,” because it ignores the consistent accuracy of Sentinel’s record-keeping. b. First District Court Opinion In its initial opinion, the district court concluded that, us- ing the ledgers and data identified by McCloskey, “it is possi- ble to identify: 1) the custodial location of every Sentinel se- curity held at [the Bank of New York] for all relevant time pe- riods; and 2) the indirect beneficial ownership interest in these securities that Sentinel assigned its customers.” Grede, 485 B.R. at 864. Nonetheless, the court rejected McCloskey’s testi- mony, calling “nonsensical” her claim to have traced “custom- ers’ indirect beneficial ownership in securities” and finding that “tracing is not possible in this case.” Id. at 878–79 (citation omitted). The district judge did not, as we read his decision, ques- tion McCloskey’s credibility. He said explicitly that he did not “disparage the work of Ms. McCloskey, who is an accountant not an attorney” and therefore “should not be expected to un- derstand arcane common law tracing rules.” Id. at 879 n.20. The judge then explained why, in his view, tracing was not possible: Nos. 16-1896 & 16-1916 35 But for tracing purposes the critical shortcom- ing of Ms. McCloskey’s report is that it fails to adequately account for the fact that none of Sen- tinel’s customers held specific ownership inter- ests in securities. Rather, they owned pro rata portions of investment portfolios, which Senti- nel was free to fill with any of the securities in its pool of assets so long as those securities met the portfolio’s investment criteria. Further, these securities were generally purchased [originally] with commingled funds … . The upshot is that the securities held in a given customer group portfolio at any time were not necessarily—in- deed, were most improbably—the converted form of the original trust property (i.e. cash de- posits) of the customers within that group. Id. at 879. The judge also asserted that the “fungible nature of cash alone makes it impossible to trace specific securities back to original customer deposits,” and thus, “commingling aside, Sentinel’s investment model makes tracing essentially impossible because, upon deposit, customer funds were im- mediately converted into an abstract ownership interest.” Id. In other words, according to the district court, “Sentinel’s pooled investment model renders tracing impracticable be- cause there is no specific form of converted trust property to trace.” Id. (emphasis omitted). c. Remand Proceedings After our remand in FCStone I, the parties resumed their battle over the reserves and submitted dueling motions for judgment and summary judgment. FCStone filed an affidavit from McCloskey addressing specifically the district court’s 36 Nos. 16-1896 & 16-1916 reasons for rejecting her tracing analysis. She contended that the district court had misunderstood both tracing principles and “how securities markets actually function.” The court, she asserted, had overlooked her testimony that customers’ cash deposits were exchanged for an ownership interest in specific, identifiable securities held in Sentinel’s inventory, which became the traceable trust res. She provided a detailed description of tracing methodology in financial services firms. McCloskey then explained why the fungible nature of cash does not defeat tracing. Customer funds were wired into seg- regated accounts, and these cash deposits were exchanged daily for ownership interests in CUSIP-identifiable securities, all verifiable through time stamps. McCloskey also challenged the district court’s conclusion that the use of a single clearing account impedes tracing. She clarified that the use of such an account is standard industry practice, pointing out that if the “use of a single clearing ac- count precluded the ability to trace, FCMs and broker-dealers nationwide would not be able to account for or hold customer assets,” a result “contrary to accounting principles and indus- try recordkeeping requirements.” Finally, she explained how the use of pooled FCM customer securities accounts is stand- ard industry practice and consistent with CFTC regulations. The district court’s remand opinion again rejected McClos- key’s testimony. The court declared tracing “difficult here, if not impossible, because Sentinel’s commingling prior to its bankruptcy filing was so appalling.” Grede, 556 B.R. at 365. The court was particularly troubled by Sentinel’s use of “SEG 1 securities in an unauthorized repo transaction,” id. at 364, but did not otherwise give reasons for rejecting McCloskey’s unrebutted analysis. Nos. 16-1896 & 16-1916 37 d. Discussion We agree with FCStone that the district court’s explanation for rejecting Frances McCloskey’s tracing analysis is not sound and would have troubling implications for protection of FCM customers more broadly. McCloskey testified that “it was actually pretty easy” to follow the trail of assets given the accuracy of Sentinel’s customer ledger. The trustee not only failed to rebut this analysis but conceded the consistency of the customer records during closing argument at trial. Instead of pointing to evidence to rebut McCloskey’s anal- ysis, the trustee continues to argue that FCStone relied on “phony” records and a “fictional, arbitrary allocation process with no basis in reality.” The trustee is arguing in essence that because Sentinel unlawfully used customer assets as collat- eral for its own borrowing, all of its records amount to noth- ing more than smoke and mirrors. The trustee also contends that Sentinel’s customer ledgers are unreliable because they do not show securities owned by the house (which seems un- surprising to us since these are customer ledgers), and because customer statements did not disclose the risk to customers from Sentinel’s use of their assets as collateral to support its own leveraged trading strategy. But the trustee cites no legal authority to show that these facts render Sentinel’s internal records meaningless, and he cites no record evidence to show that McCloskey’s tracing analysis was flawed. In our view, McCloskey’s forensic analysis therefore remains unrebutted. The trustee insists that tracing is “improper” because it is “merely fortuitous” that Sentinel, on the eve of bankruptcy, had swapped improperly collateralized SEG 1 securities for assets owned by SEG 3 clients. This argument, though central to the trustee’s position and the district court’s analysis, is a 38 Nos. 16-1896 & 16-1916 red herring. All parties to this case agree that Sentinel broke the law by using client assets as collateral for its Bank of New York loan. (Two of Sentinel’s executives are serving prison sentences, after all.) But that fact does not mean that FCStone cannot prove what it owned. Sentinel risked customer assets by pledging them as collateral, but that misconduct did not affect McCloskey’s ability to trace those assets. The fact that SEG 3 customers happened to be the last victims of Sentinel’s machinations does not confer upon the district court broad equitable discretion to remedy their injury at the SEG 1 cus- tomers’ expense. The trustee also argues that the district judge found “that FCStone’s tracing expert lacked credibility,” but the judge said no such thing. Rather, the judge characterized McCloskey’s work as valuable to an “understanding of the facts” and ex- pressly avoided disparaging her work or her credibility. Grede, 485 B.R. at 879 n.20. The judge rejected McCloskey’s conclusion that beneficial interests in pooled securities can be traced easily because, as “an accountant not an attorney,” McCloskey “should not be expected to understand arcane common law tracing rules.” Id. With respect, that assertion is mistaken: McCloskey is a forensic accountant, an expert in fi- nancial tracing. As an expert witness in this case, it was her task to break down complex accounting principles for the trier of fact. She did so, tracing customer assets through the pooled investment portfolios for each group of customers, just as FCMs and other financial institutions do routinely. The ability to do so is critical to the day-to-day operation of FCMs and commodity markets and to customer confidence that their property will be protected in the event of an FCM bank- ruptcy. Nos. 16-1896 & 16-1916 39 In urging us to affirm the district court’s ruling on FCStone’s cross-appeal, the trustee argues that mixed ques- tions of law and fact not involving constitutional issues are generally reviewed for clear error, see United States v. Freder- ick, 182 F.3d 496, 499 (7th Cir. 1999), and that we accordingly owe substantial deference to the district court’s findings. The trustee’s explanation of the standard of review is consistent with the weight of authority. See, e.g., Isby v. Brown, 856 F.3d 508, 521 (7th Cir. 2017); Muhammad-Ali, 832 F.3d at 760; Trovare Capital Group, LLC v. Simkins Industries, Inc., 794 F.3d 772, 778 (7th Cir. 2015); Morisch v. United States, 653 F.3d 522, 528 (7th Cir. 2011). But see Mungo v. Taylor, 355 F.3d 969, 974 (7th Cir. 2004) (“Both questions of law and mixed questions of law and fact … are reviewed de novo.”); FCStone I, 746 F.3d at 251 (same); In re Longview Aluminum, L.L.C., 657 F.3d 507, 509 (7th Cir. 2011) (same). Our primary disagreement with the district judge’s analy- sis on this tracing issue, however, concerns not his factual findings (most of these are undisputed) but rather his legal conclusions stemming from and relating to those findings. Those legal conclusions are subject to de novo review. The district judge concluded that tracing is impossible be- cause Sentinel’s customers held only pro rata interests in a pooled investment portfolio (i.e., a share of the securities allo- cated to each investment group) rather than discrete interests in particular securities. Sentinel’s investment model does not render tracing impossible. It is undisputed that pooling and allocation is standard industry practice, and CFTC regula- tions expressly permit FCM customer funds to be held and invested in this manner. See 17 C.F.R. §§ 1.20–1.23. The district court’s grounds for rejecting McCloskey’s tracing run contrary 40 Nos. 16-1896 & 16-1916 to the regulations that protect FCM customers’ funds. If ac- cepted more broadly, that view could disrupt the futures mar- ket by thwarting the ability of investors to trace assets held in a trust in which multiple beneficiaries maintain undivided in- terests. The likely result would be to undermine confidence of FCM customers that they would actually receive the prom- ised statutory protection in a future FCM bankruptcy. The district court also concluded that the fungible nature of cash renders tracing impossible, a contention the trustee defends by insisting that “because cash is fungible, FCStone is necessarily confusing actual tracing with tracing with the benefit of a fiction.” That logic would hold, however, only if Sentinel maintained customer assets as one undifferentiated pool of cash—similar to the original Ponzi scheme discussed in Cunningham, 265 U.S. at 7–9. Instead, Sentinel exchanged customer deposits for a beneficial ownership interest in iden- tifiable securities on a daily basis. The trustee makes much of the fact that securities allocated to customers often were pur- chased by Sentinel much earlier. But the fact that Sentinel used a buy-and-hold strategy for its securities is irrelevant. The process of converting cash deposits into identifiable secu- rities was unaffected by whether Sentinel already owned the securities or purchased them on the open market in response to new customer deposits. For similar reasons, the district court erred by concluding that Sentinel’s use of a single clearing account impeded the ability to trace customer ownership of securities. McCloskey testified not only that use of a single clearing account is stand- ard industry practice but also that if this practice impeded the ability to trace, “FCMs and broker-dealers nationwide would not be able to account for or hold customer assets,” a result Nos. 16-1896 & 16-1916 41 “contrary to accounting principles and industry recordkeep- ing requirements.” The trustee did not address this problem on appeal. The trustee has cited as supplemental authority our deci- sion in Bloom, 846 F.3d 243, which was issued after the briefing in these appeals. See Fed. R. App. P. 28(j). He contends that our “detailed discussion of Sentinel’s fraudulent operation supports the Trustee’s position that FCStone failed to meet its burden of tracing its property to the reserve account because Sentinel’s fraudulent conduct made tracing impossible with- out the benefit of tracing fictions.” On the contrary, our dis- cussion in Bloom undermines James Feltman’s opinion that Sentinel’s allocation of securities within its customer ledgers amounted to a fiction. Eric Bloom, Sentinel’s former president and CEO, challenged the sufficiency of the evidence support- ing his fraud conviction. The government had offered three theories of fraud, one of which was that Bloom had intention- ally manipulated customer yield rates to inflate the returns to SEG 1 customers while attributing less interest to SEG 3 than those customers’ securities actually earned. Bloom, 846 F.3d at 252. To prove its theory, the government had to show which securities were owned by each customer group on a given day and to compare the market interest earned by those securities with the amounts reflected on customer statements. Although the jury’s guilty verdicts in Bloom do not establish that McCloskey’s tracing analysis must be accepted, these verdicts do undercut the trustee’s position that Sentinel’s fraud ren- dered tracing impossible. In short, the district court’s rationale for rejecting McClos- key’s detailed tracing analysis turned on flawed legal conclu- 42 Nos. 16-1896 & 16-1916 sions, not factual determinations, and the trustee failed to re- but FCStone’s evidence of actual tracing. FCStone is therefore entitled to judgment on Count III, and FCStone and the other SEG 1 Objectors are entitled to share pro rata in the SEG 1 re- serve. C. Disputed Claims Reserve Before we conclude, we address briefly the Section 7.20(b) “Disputed Claims Reserve,” a separate reserve fund estab- lished under the confirmed Chapter 11 plan that contained about $3.7 million as of September 30, 2014. FCStone acknowl- edges the Disputed Claims Reserve in its appellate brief, but it does not discuss the reserve at length, and the trustee ig- nores the reserve completely. The Disputed Claims Reserve consists of funds that “any Citadel-Beneficiary Customer Claim which voted against the Plan and/or lodged objections thereto[] would be entitled to receive” but for language in Section 4.5(a) that requires SEG 1 Citadel sale beneficiaries to wait on further distributions until other customers catch up to their level of recovery. The Dis- puted Claims Reserve was designed to capture the pro rata portions of litigation recoveries and similar distributions that SEG 1 Objectors would have received had the parties agreed up front that the Citadel sale proceeds were SEG 1 trust prop- erty (and therefore should not count against the SEG 1 Objec- tors’ pro rata share in the property of the estate). As discussed in Part II-B, however, we conclude that the SEG 1 reserve funds are trust property belonging to the SEG 1 Objectors both because these statutory trust claimants are entitled to the benefit of tracing conventions and because FCStone has shown that it is possible to trace portions of the Nos. 16-1896 & 16-1916 43 reserve back to the customers’ initial investment. Likewise, FCStone has shown that it is possible to trace the $297 million post-petition transfer to the Citadel security sale and further to trace the beneficial ownership of those securities all the way back to the dates they were first acquired by Sentinel. Those securities were overwhelmingly allocated to (and segregated for) SEG 1 customers as of the Citadel sale date. The confirmed Chapter 11 plan accounted for the possibil- ity that the courts might ultimately side with the SEG 1 Ob- jectors on the property-of-the-estate dispute. In addition to Section 7.20(c)(i), which, as discussed above, modifies the plan’s distribution provisions if a reviewing court determines that the SEG 1 reserve is not property of the estate, Section 7.20(c)(ii) provides: In the event the Court determines that the Cita- del Sale Distributions did not constitute distri- butions of property of the estate, the Claims of Citadel-Beneficiary Customers shall, to the ex- tent such Claims become Allowed Claims, be entitled to pro rata distributions with all other Holders of Allowed … Claims with respect to all Property other than Customer Property, without regard to … Plan provisions which pro- vide that until all Holders of Allowed … Cus- tomer Claims that are NonCitadel-Beneficiary Customers shall have received a Percentage Re- covery on account of such Claims equivalent to the applicable Citadel-Beneficiary Customer, such Citadel-Beneficiary Customer shall not be entitled to a distribution. 44 Nos. 16-1896 & 16-1916 Section 7.11, which governs disputed claims reserves gener- ally, similarly provides that “to the extent any … Disputed Claim becomes an Allowed Claim by Final Order, the relevant portion of the Cash held in the Disputed Claims Reserve therefor shall be distributed … to the Claim Holder in a man- ner consistent with distributions to similarly situated Al- lowed Claims.” In light of these provisions and our conclu- sions about the post-petition transfer and the SEG 1 reserve funds, the Section 7.20(b) Disputed Claims Reserve should be liquidated and the funds disbursed to SEG 1 Objectors who would have received these funds but for the property-of-the- estate dispute. For the reasons we have explained, we AFFIRM the dis- trict court’s judgment as to Counts I and V of the trustee’s op- erative Second Amended Complaint. We REVERSE with re- spect to Count III and REMAND with instructions to enter judgment for FCStone on that count and for further proceed- ings consistent with this opinion.