Doctors Hospital of Hyde Park was founded (as “Illinois Central Hospital”) to provide medical care as a fringe benefit for workers of the Illinois Central Railroad. Construction began in 1914; the architects were Schmidt, Garden & Martin. The building is on several lists of memorable designs. But the Illinois Central, like other railroads, eventually decided that it did not have a comparative advantage in providing medical care and sold the business.
Between 1992 and 2000 the Hospital was a Subchapter S corporation controlled by James Desnick, an ophthalmologist with a checkered past. Desnick once operated a chain of eye-care clinics, whose business practices garnered adverse publicity. See Desnick v. American Broadcasting Cos., 233 F.3d 514 (7th Cir.2000). Following charges of misconduct during the 1980s, Desnick gave up his medical practice in 1991 and bought the Hospital the next year. In 1999 and 2000 Desnick paid civil penalties of some $18.5 million to the Medicare and Medicaid programs on account of the Hospital’s excessive bills — not only “upcoding” to put services in categories that led to greater reimbursement, but also claims for medically unnecessary *690procedures or work never done at all. The Hospital also was inefficient, and not only because the old building’s design is not well suited to modern medicine: patient stays were significantly longer than the national average. Because third-party payments often cover a particular procedure rather than the number of days spent in a hospital room, this led to lower average revenues per patient-day. Doctors Hospital closed its doors in 2000; the building has been vacant since then. (There are plenty of other hospital beds in or near Hyde Park at the University of Chicago’s sprawling medical center plus the Jackson Park Hospital and Medical Center.)
Like other medical providers, the Hospital furnished services well before it received payment from patients or their insurers. This created a cash-flow problem, which the Hospital addressed by borrowing money. The current dispute arises from two of these loans. The Hospital’s trustee in bankruptcy proposes to recover, as fraudulent conveyances, some of the payments made during the last years before the Hospital entered bankruptcy.
In March 1997 Daiwa Healthco extended a revolving $25 million line of credit to MMA Funding, L.L.C., which made the money available to the Hospital for operating expenses. (Desnick owned 99% or more of MMA Funding and all other Hospital-related entities mentioned in this opinion. Anyone interested in details can consult the lengthy opinions of the bankruptcy judge and district judge. To make this opinion manageable, we simplify the facts ruthlessly.) The Hospital transferred all of its current and future accounts receivable to MMA Funding, which gave Daiwa a security interest in them. The plan of this transaction was to use MMA Funding as a “bankruptcy-remote vehicle” so that Daiwa could be assured of repayment even if the Hospital entered bankruptcy.
In August 1997 Nomura Asset Capital Corporation loaned $50 million to the Hospital through HPCH LLC, which owned the Hospital’s building and land. As part of this transaction, the Hospital promised to pay HPCH additional rent. HPCH gave Nomura a security interest in the incremental rent, which was to be transferred to MMA Funding.
The Daiwa line of credit lasted through March 2000; its termination caused insuperable cash-flow problems that led the Hospital to file for bankruptcy. The No-mura loan was securitized before the end of 1997. It was sold to a third party that packaged several billion dollars of commercial credit for resale to investors. The assets — borrowers’ notes and the security interests backing them up — were transferred to a trust, of which LaSalle National Bank is the trustee and Orix Capital Markets the servicer. (On the role of the servicer in securitization, see CWCapital Asset Management, LLC v. Chicago Properties, LLC, 610 F.3d 497 (7th Cir.2010).) Nomura was reimbursed and has no stake in the current dispute, except to the extent that it may be an investor in the pool. Because two trustees are opponents in this appeal — LaSalle Bank as trustee of the investment pool, and Gus Paloian as trustee of the Hospital’s estate in bankruptcy — we refer to each by name, while recognizing that each is a litigant only in a trustee’s capacity.
In multiple adversary proceedings, a bankruptcy judge concluded after a trial that the Hospital was insolvent no later than August 1997 and that the increased rental rate for the lease of the building and grounds was in reality debt service by the Hospital. See, e.g., United Airlines, Inc. v. HSBC Bank USA, N.A., 416 F.3d 609 (7th Cir.2005) (discussing the circum*691stances under which a lease may be re-characterized as a secured loan). The judge concluded that the repayments on the Nomura loan were fraudulent conveyances, which must be returned to the estate. The investment pool then would receive the same fraction of its claims as other creditors. Although 11 U.S.C. § 550(b)(1) prevents recovery by a bankruptcy estate when the transfer satisfies an antecedent debt, LaSalle Bank has not relied on this provision. Its briefs do not say why, and we do not speculate.
The conclusion about the Hospital’s insolvency led to only partial victory for trustee Paloian, however. The bankruptcy court concluded that as of July 1998, when the Hospital and its affiliates finally started using the precise cash-routing instructions in the loan agreements by sending the lease payments directly to LaSalle Bank, the payments were being made with MMA Funding’s assets rather than the Hospital’s and thus could not be avoided in the bankruptcy. The judge concluded that all payments from July 1998 forward are outside the bankruptcy. Paloian accepts this conclusion with respect to repayments on the Daiwa loan but not with respect to repayments on the Nomura loan.
The bankruptcy judge’s findings and conclusions appear at 360 B.R. 787 (Bankr. N.D.Ill.2007), additional findings made and reconsideration denied, 373 B.R. 53 (Bankr.N.D.Ill.2007). A district judge affirmed. 406 B.R. 299 (N.D.Ill.2009). Both the bankruptcy judge and the district judge issued several additional opinions, but these three are the only decisions that matter now. Paloian and LaSalle Bank have filed cross-appeals. A related opinion of this court appears at In re Doctors Hospital of Hyde Park, Inc., 474 F.3d 421 (7th Cir.2007) (holding that the bankruptcy and district judges did not abuse their discretion in approving a settlement under which Desnick and entities he controls paid $6 million in exchange for a release of further liability to the bankruptcy estate).
LaSalle Bank’s principal argument on appeal is that it is not an “initial transferee” of the funds, for the purpose of § 550(a)(1). Section 550(a) allows preferential transfers to be recouped from initial transferees, recipients from initial transferees, or “the entity for whose benefit such transfer was made”. Relying on Bonded Financial Services, Inc. v. European American Bank, 838 F.2d 890 (7th Cir.1988), LaSalle Bank contends that it was simply a conduit for placing the money in the trust. We held in Bonded Financial Services that, when a debtor’s check transfers money to a checking account, the “initial transferee” is the bank’s customer, who possesses control over the funds, rather than the bank that holds money subject to its customer’s orders. LaSalle Bank sees itself as agent of the pool’s investors and therefore as a similarly inappropriate target of a turnover order. And if, as it contends, the Bank is not the “initial transferee,” this whole proceeding is at an end.
The Bankruptcy Code does not define “initial transferee”; Bonded Financial Services adopted an approach that tracks the function of the bankruptcy trustee’s avoiding powers: to recoup money from the real recipient of preferential transfers. In Bonded Financial Services, that recipient was the bank’s customer, who had full control over the balance in the checking account. In this situation, the real recipient is LaSalle Bank, which is the trustee of the securities pool. In American law, a trustee is the legal owner of the trust’s assets. Wellpoint, Inc. v. CIR, 599 F.3d 641, 648 (7th Cir.2010); Restatement (Third) of Trusts § 2 comment d, § 42. Although LaSalle Bank has duties to the trust’s beneficiaries (the investors) concerning the application of *692funds, the assets’ owner remains the appropriate subject of a preference-avoidance action. If LaSalle Bank must hand $10 million over to the bankruptcy estate, it will draw that money from the corpus of the trust, not from the Bank’s corporate assets. This means that the money really comes from the trust’s investors — the persons “for whose benefit [the] transfer was made”. Instead of requiring the bankruptcy trustee to sue thousands of investors who may have received interest payments that were increased, slightly, by money from the Hospital’s coffers, a single suit suffices. If the Hospital had made a preferential transfer to Exxon, it would be appropriate to recover that transfer from Exxon rather than the millions of people who hold stock in Exxon. Similarly with a preferential transfer to a hedge fund. Instead of suing each investor, the bankruptcy estate could recover from the fund. Likewise, a preferential transfer to a trust is appropriately recovered from the trustee, who will charge it to the trust and thus create the appropriate economic incidence.
We cannot find any appellate decision on the question whether a trustee for a secu-ritized investment pool is an “initial transferee” under § 550(a). But lots of decisions hold that an entity that receives funds for use in paying down a loan, or passing money to investors in a pool, is an “initial transferee” even though the recipient is obliged by contract to apply the funds according to a formula. See, e.g., In re Columbia Data Products, Inc., 892 F.2d 26, 28 (4th Cir.1989); In re Chase & Sanborn Corp., 904 F.2d 588, 599-600 (11th Cir.1990); In re Incomnet, Inc., 463 F.3d 1064, 1071-76 (9th Cir.2006). All of these courts say that they are adopting and applying the approach that this circuit devised in Bonded Financial Services. We agree with that assessment and shall not create a conflict among the circuits on the question how to interpret one of our own opinions.
We have for decision four appeals that collectively present more than 20 issues. The appellate briefs approach 300 pages. The bankruptcy and district judges issued opinions thick enough to outweigh a dictionary. Because most issues have been satisfactorily resolved, in published opinions, at two levels of the federal judiciary, it is unnecessary to traipse through all of them again. Having disposed of the one issue (the “initial transferee” question) that might have ended the case, we address only two more. The first question is whether the Hospital was insolvent in August 1997. If not, the payments cannot be recovered under the trustee’s avoiding powers (though perhaps it became insolvent at some later time before filing for bankruptcy in April 2000). The second question is whether the transfer of the accounts receivable to MMA Funding was a true sale. If not, then MMA Funding did not serve as the bankruptcy-proofing intermediary that the lenders desired. In bypassing other questions, we do not necessarily approve the bankruptcy judge’s or district judge’s reasoning; we approve only the result. And some of the results (such as several of the questions about interest on the avoided transfers) are approved only because vital arguments have not been preserved for appellate decision. The subjects that this opinion pretermits are the law of the case, but not the law of the circuit. (This is the approach that Fed. R.App. P. 32.1 and Circuit Rule 32.1 adopt for nonprecedential orders. We have elected not to discuss, in this prece-dential opinion, issues that may be significant to the parties but would not contribute to the stock of precedents.)
Until after it filed for bankruptcy, the Hospital was current in paying its *693creditors and had consistently positive financial statements and EBITDA (earnings before interest, taxes, depreciation, and amortization). Yet the bankruptcy judge concluded that it had been insolvent for almost three years. How was that possible?
Trustee Paloian presented testimony from accountants who calculated solvency using several methods; the one the bankruptcy judge adopted is a diseounted-cash-flow analysis. The analyst projects a firm’s net cash flows into the future, then discounts the stream of income to present value. If the present value of the income plus the firm’s physical assets exceeds the firm’s obligations, it is solvent. Discounted-cash-flow analysis is sensitive to assumptions about the discount rate and expected future income, but these need not detain us. The discounted-cash-flow analysis showed that the Hospital was solvent in August 1997 — indeed, comfortably solvent. The bankruptcy judge then subtracted $18.5 million, the amount that federal audits later determined the Medicare and Medicaid systems had overpaid. The rationale for this subtraction is that by mid-1997 investigations were under way, and a chargeback was inevitable. But this adjustment was not enough to turn the Hospital’s bottom line red. What did the trick was lopping 40% off the portion of the Hospital’s assets that represented the present value of future income. The rationale for this adjustment was that the Hospital was a Subchapter S corporation that did not pay taxes, and a taxable buyer would reduce the purchase price in recognition of the fact that its profits must be shared with state and federal treasuries.
LaSalle Bank observes that the bankruptcy judge made a mistake when determining the reserve (a contingent liability) for reimbursing Medicare and Medicaid: the judge did not determine the expected value of the payment. Instead the judge used hindsight and added $18.5 million to the liability side of the balance sheet. Hindsight is wonderfully clear, but in determining the Hospital’s solvency in mid-1997 it was necessary to determine the expected value of this liability as of mid-1997, not the actual value as of 1999 or 2000. Hindsight bias is to be fought rather than embraced. See Boyer v. Crown Stock Distribution, Inc., 587 F.3d 787, 794-95 (7th Cir.2009). Paloian contends that LaSalle Bank’s lawyers did not make this point with sufficient clarity and therefore have forfeited the issue on appeal. We need not determine whether the Bank forfeited its argument about the right way to value contingent liabilities, because there is an even more glaring error: contingent liabilities must be matched against contingent assets.
The missing asset was Desnick’s wealth. If Desnick caused the Hospital to submit excessive claims for federal payments, then he and the Hospital were jointly and severally liable for restitution. So if $18.5 million goes on the liability side of the Hospital’s balance sheet, the asset side must contain an estimate (as of mid-1997) of how much Desnick would chip in, either directly or via contribution or indemnity (for, if the Hospital paid, then it would have a claim against Desnick). As it happened, Desnick paid the whole $18.5 million out of his own resources. Paloian contends that none of the money that Des-niek supplied belonged on the asset side of the balance sheet as of mid-1997, because it was speculative how much he could or would pay. Yet Desnick’s personal wealth is not pie in the sky; it is the sort of thing that banks would loan money against (and did). If Desnick had made out a note, in the Hospital’s favor, for $18.5 million, a court would not ignore it when toting up the Hospital’s assets. The judge would discount the note to reflect the probability *694that it could be collected. The discount might be substantial, but the court would not value the note at zero. Yet that’s what the bankruptcy court did: it valued contingent liabilities at 100<t on the dollar and contingent assets at 0$ on the dollar. The treatment must be symmetrical. (So too with hindsight: If a court uses hindsight to value the liability at $18.5 million, it must use hindsight to value Desnick’s share at $18.5 million, for a net zero effect on the Hospital’s balance sheet.)
The 40% reduction likewise was a mistake. Such a reduction has two potential bases: the illiquidity of the Hospital’s shares, and the fact that some potential buyers would have to pay taxes that the Hospital itself did not (because the profits of a Subchapter S corporation are taxable only as income to its shareholders). Neither of these has anything to do with a corporation’s solvency. They concern the market value of its securities, not the state of its balance sheet. Cf. Gross v. CIR, 272 F.3d 333, 351-56 (6th Cir.2001).
Take tax effects. A Subchapter S corporation does not pay taxes. Some potential buyers do. But how much a buyer will pay for a revenue stream does not tell us whether a firm is insolvent, except indirectly: If a buyer will pay a positive price for the firm’s stock, then it is very likely to be solvent. (“Very likely” rather than “certain” because stock has an option value. Even after a firm is in bankruptcy, its stock will sell for a small price, reflecting the probability that the firm will be reorganized and old equity investors be given some stake in the reorganized firm.) And questions of income tax arise only when a firm is profitable, because the income tax is imposed on net rather than gross receipts. So whether an outsider would have paid $50 million or only $30 million (a 40% discount) for all stock in the Hospital, either price implies that the Hospital’s assets are worth more than its liabilities. On top of all this, most potential buyers for hospitals are themselves nonprofit organizations that do not pay income tax. These organizations, competing against one another to buy either stock or assets in a hypothetical sale, would not reduce their bids on account of income taxes.
Much the same can be said about an illiquidity discount. Suppose a closely held corporation has $1 million in profits annually, 1,000 shares of stock, and thus $1,000 in annual profits per share, but does not pay dividends. How much is the share worth? If the stock were traded in a liquid market, the investor could get something close to the present value of the income stream (plus the value of any nondepreciating assets). If the discount rate is 5%, and the corporation is expected to stay in business for 20 years, the value per share would be around $12,500. When the firm is closely held, though, buyers may not be available. What’s more, many closely held firms forbid the sale of stock to outsiders. The owner of a share then might accept $10,000 or even $8,000, rather than the amount implied by a discounted-cash-flow analysis. So when shares are appraised — for purposes of the estate tax, valuation when shareholders object to a merger, or repurchase by the issuer under a buy/sell clause — the appraisal usually is less than the hypothetical price in the (nonexistent) liquid market. But this has nothing to do with whether the firm is solvent. The thing being adjusted is the anticipated selling price per share, not the asset side of a balance sheet.
We asked counsel at oral argument whether they knew of any decision, from any court, that applied either an illiquidity discount or a tax-effect discount to the asset side of a corporation’s balance sheet for the purpose of figuring out whether the firm is solvent. Neither side’s lawyers *695knew of any example, other than the decision under review. Our own research did not turn one up. It follows that the Hospital was solvent in August 1997 and that the ensuing months’ debt service (which is how the bankruptcy judge understood the above-market rent payments) cannot be recaptured as a fraudulent conveyance.
Unfortunately, the bankruptcy judge did not determine whether the Hospital was insolvent at some later time between August 1997 and the bankruptcy petition in April 2000. The bankruptcy judge will need to address that subject, if trustee Paloian believes that payments made during some shorter period are within the scope of the avoiding powers in bankruptcy. And if further proceedings ensue, one issue that is sure to recur is whether Dai-wa and Desnick succeeded in making MMA Funding a bankruptcy-remote vehicle. For, if they did not, and the Hospital went insolvent before April 2000, then payments routed through or for the account of MMA Funding potentially could be recaptured for the benefit of creditors in general (if § 550(b) does not foreclose relief).
The idea behind the bankruptcy-remote vehicle is that, if a debtor sells particular assets to a separate corporation, the lender can rely on those assets without the complications (such as preference-recovery actions) that attend bankruptcy. See, e.g., Kenneth N. Klee & Brendt C. Butler, Asset-Backed, Securitization, Special Purpose Vehicles and Other Securitization Issues, ALI-ABA Course of Study Materials SJ082 (June 2004). Bankruptcy-remote entities are among several devices that borrowers and lenders have adopted to make corporate reorganization more a matter of contract and less a matter of judicial discretion. See Alan Schwartz, A Contract Theory Approach to Business Bankruptcy, 107 Yale L.J. 1807 (1998); Robert K. Rasmussen, Debtor’s Choice: A Menu Approach to Corporate Bankruptcy, 71 Tex. L.Rev. 51 (1992). See also Douglas G. Baird & Robert K. Rasmussen, The End of Bankruptcy, 55 Stan. L.Rev. 751 (2002).
To make the idea work, the separate entity must be, well, separate. It must buy assets (here, accounts receivable). It must manage these assets in its own interest rather than the debtor’s. It must observe corporate formalities, to prevent the court from rolling it back into the debtor under the approach of a decision such as United Airlines, which holds that debtors and creditors can’t evade bankruptcy law through clever choice of words, but must structure their transactions so that their economic substance lies outside particular sections of the Bankruptcy Code.
If Daiwa had loaned $25 million to Vehicle, a corporation independent of Desniek, which then purchased the Hospital’s accounts receivable for $22 million (using the proceeds of the loan) and stood to make a profit, or suffer a loss, depending on how much eventually came in, there would be little ground to treat Vehicle’s payments on the loan as preferential transfers by the Hospital. The transfer by the Hospital would have occurred with the initial sale of the receivables, and, if that sale predated the Hospital’s insolvency (or the bankruptcy filing) by enough time, it would be outside the bankruptcy trustee’s avoiding powers, even though particular payments occurred within the look-back periods (90 days or one year under 11 U.S.C. § 547(b), two years under § 548(a)). And the parties agree that, if MMA Funding became a legitimate bankruptcy-remote vehicle as part of the Daiwa loan, this prevents recovery of payments made on the Nomura loan from July 1998 forward.
As far as we can tell from this record, however, MMA Funding lacked the usual *696attributes of a bankruptcy-remote vehicle. It was not independent of Desnick or the Hospital; Desnick owned MMA Funding (99% of which was owned by the Hospital, and 1% of which was owned by a firm that Desnick owned directly or through some trusts), and MMA Funding operated as if it were a department of the Hospital. It did not have an office, a phone number, a checking account, or stationery; all of its letters were written on the Hospital’s stationery. It did not prepare financial statements or file tax returns. It did not purchase the receivables for any price (at least, if it did, the record does not show what that price was). Instead of buying the receivables at the outset, MMA Funding took a small cut of the proceeds every month to cover its (tiny) costs of operation. The Hospital continued to carry the accounts receivable on its own books, as a corporate asset; it told other creditors that Daiwa had a security interest in the receivables, which is of course the sort of structure that makes the payments amenable to a preference-recovery action whether or not the receivables are remitted to a lockbox at a bank.
There is scarcely any evidence in this record that MMA Funding even existed, except as a name that Daiwa’s and Desnick’s lawyers put in some documents. Daiwa can’t complain; it knew that MMA Funding was a shell or could have found out easily enough. See Fusion Capital Fund II, LLC v. Ham, No. 09-3723, 614 F.3d 698, 2010 WL 2990817 (7th Cir. Aug. 2, 2010). But a trustee in bankruptcy can step into the shoes of any hypothetical lien creditor, see 11 U.S.C. § 544 — which for current purposes may mean a creditor ignorant of the contracts signed by the Hospital, Daiwa, Nomura, LaSalle Bank, and MMA Funding. If a hypothetical creditor could have obtained an interest in assets that the Hospital’s books declared belonged to it, then a bankruptcy trustee can maintain an avoidance action. And the interests of these outside creditors can’t be ignored. The bankruptcy and district judges observed that treating MMA Funding as a bankruptcy-remote vehicle allowed Daiwa and Nomura to charge lower rates of interest — which is true enough but overlooks the fact that, if some creditors are protected from preference-recovery actions and thus can charge lower interest, other creditors bear higher risk and must charge higher interest. The net effect for operating firms is unclear. See Barry E. Adler, A Re-Examination of Near-Bankruptcy Investment Incentives, 62 U. Chi. L.Rev. 575 (1995). The Code allows the trustee to look out for the interests of these other creditors, who may not appreciate that they should have charged extra to offset the effects of a bankruptcy-remote vehicle that was hidden in the weeds.
Perhaps LaSalle Bank can offer on remand evidence to show that there was a bona fide sale of accounts receivable from the Hospital to MMA Funding in March 2007, and that MMA Funding was more than a name without a business entity to go with it. Or perhaps the Bank could contend that the hypothetical lien creditor must be charged with knowledge of those aspects of the earlier transactions that were matters of public record. See In re Professional Investment Properties, 955 F.2d 623, 627-28 (9th Cir.1992); Collier on Bankruptcy ¶ 544.02[2]. But the first task on remand will be to determine whether the Hospital was insolvent at any time before filing for bankruptcy. Unless it was, nothing else matters.
The judgment of the district court is vacated, and the case is remanded with instructions to remand to the bankruptcy court for proceedings consistent with this opinion.