Opinion for the Court filed by Circuit Judge STEPHEN F. WILLIAMS.

STEPHEN F. WILLIAMS, Circuit Judge:

In January 1990 AlliedSignal, a company manufacturing aerospace and automotive products, decided to sell its interest in Union Texas Petroleum Holdings, Inc., an oil, gas, and petrochemical company. Anticipating a large capital gain, it sought to reduce the resulting tax burden by entering into a set of transactions via a partnership with several foreign corporations. The transactions took advantage of provisions of the Internal Revenue Code (and related regulations) designed to yield reasonable results when property is sold on an installment basis and the value of the installment payments cannot be known in advance. With the help of these provisions, transactions that in substance added up to a wash were transmuted into ones generating tax losses of several hundred million dollars; the offsetting gains were allocated to foreign entities not subject to United States income tax at all.

The Commissioner of Internal Revenue in 1996 issued a notice of final partnership administrative adjustment, reallocating to AlliedSignal much of the capital gain accrued by the partnership. ASA, via its “Tax Matters Partner,” AlliedSignal, petitioned for relief in the Tax Court, which agreed with the Commissioner that Allied-Signal had not entered into a bona fide partnership and upheld the Commissioner’s determination. ASA Investerings Partnership, AlliedSignal, Inc., Tax Matters Partner v. Commissioner, 76 T.C.M. (CCH) 325 (1998) (“Tax Court Decision”). We affirm.

* * *

The hardest aspect of this case is simply getting a handle on the facts. To make them manageable, we first discuss the principal tax provisions in question and then show their application through a series of examples, ending with a simplified version of the transactions here. Only then do we lay out the exact transactions themselves.

Under the general provisions of the Internal Revenue Code (“IRC”), gains and losses are generally “realized” in the year that they are received or incurred. See 26 U.S.C. § 1001 (1994). A sale for future payments, however, presents several difficulties, among them that the deferred payment may be contingent in amount or otherwise not susceptible to accurate valuation. Section 453 of the Internal Revenue Code, 26 U.S.C. § 453, provides methods for taxation of such an “installment sale,” defined as “a disposition of property where at least 1 payment is to be received after the close of the taxable year in which the disposition occurs.” Id. § 453(b)(1). It specifies the “installment method” for such a sale, providing that “the income recognized for any taxable year from a disposition is that proportion of the payments received in that year which the gross profit ... bears to the total contract price.” Id. § 453(c). Thus, if A owns a building with a basis of $100, and sells it for $300 to be paid in five $60 annual installments, A recognizes a taxable gain of $40 each year. The proportion of “gross profit” to “total contract price” is 2 %o or two thirds, so the income recognized for each year is two thirds of the receipts of that year.

In 1980 Congress expanded § 453, authorizing the Secretary to make the install- *507 merit method available to deferred payment transactions for which the sales price is indefinite, or subject to a contingency. Section 453(j) (previously § 453(i)) mandates that the Secretary shall promulgate regulations “providing for ratable basis recovery in transactions where the gross profit or the total contract price (or both) cannot be readily ascertained.” In response, the Treasury promulgated Temp. Treas. Reg. § 15A.453-l(c)(3)(i) (1981), which provides that in contingent payment sales (and subject to irrelevant exceptions), “the taxpayer’s basis ... shall be allocated to the taxable years in which payment may be received under the agreement in equal annual increments.”

Under these regulations, the taxpayer will have a recognized gain in years when payment from the sale exceeds the basis recovered; in years when payment is less than the basis recovered, “no loss shall be allowed unless the taxable year is the final payment year under the agreement or unless it is otherwise determined ... that the future payment obligation under the agreement has become worthless.” Id.

The following examples should illustrate the ratable basis recovery rule. Property owner, A, sells a house with a basis and current value of $1 million in exchange for an instrument that will pay unpredictable amounts (e.g., a share of the property’s gross profits) over a five-year period. In any year in which the payout equals or exceeds $200,000, A will recover $200,000 in basis under the ratable basis recovery rule and will have a taxable gain equal to the difference between the amount received from the note and $200,000. In a year in which the payout is less than $200,-000, A will not report a loss, but instead will recover a portion of the basis equal to the payout; the unused basis will then be carried forward to the next year. See id. § 15a.453-l(c)(3), example (2). Under the rule just quoted above, unused basis would be recovered only in the last year of scheduled payout.

The rule works similarly when the seller receives both an instrument with indefinite value and an immediate payment of cash. In a variation on the preceding case, for example, suppose A sells the property for a $500,000 cash payment and an indefinite five-year instrument. Once again, the basis is recovered over the course of five years. In the first year, A recovers $200,-000 in basis; because he has received $500,000 that year, he must report a gain of $300,000. If A sells the note in Year 2 for $500,000, he can report a loss of $300,-000, equal to the difference between the remaining basis in the note ($800,000) and the $500,000 he has received in exchange for the note. 1 In this example, of course, the results are rather unappealing to the taxpayer: although he had no real gain, he recognized a nominal one early, offset by an equal tax loss — but one that was deferred and therefore not a complete offset. Because of the rule against any recovery of basis in excess of gross receipts in any year except the last, the taxpayer cannot manipulate the timing in his favor.

But suppose A finds a way of allocating the nominal tax gain to a tax-free entity, reserving for himself a nominal tax loss? Here is how he might do it: He forms a partnership with a foreign entity not subject to U.S. tax, supplying the partnership with $100,000 and inducing the “partner” to supply $900,000. The “partnership” buys for $1,000,000 property eligible for installment sale treatment under § 453, and, as the ink is drying on the purchase documents, sells the property, as in the last example, for $500,000 in cash and an indefinite five-year debt instrument. The cash payment produces a gain of $300,000, 90% of which goes to the nontaxable foreign entity. Then ownership adjustments are made so that A owns 90% of the partnership. In Year 2 the instrument is *508 sold, yielding a tax loss of $300,000, 90% of which is allocable to A. Presto: A has generated a tax loss of $240,000 ($270,000 loss in Year 2, offset by $30,000 gain in Year 1), with no material change in his financial position — other than receipt of the valuable tax loss. This example is AlliedSignal’s case, stripped to its essentials.

Now for the specifics of this case: In 1990, AlliedSignal anticipated that it would soon realize a capital gain of over $400 million from the sale of its interest in Union Texas Petroleum Holdings, Inc. In February, AlliedSignal approached Merrill Lynch & Co. to discuss a set of transactions that Merrill had developed to create tax losses to shelter anticipated capital gains.

Under the plan AlliedSignal would form a partnership with a foreign entity, which in turn would supply the majority of the capital for and assume a majority stake in the partnership. In Year 1, the partnership would purchase short-term private placement notes (“PPNs”), which can be sold under the installment method of accounting provided for in IRC § 453. See 26 U.S.C. § 453(k)(2)(A) (implying that the sale of “stock or securities which are [not] traded on an established securities market” would be subject to the installment method). Several weeks later, the partnership would sell the instruments for 80% cash and 20% debt instruments, which would pay out over several years and thus would be subject to the ratable basis recovery rule. As in the example above, the partnership would report a large gain in the first year, equal to the difference between the substantial cash payment and the small share of basis recovered that year. The gain would be allocated according to each partner’s interest, with the tax-exempt foreign partner receiving the lion’s share.

The next year, AlliedSignal would acquire a majority interest in the partnership, and sell the debt instruments. The sale would create a large tax loss because the basis available for recovery would far exceed the value of the instruments.

Merrill would serve as the partnership’s financial adviser and, for a $7 million fee, would recruit the foreign partner and arrange for the subsequent investments. Tax Court Decision, 76 T.C.M. at 326. Merrill would also “structure and enter into the requisite swap transactions” with the banks, “[t]o ensure a market for [the] issuance and sale” of the PPNs and the debt instruments to be received on their sale. Id. In exchange for serving as the partnership’s financial intermediary, Merrill would receive roughly $1 to 2 million on the initial sale of PPNs, and $200,000 to $400,000 on the sale of the subsequent debt instruments. Id. The foreign partner would receive the greater of $2,850,-000 or 75 basis points (1 basis point = .01%) over the London International Bank Offering Rate (“LIBOR”) on any funds contributed to the partnership, as well as reimbursement of all partnership expenses incurred. Id.

AlliedSignal and Merrill followed the proposal to the letter. Merrill contacted Algemene Bank Netherlands N.V. (“ABN”), one of the Netherlands’ largest commercial banks. Id. at 327. ABN had previously participated in similar Merrill transactions, and anticipated that this partnership would strengthen its preexisting lending relationship with AlliedSignal. Id. On April 5, 1990 Johannes den Baas, Vice President of Corporate Finance for ABN New York, an ABN affiliate, requested authorization to enter into this venture. Id. Den Baas recommended the creation of two “special purpose corporations” (“SPCs”), to which ABN would lend $990 million, and which would then contribute this money to the venture. Id. The purpose of this structure, den Baas said, was (1) to permit ABN, which would otherwise be a general partner in the venture with AlliedSignal, to limit its exposure to liability, and (2) to facilitate ABN’s shifting part of the loan to other banks for various reasons.

*509 On April 17 and 18, den Baas and another representative of an ABN affiliate met in Bermuda with a representative of Al-liedSignal. Id. Both sides agreed that AlliedSignal would pay all partnership expenses, as well as ABN’s costs funding the requisite loans to the partnership (approximately LIBOR), plus 75 basis points. Id. at 328. 2 The precise amount, of course, would depend on the amount that ABN contributed and how long the partnership, to be known as ASA Investerings, held those funds. In response to den Baas’s request that AlliedSignal pay $5 million up-front, the parties agreed that AlliedSig-nal would instead periodically make “premium” payments upon the occurrence of certain events. Id. The agreements reached during these negotiations were referred to by the Tax Court as the “Bermuda Agreement.”

ABN’s Risk Management Division expressed concern that a loss might arise out of the sale of the PPNs. Id. at 327. Den Baas responded by assuring these officials that any such loss would be added to the value of the subsequent debt instruments, and would in turn be borne by AlliedSignal on liquidation of those instruments. But there could be no written agreement to this effect, explained den Baas, because “in that case it would not be a matter of a general partnership.” Id.; Memo from den Baas to Jos Albers, 4/22/90, Joint Appendix (“J.A.”) 676. Den Baas nonetheless assured the authorities at ABN of his confidence that AlliedSignal would bear any such loss:

[T]he fact that the client is the only one who is interested in such a sale and wants to obtain the installment note, whereby the SPCs have a right of veto during the entire procedure — as is, in fact, set forth in the partnership document — makes ABN N.Y. Corporate Finance more than confident that the client will continue to have this loss charged to his account in the future as well.

Id. The loan-approving committees later authorized ABN’s participation.

On April 19 ASA Investerings Partnership was formed. Tax Court Decision, 76 T.C.M. at 328. It consisted of AlliedSig-nal, AlliedSignal Investment Corporation (“ASIC”), a wholly-owned subsidiary of Al-liedSignal, and the two SPCs, Barber Corp. N.V. and Domniguito Corp. N.V., which were controlled by foundations in turn controlled by ABN. Barber and Dominguito entered into revolving credit agreements with ABN. The foundations which owned the SPCs granted ABN an irrevocable option to buy shares of the respective SPCs at par value. Id.

On April 19th and 24th, the partners contributed a total of $850 million, with each partner receiving a partnership interest in accordance with its contribution. AlliedSignal’s share was 10%, the SPCs’ 90%. In May 1990, the parties supplemented their contributions (in the same ratio), bringing the total contribution to $1.1 billion.

On April 25, 1990 ASA purchased from two Japanese banks $850 million of 5-year floating rate notes which were not traded on an established securities market — the planned PPNs. Id. at 329. On May 8, the partnership met in Bermuda and decided to sell the PPNs for 80% cash and 20% LIBOR notes. Id. Between May 17 and May 24 (i.e., within one month of purchase), ASA sold the PPNs to two banks in exchange for a little under $700 million in cash and 11 notes. (With the cash it bought time deposits and 30-day commercial paper.) These notes each made 20 quarterly payments consisting of the 3- *510 month LIBOR rate, calculated at the beginning of each payment period, applied to 25% of the notional principal amount, which in this case was $434,749,000. Id. The LIBOR notes did not require a return of principal at maturity; rather, the quarterly payments of interest on the “notional amount” can be seen as both interest and return of actual principal. ABN entered into swap transactions with Merrill, Barber and Dominguito to hedge ABN’s interest rate risk relating to the LIBOR notes. Merrill also entered into swap transactions with the banks from whom it bought the LIBOR notes (as it had with the PPNs) to induce their participation. Merrill’s transaction costs in selling the PPNs were $6,375 million. Id. This was added to the value of the LIBOR notes in ASA’s partnership books, so that AlliedSignal would bear the costs of the sale when it acquired the LIBOR notes from the partnership. Id. at 329-330.

Because the LIBOR notes provided for 20 quarterly payments at a variable rate, and the PPNs were not sold on an established securities market, the sale of the PPNs qualified for use of the installment method. For the taxable year ending May 31, 1990, ASA recovered jé of the basis in the PPNs (because the completion of the scheduled payout would occur in the sixth tax year after the sale, and reported $549,-443,361 in capital gains (= $681,300,000 in cash minus %($851,139,836))). 3 Id. at 331. The gain was allocated according to partnership interest, 90% to the SPCs and 10% to AlliedSignal and ASIC. Id.

On August 2, 1990, AlliedSignal issued $435 million in commercial paper, and with the proceeds purchased Barber’s entire interest in ASA for about $400 million and a 13.43% interest in ASA from Dominguito for about $150 million. Between November 1991 and April 1992, ASA further reduced Dominguito’s interest to 25%. AlliedSig-nal also paid a $4.4 million “premium” payment, representing a portion of the $5 million requested by den Baas. After this purchase, AlliedSignal and ASIC entered into several swaps with Merrill to hedge their interest in the LIBOR notes. Id. at 330. Between August 31 and September 28 AlliedSignal borrowed $435 million from ASA, using the proceeds to repay the debt incurred August 2; this indebtedness was paid off May 1,1992.

On August 21 the partnership authorized a distribution of the LIBOR notes to AlliedSignal and ASIC, and about $116 million in cash and commercial paper to Dominguito. Id. at 331. During 1990, AlliedSignal and ASIC sold a fraction of the LIBOR notes with a basis of $246,520,-807, for a total of $50,454,103, and reported a capital loss equal to the difference, $196,-066,704. Id. That year, AlliedSignal also reported a capital gain of $53,926,336, its share of ASA’s capital gain from the sale of the PPNs. Id.

On December 5, 1991, AlliedSignal made a direct payment of $1,631,250 to Domin-guito, representing: the difference between ABN’s funding costs and the SPCs’ combined income allocations; interest on $92 million of ABN’s funds that remained in ASA beyond the agreed upon date; and the difference (plus interest) between the $5 million up-front payment that ABN had requested and the $4.4 million it had previously paid. Id.

Prior to liquidation, ABN and AlliedSig-nal agreed that ABN would refund $315,-000, reflecting excesses of the SPCs’ income allocations over funding costs, and of ASA’s returns from November 22, 1991 through April 30, 1992 over LIBOR. Id. at 332. On June 1, 1992, the partners liquidated ASA. On November 31, Allied-Signal sold its remaining LIBOR notes for $33,431,000, and recovered the remainder of the basis, $429,655,738, taking a capital loss of $396,234,738. Id.

In a notice of final partnership administrative adjustment the Commissioner dis *511 allowed ASA’s capital gain, reallocating it to AlliedSignal and ASIC and thus in substance cancelling out the tax losses otherwise enjoyed by AlliedSignal. Id. at 333. The Commissioner relied on alternative grounds: first, ASA was not a bona fide partnership, and Barber and Dominguito were not correctly deemed partners; and second, the transactions lacked “economic substance.” Id.

* *

The Tax Court agreed with the Commissioner that ASA was not a valid partnership for tax purposes, and thus did not reach the economic substance argument. At the outset, the court disregarded Barber and Dominguito “[f]or purposes of [its] analysis,” Tax Court Decision, 76 T.C.M. at 333, finding that they were merely ABN’s agents. First, they were thinly capitalized, and created just for purposes of the venture. Second, the parties themselves treated ABN as the partner, disregarding Barber and Dominguito. Third, Barber and Dominguito were “mere conduits,” to whom ABN lent funds and in whom ABN owned options to purchase shares for a de minimis amount.

Having found ABN to be the relevant party, the Tax Court turned its attention to the question whether AlliedSignal and ABN entered into a bona fide partnership. Although the Internal Revenue Code provides that “the term ‘partnership’ includes a syndicate, group, pool, joint venture, or other unincorporated organization through or by means of which any business, financial operation, or venture is carried on,” 26 U.S.C. § 761, the court set out to determine whether the formal partnership had substance, quoting the Supreme Court’s language in Commissioner v. Culbertson, 337 U.S. 733, 742, 69 S.Ct. 1210, 93 L.Ed. 1659 (1949), which said that the existence of the partnership (for tax purposes) would depend on whether, “considering all the facts ... the parties in good faith and acting with a business purpose intended to join together in the present conduct of the enterprise.” Applying this test, the Tax Court concluded that AlliedSignal and ABN did not have “the requisite intent to join together for the purpose of carrying on a partnership.” 76 T.C.M. at 335.

* * *

We review decisions of the Tax Court “in the same manner and to the same extent as decisions of the district courts in civil actions tried without a jury.” 26 U.S.C. § 7482. Factual findings are reviewed for clear error, see Commissioner v. Duberstein, 363 U.S. 278, 291, 80 S.Ct. 1190, 4 L.Ed.2d 1218 (1960), and determinations of law de novo. See United States v. Popa, 187 F.3d 672, 674 (D.C.Cir.1999). We have held that in tax cases mixed questions of law and fact are to be treated like questions of fact. See Fund for the Study of Economic Growth and Tax Reform v. IRS, 161 F.3d 755, 759 (D.C.Cir.1998) (citing Duberstein, 363 U.S. at 289 n. 11, 80 S.Ct. 1190). Petitioner poses challenges to all three types of decisions constituting the Tax Court judgment.

Much of petitioner’s opening brief is directed to an attack on the Tax Court’s reasoning. We confess that some of that reasoning seems misdirected. For example, the court seemed to believe that because ABN and AlliedSignal had “divergent business goals,” 76 T.C.M. at 333, they were precluded from having the requisite intent to form a partnership. We agree with petitioner that partners need not have a common motive. In fact, the desirability of joining complementary interests in a single enterprise is surely a major reason for creating partnerships. But we see no reason to think that this view, mentioned only in the opinion’s initial summary and concluding paragraph, was essential to the court’s conclusion.

Some of petitioner’s argument seems an exercise in semantic ju jitsu. It argues that we may not consider whether the partnership was a “sham” because the Tax Court (a) explicitly refused to consider that, and (b) never used the word “sham.” The first point is false, the second irrele *512 vant. Although the Tax Court said that it would not consider whether the transactions at issue lacked “economic substance,” id,., its decision rejecting the bona fides of the partnership was the equivalent of a finding that it was, for tax purposes, a “sham.”

Getting to the controlling issue, petitioner argues that under the standard established in Moline Properties v. Commissioner, 319 U.S. 436, 63 S.Ct. 1132, 87 L.Ed. 1499 (1943), the partnership cannot be regarded as a sham. The Court there said that a corporation remains a separate taxable entity for tax purposes “so long as [its] purpose is the equivalent of business activity or is followed by the carrying on of business by the corporation.” 319 U.S. at 439, 63 S.Ct. 1132. The Tax Court has since applied Moline to partnership cases. See Bertoli v. Commissioner, 103 T.C. 501, 511-12,1994 WL 579942 (1994).

Petitioner views Moline as establishing a two-part test, under which a tax entity is accepted as real if either. (1) its purpose is “the equivalent of business activity” (not tax avoidance), or (2) it conducts business activities. Moline, 319 U.S. at 439, 63 S.Ct. 1132. Because ASA “engaged in more than sufficient business activity to be respected as a genuine entity,” petitioner argues that ASA was a partnership under the second alternative. Petitioner’s Reply Br. at 12. We agree if engaging in business activity were sufficient to validate a partnership ASA would qualify. It was infused with a substantial amount in capital ($1.1 billion), and invested it in PPNs, LIBOR notes, and other short-term notes over a period of two years. In fact, however, courts have understood the “business activity” reference in Moline to exclude activity whose sole purpose is tax avoidance. This reading treats “sham entity” cases the same way the law treats “sham transaction” cases, in which the existence of formal business activity is a given but the inquiry turns on the existence of a nontax business motive. See Knetsch v. United States, 364 U.S. 361, 364-66, 81 S.Ct. 132, 5 L.Ed.2d 128 (1960). Thus, what the petitioner alleges to be a two-pronged inquiry is in fact a unitary test — whether the “sham” be in the entity or the transaction — under which the absence of a nontax business purpose is fatal. 4

Shortly after Moline the Second Circuit held per Judge Learned Hand in National Investors Corp. v. Hoey, 144 F.2d 466 (2d Cir.1944), that the retention and sale of securities, after the date when the corporate holding had served its nontax goals, could not be considered for tax purposes. There an investment trust corporation proposed to merge with its subsidiaries. To this end it created a new corporation into which it transferred its interests in the subsidiaries in exchange for 10 shares of stock. But the stockholders decided to reject the plan to unify the four corporations. The investment trust then liquidated the new corporation, starting with an exchange of 10% of the new corporation’s shares and taking a deduction based on the difference between the value of 10% of the shares when originally issued to the trust, and 10% of their reduced value a year later. In initiating even this 10% liquidation, however, it delayed for some time after the shareholders’ vote. The court held that the trust was entitled to a deduction only for value decreases incurred from the time of the original transfer of assets to the corporation to “a reasonable time” after the stockholders rejected the plan. Id. at 468. Thereafter, “there was no longer any business for [the corporation] to do.” Id. Retention of the corporation merely for the purpose of tax minimization was not enough. The court explicitly read the *513 cases as saying that “the term ‘corporation’ will be interpreted to mean a corporation which does some ‘business’ in the ordinary meaning, and that escaping taxation is not ‘business’ in the ordinary meaning.” Id. So too, for ASA: Although its investment in LIBOR notes might have had a business purpose, the prior three-week investment in and subsequent sale of PPNs was, like the retention of assets in Hoey, a business activity merely conducted for tax purposes. 5 Moreover, as discussed later, AlliedSig-nal’s interest in any potential gain from the partnership’s investments was in its view at all times dwarfed by its interest in the tax benefit.

The Ninth Circuit has similarly held that “business activity” is inadequate in the absence of a nontax business purpose. In Zmuda v. Commissioner, 731 F.2d 1417 (9th Cir.1984), the taxpayers argued that the Tax Court incorrectly applied the “economic substance” rule rather than the “business purpose” rule. The court found that the taxpayer’s argument “attempts to create a distinction where none exists. There is no real difference between the business purpose and the economic substance rules. Both simply state that the Commissioner may look beyond the form of an action to discover its substance.” Id. at 1420. The court went on to say:

The terminology of one rule may appear in the context of the other because they share the same rationale. Both rules elevate the substance of an action over its form. Although the taxpayer may structure a transaction so that it satisfies the formal requirements of the Internal Revenue Code, the Commissioner may deny legal effect to a transaction if its sole purpose is to evade taxation.

Id. at 1421. At issue was the validity of certain trusts, so the court’s equation of the “transaction” test and the “entity” test was clearly a holding.

We note that the “business purpose” doctrine is hazardous. It is uniformly recognized that taxpayers are entitled to structure their transactions in such a way as to minimize tax. When the business purpose doctrine is violated, such structuring is deemed to have gotten out of hand, to have been carried to such extreme lengths that the business purpose is no more than a facade. But there is no absolutely clear line between the two. Yet the doctrine seems essential. A tax system of rather high rates gives a multitude of clever individuals in the private sector powerful incentives to game the system. Even the smartest drafters of legislation and regulation cannot be expected to anticipate every device. The business purpose doctrine reduces the incentive to engage in such essentially wasteful activity, and in addition helps achieve reasonable equity among taxpayers who are similarly situated — in every respect except for differing investments in tax avoidance.

Thus the Tax Court was, we think, sound in its basic inquiry, trying to decide whether, all facts considered, the parties intended to join together as partners to conduct business activity for a purpose other than tax avoidance. Its focus was primarily on ABN, curiously. As we shall discuss later, the absence of a nontax business purpose was even clearer for Al-liedSignal. Nonetheless, given ABN’s protection from risk, and even from the borrowing costs of providing its capital contribution, there was no clear error in the finding that its participation was formal rather than substantive. 6 Petitioner *514 alleges two primary ways in which the Tax Court erred in this regard.

First, petitioner says that the Tax Court incorrectly found that Barber and Domin-guito were mere agents of ABN rather than partners in their own right. But this issue of classification makes no material difference. Once the court decided that the SPCs were mere conduits for ABN, it shifted its focus to whether AlliedSignal and ABN (rather than the SPCs) formed a bona fide partnership. There was certainly no clear error in the court’s view of the SPCs as being within the complete control of ABN, and there is no indication as to how the SPCs’ intent as to the “partnership” differed from that of ABN.

Second, petitioner argues that the Tax Court erred by finding that ABN did not share in profits and losses because it received a specified return from AlliedSignal and hedged against risk through swap transactions with Merrill. On the profit side, we find no clear error in the court’s findings that the direct payments made to ABN were to compensate it merely for its funding costs. Petitioner , says that there was no explicit agreement that ABN would receive a return of LIBOR plus 75 basis points, pointing to the fact that the payments actually made by AlliedSignal to ABN did not produce such a return. See Petitioner’s Brief at 59; supra note 2. But petitioner makes no argument that these payments were related to the success of the partnership’s investments (i.e., the PPNs and LIBOR notes), and under the circumstances it is reasonable to infer. that they were made pursuant to a prearranged agreement to compensate ABN for its funding costs (plus some amount above LIBOR, even if not 75 basis points).

Den Baas’s testimony, moreover, confirms that ABN could make no profit from the transaction: any potential profit from the LIBOR notes would be offset by losses from the concomitant swap transactions. Petitioner cites Hunt v. Commissioner, 59 T.C.M. (CCH) 635 (1990), for the proposition that a guaranteed return is not inconsistent with partnership status. In Hunt, however, both parties had a bona fide business purpose for entering into the partnership, and unlike in the case at hand, the guaranteed return created a floor but not a ceiling. See id. at 648.

Turning to the risk of loss, we agree with the Tax Court that any risks inherent in ABN’s investment were de minimis. As a preliminary matter, the court did not err by carving out an exception for de minimis risks, as no investment is entirely without risk. Unless one posited a particular asset (such as the dollar) as the sole standard, its value could change in relation to the values of other assets, and treating one— even the dollar — as the sole' standard would be arbitrary. The Tax Court’s decision not to consider ABN’s “de minimis” risk is also consistent with the Supreme Court’s view, albeit in the “sham transaction” context, that a transaction will be disregarded if it did “not appreciably affect [taxpayer’s] beneficial interest except to reduce his tax.” Knetsch, 364 U.S. at 366, 81 S.Ct. 132 (emphasis added) (quoting Gilbert v. Commissioner, 248 F.2d 399, 411 (2d Cir.1957) (Hand, C.J., dissenting)).

There was no clear error in the Tax Court’s determination that at no point during the transaction did ABN assume greater than de minimis risk. The PPNs were essentially risk free: not only were they issued by banks with the highest credit ratings but they were held for a mere three weeks. Moreover, because of the side agreement under which any loss on the PPNs would be embedded in the value of the LIBOR notes, AlliedSignal would bear any shortfall over the brief period in which PPNs were held. Petitioner argues that ABN was subject to the risk that AlliedSignal would ultimately decide not to acquire the LIBOR notes. This seems *515 unlikely to the point of triviality, however, for two reasons: first, this step was integral to AlliedSignal’s tax objective, and to the entire transaction; and second, at the point when the LIBOR notes would be distributed, Dominguito still owned over 40% of ASA, and according to the partnership agreement, any act of the partnership committee would require the consent of partners whose interest equaled or exceeded 95%. Nor was there any real hazard that AlliedSignal might agree to distribute the LIBOR notes but refuse to make the adjustment for any loss on the PPNs: as den Baas had stated in a memorandum to ABN officials, the SPCs could counter by refusing consent for the distribution of the notes altogether.

The LIBOR notes certainly had greater inherent risk than the short-term PPNs; ABN, however, entered into hedge transactions outside the partnership to reduce the risk to a de minimis amount. In fact, the correlation between the swaps and the LIBOR notes was 99.999%, i.e., the company succeeded in hedging all but a de minimis amount of the risk associated with the LIBOR notes. Petitioner concedes that “the hedges provided the ABN parties with substantial protection from fluctuations in LIBOR Note value due to movements in interest rates.” Petitioner’s Reply Br. at 18. But ASA nevertheless contends that ABN still bore the credit risk associated with the issuers of the LI-BOR notes and with Merrill, which provided the hedges. Once again, we And that the Tax Court correctly found the risk to be de minimis: the LIBOR notes were issued by banks having a credit rating of AAA and AA+, and were held by ASA for only three months. So too, the risk associated with the swaps with Merrill (a single-A rated institution) was de minimis. Finally, there was little, if any, risk associated with the commercial paper that ASA held at this point, which although un-hedged, was found by the Tax Court to be “AAA-rated, short-term, and from multiple issuers.” Tax Court Decision, 76 T.C.M. at 335.

Petitioner argues that ABN’s side-transactions should not be considered in deciding whether a bona fide partnership was formed. It draws an analogy to a partnership which owns an uninsured building later destroyed by fire; this partnership is bona ñde even if one partner had insured against his portion of the loss. The analogy, though imaginative, is not very telling. The insurance in the hypothetical is comparatively narrow, leaving a considerable range of potential business hazards and opportunities for profit. Contrast den Baas’s observation at the outset of the present plan: “Credit risk: The structure demands that virtually no credit risk will be taken in the partnership since any defaults on the principal of the investments will jeopardize the objective... Market interest rate risk: ABN New york [sic] will take care of perfect hedges in order to protect the bank from the changes in the value of the underlying securities.... due to interest rate fluctuations.” J.A. 680-81. We note, moreover, that petitioner directs us to no evidence that ABN even bore the cost of these hedges. Given that Merrill, which had orchestrated the entire transaction and to whom AlliedSignal had paid a substantial sum, engaged in the swap transactions, it is likely that AlliedSignal assumed the costs of the swaps. A partner whose risks are all insured at the expense of another partner hardly fits within the traditional notion of partnership.

Petitioner argues that the Tax Court also erred in determining that ABN’s capital contribution constituted a loan. We need not pass on this question because it is quite peripheral to the central issue of whether the parties entered into a bona fide partnership.

We noted earlier that the Tax Court’s focus on ABN’s intentions was a little puzzling. AlliedSignal, after all, was the driving force and AlliedSignal focused on tax minimization to the virtual exclusion of ordinary business goals. Of course no one *516 wants to pay more than necessary, even for a very profitable tax minimization scheme, and the petitioner argues that even with the very substantial transaction costs associated with the partnership, Al-liedSignal had grounds for expecting that it would come out more than $15 million ahead. As it proved, transaction costs were almost $25 million rather than the roughly $12 million it anticipated, Tax Court Decision, 76 T.C.M at 326, 332, so the ultimate financial gain, according to the parties, was actually about $3.6 million. The expected gain turned on the belief of Roger Matthews, AlliedSignal’s assistant treasurer — which proved correct — that interest rates would shift in such a way that, when all the swaps were taken into account, AlliedSignal would benefit. But this evidence says nothing about AlliedSig-nal’s use of the elaborate partnership— with a pair of partners concocted for the occasion. There is no reason to believe that AlliedSignal could not have realized Matthews’s interest rate play without the partnership at far, far lower transactions costs. For the deal overall, the most telling evidence is the testimony of AlliedSig-nal’s Chairman and CEO, who could not “recall any talk or any estimates of how much profit [the transaction] would generate.” The Tax Court concluded that none of the supposed partners had the intent to form a real partnership, a conclusion that undoubtedly encompasses AlliedSignal. And the record amply supports that finding.

The decision of the Tax Court is

Affirmed.

Notes

. The regulation does not appear to provide expressly that in the event that the taxpayer completely liquidates the instrument before the end of scheduled payout he may recover all unused basis in that year. Both parties agree, however, that this is the case.

. Petitioner acknowledges that ABN might have "had in mind a target return of LIBOR plus 75bp on the amount it invested,” Petitioner’s Initial Br. at 59, but argues that the Tax Court erred in finding that the parties actually entered into such an agreement. Petitioner points to the fact that AlliedSignal's payments were negotiated rather than specified in advance, and were not based on a LIBOR plus 75 basis points return. This dispute is, however, immaterial. See infra p. 514.

. The parties stipulated in the Tax Court that ASA had erred in calculating the gain on its 1990 tax return and that the correct amount was $539,364,656.

. Indeed, one might logically enough place the Tax Court’s findings here under the “sham transaction” heading, viewing the formation of the partnership as the transaction. Because of the ultimate unity of the tests, however, there is no need to address this formulation.

. The PPNs cost $850 million. When an expert was later engaged by AlliedSignal to evaluate the partnership's gains and losses, AlliedSignal asked that he assign to the LI-BOR notes a value which, together with the cash, would bring the total value of the proceeds of the PPNs to $850 million. See J.A. 1343. AlliedSignal evidently did not believe that the initial investment in PPNs increased the return from the transactions in the aggregate.

. Although petitioner argues that ABN's purpose was not tax-avoidance, but rather "included a desire to enhance its business *514 relationship with AlliedSignal,” Petitioner's Initial Br. at 41 n.19, the desire to aid another party in tax avoidance is no more a business purpose than actually engaging in tax avoidance.