Full opinion text
ANDERSON, Circuit Judge: The SEC brought this enforcement action against defendants Steven Wyer, Kurt Beasley, and Merchant Capital, LLC (“Merchant”), alleging violations of the registration and antifraud provisions of the federal securities laws. Wyer and Beasley, through Merchant, sold interests in twenty-eight registered limited liability partnerships (“RLLPs”) to 485 persons. The SEC asserted that these interests were “investment contracts” within the meaning of the federal securities laws, and that the defendants had committed securities fraud in marketing the interests. The district court concluded that the interests were not investment contracts and, regardless, that the defendants had not committed securities fraud. We reverse in part, vacate in part, and remand for further proceedings. I. Facts Wyer and Beasley formed Merchant in order to participate in the business of buying, collecting, and reselling charged-off consumer debt from financial institutions such as banks and credit card companies. Because this case depends to a significant extent on the nature of this industry, we will set out its characteristics in some detail. When a consumer is delinquent on a credit account, the company that provided the account begins by trying to collect the debt itself. After 180 days, however, the company normally sells the debt to a wholesale purchaser. Most sales of debt occur in large pools. Some of these pools are sold at auction. Others are sold pursuant to long-term contracts with large purchasers, so-called “forward-flow contracts.” Forward-flow contracts are attractive to the seller because they provide a consistent and reliable way to get rid of debt. They are attractive to the wholesaler because they provide a reliable supply and also typically guarantee that the accounts are a representative sample of the company’s debt pool and that none of the debtors are deceased or bankrupt. After auction and forward-flow contracts dispose of the large pools, a credit card company or bank may have small amounts of debt left over. It will generally sell the remaining debt in small pools in what are known as “one-off’ sales. These sales may occur in parcels as small as $25,000, and have none of the guarantees that are present in the long-term forward-flow contracts. While some wholesalers attempt to collect the debt themselves, most purchase debt and then outsource the collection of that debt to attorneys and collection agencies. For these wholesalers, the key business decision is determining the price at which they are willing to buy particular pools of debt from the issuer. As established at trial, the appropriate price for a pool of debt depends on the many factors that determine how likely it is that the debt in the pool will be collected. These factors include (1) who the issuer is, (2) how hard the issuer tries to collect the debt before selling it, (3) how the issuer selects the accounts it sells from the general pool of debtor accounts, (4) whether the selection produces a fair representation of the issuer’s general debt profile, (5) the geographic distribution of the accounts, (6) the laws of the states where the accounts are located, (7) the terms and conditions of the particular debtor accounts, (8) how old the accounts are, (9) how recently a payment was made on an account, (10) how many payments were made on an account, and (11) whether a payment was ever made on an account at all. Wholesalers often have proprietary computer programs that assign a weight to the various factors, assess the characteristics of the particular debt pool, and thereby estimate how valuable the pool is. The wholesaler makes its profit by buying pools that seem profitable; outsourcing the collection for a certain period of time; and then reselling the debt on the secondary market for an even lower price than it paid for the debt. Wyer and Beasley had no prior experience in the debt purchasing industry. Wyer was formerly a principal in a securities firm. His most recent business was conducting direct marketing for financial institutions. Wyer had declared personal bankruptcy because that business defaulted on certain obligations that he had personally guaranteed. Beasley was a lawyer and CPA whose practice focused on banking, asset protection, and corporate representation. He had previously done legal work for Wyer. Wyer became interested in the debt purchasing business through conversations with a participant in the industry, Fred Howard. Howard was principal owner of New Vision Financial (“New Vision”), a wholesale debt purchaser. Wyer sought to enter a business where his personal relationships with financial institutions would be valuable, and through discussions with Howard, he believed debt purchasing was such a business. Howard had previously raised funds by selling RLLP interests, but had abandoned that model because a number of states were investigating whether the RLLP interests were securities under state law. Howard provided Wyer with business models and legal opinions on the status of the RLLP interests under the securities laws. He also informed Wyer and Beasley of the existence of the state securities investigations. Wyer and Beasley formed Merchant, with Wyer owning seventy-five percent, and Beasley twenty-five percent. Wyer and Beasley planned to raise funds through Merchant and then buy fractional shares in debt pools ultimately purchased by New Vision. New Vision would aggregate money from Merchant and other sources, purchase debt pools through auction and forward-flow contracts, and then outsource the collection of the debt to Enhanced Asset Management (“EAM”), a collection company. To formalize this relationship, Merchant entered into a services contract with New Vision. Merchant began raising money in November 2001 by soliciting members of the general public to become partners in Colorado Registered Limited Liability Partnerships. Merchant employed a network of recruiters to sell the RLLP interests, and provided the recruiters with scripts. These recruiters informed potential partners that, while they would be expected to participate in the operation of their partnership, their actual duties would be limited to checking a box on ballots that would be periodically sent to them. Merchant’s offering materials also represented that the RLLP interests were not securities and that the federal securities laws did not apply to the interests. Each RLLP was to have no more than 20 partners. Merchant eventually organized twenty-eight RLLPs, containing 485 partners, with a total capitalization of over $26 million. The eventual RLLP partners were all members of the general public with no demonstrated expertise in the debt purchasing business, and included a nurse, a housewife, and a railroad retiree. Each partner had a net worth of at least $250,000, and more than seventy-five percent had a net worth exceeding $500,000. Further, ninety percent of the partners self-reported business experience between “average” and “excellent.” Two-thirds of the investors invested through their IRA accounts. The partnerships were marketed and sold as freestanding entities. Merchant did not disclose the existence of the other partnerships or the relationship with New Vision to the RLLP partners. Despite the partnerships’ formal independence, Wyer testified at trial that Merchant planned from the beginning to pool the money collected from the RLLPs in order to purchase fractional interests in debt pools owned by New Vision. Under the partnership agreement provided by Merchant, each partnership was scheduled to participate in the debt purchasing, collection, and resale business for three years, at which point it would be dissolved. The managing general partner (“MGP”) would be the operational head of the partnership, with sole authority to bind the partnership. At the end of the three years, the partnership assets would be distributed to the partners and the MGP. The partnership agreement also provided that partners could elect to receive returns on capital equal to 3.6 percent of their contribution per quarter for three years, or else a 16.5 percent annual return on capital payable at the end of the three years. The MGP would collect fees on each transaction in debt. In addition, the MGP would participate in any profits over and above the 14.4 or 16.5 percent annual return. Partners would receive fifty percent of such profits, and the MGP would receive the other half. The partnership materials told the partners that they were expected to have an active role in managing the business of their partnership. The agreement reserved certain powers for the partners. They had the ability to select the MGP. They had the exclusive right to approve any act obligating the partnership in an amount exceeding $5,000. They could remove the MGP upon a unanimous vote, for cause. Finally, they could inspect books and records and participate in various committee meetings. Merchant prepared all the partnership materials and was the sole business contact for all of the partners. It was the only candidate for MGP and was named on one-hundred percent of the ballots. Merchant thus became MGP for each of the twenty-eight RLLPs. In practice, the partners exercised little control over the operations of their partnerships. Merchant, as MGP, had sole authority to bind the partnerships and made the key business decisions, such as whether and when to purchase the shares in New Vision debt pools. Merchant submitted ballots to the partners, but those ballots contained only the name of the issuer, the face value of the pool, and the price per dollar of debt. Moreover, until the SEC began its investigation, the ballots had only a signature line. If a ballot was not cast, the partnership agreement provided that it was voted in favor of management’s proposal. In addition, Merchant frequently disregarded the balloting process. The SEC’s expert surveyed the voting records of five partnerships, and his unrebutted testimony showed that Merchant purchased more debt than the partners authorized thirty times; purchased debt before the ballots had been sent six times; and purchased debt before the ten-day return period had expired seventy-three times. Merchant also failed to operate the business successfully. As early as June 2002, Merchant and its principals knew that the existing partnerships were performing poorly. The benchmark for performance goals was the capital contribution of each partnership plus the 14.4% or 16.5% elected return on capital. By June 2002, the partnerships were performing at 89% of benchmark. By nine months, the partnerships were at 77%; at fifteen months, 62%. Some partnerships were doing far worse. An email revealed that the principals knew in June 2002 that RLLP-5 was collecting at 55% of benchmark. Merchant nevertheless continued to sell the partnerships, with the same partnership materials, through November 2002. Despite the poor performance, no partnership replaced Merchant during the period before the SEC began its investigation. One barrier to replacement was the fact that the partnerships did not own their own debt pools. Rather, Merchant had invested partnership funds in fractional interests of debt pools purchased and owned by New Vision. After the SEC’s investigation began, Merchant devised a method to make it easier for partnerships to repossess their debt pools. Together with EAM and New Vision, Merchant developed a process, called “n-select,” by which a share in a fractionalized debt pool could be converted into whole debtor accounts. The n-select process selected a random assortment of accounts that came as close as possible to the partnership’s share in the debt pool. The difference was then reconciled with cash. N-select made it theoretically possible for a partnership to receive an identifiable set of debtor accounts to approximate its capital contribution, and thus operate as an independent business. RLLP-19 took advantage of the n-select process to remove Merchant as MGP in August 2004, near the end of its three-year life, and received possession of its debtor accounts from Merchant. By that point, the relationship with New Vision had ended. Merchant had replaced New Vision with a company called Trilogy Capital Management in January 2003, and then with Merchant Management, a company owned by Merchant, in November 2003. In June 2004, two months before RLLP-19 withdrew, Merchant informed the partnerships that liquidation was preferable to continuing as a going concern. At the time RLLP-19 withdrew, Merchant Management was in possession of the pools of debt. II. Proceedings in the district court and scope of review The SEC brought this action on November 4, 2002. Merchant consented to a temporary restraining order that precluded it from selling further RLLP interests. Merchant had already received a cease and desist order from the State of California in October 2002, enjoining the sale of unregistered securities and precluding it from selling the interests in that state. The SEC alleged violations of the registration and antifraud provisions of the federal securities laws, and sought injunctive relief, disgorgement, and penalties. The district court conducted an evidentiary hearing on the SEC’s application for a preliminary injunction, and denied the application on May 1, 2003. After more discovery, a bench trial took place in January 2005. In November 2005, the district court denied the SEC’s application for in-junctive relief, disgorgement, and civil penalties, and entered judgment in favor of the defendants. It concluded that the RLLP interests were not investment contracts and therefore not securities. Alternatively, the district court concluded that the defendants had not committed securities fraud. On appeal, the SEC challenges both of these determinations. The meaning of “investment contract” is a question of law reviewed de novo. SEC v. Unique Fin. Concepts, Inc., 196 F.3d 1195, 1198 (11th Cir.1999). We review for clear error the factual findings underlying the district court’s determination that the RLLP interests were not investment contracts. See Mitchell v. Hillsborough County, 468 F.3d 1276, 1282 (11th Cir.2006). We also review the district court’s findings on materiality and scienter (which underlie the securities fraud determination) for clear error. Lucas v. Fla. Power & Light Co., 765 F.2d 1039, 1040-41 (11th Cir.1985). A finding will be held clearly erroneous only when “the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” United States v. U.S. Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948). III. Status of RLLP interests under the federal securities laws The key issue in this case is whether the RLLP interests marketed by Merchant were “investment contracts” covered by the federal securities laws. Both the Securities Act and the Exchange Act define “securities” to include “investment contracts.” 15 U.S.C. §§ 77b(a)(1), 78c(a)(10). An investment contract is “a contract, transaction, or scheme whereby a person invests his money in a common enterprise and is led to expect profits solely from the efforts of the promoter or a third party.” SEC v. W.J. Howey Co., 328 U.S. 293, 298-99, 66 S.Ct. 1100, 1103, 90 L.Ed. 1244 (1946). The issue in this case is whether the RLLP partners were led to expect their profits solely from the efforts of Merchant. Under this prong of Howey, “solely” is not interpreted restrictively. “The Supreme Court has repeatedly emphasized that economic reality is to govern over form and that the definitions of the various types of securities should not hinge on exact and literal tests.” Williamson v. Tucker, 645 F.2d 404, 418 (5th Cir. May 20, 1981). An interest thus does not fall outside the definition of investment contract merely because the purchaser has some nominal involvement with the operation of the business. Rather, “the focus is on the dependency of the investor on the entrepreneurial or managerial skills of a promoter or other party.” Gordon v. Terry, 684 F.2d 736, 741 (11th Cir.1982). A general partnership interest is presumed not to be an investment contract because a general partner typically takes an active part in managing the business and therefore does not rely solely on the efforts of others. Williamson, 645 F.2d at 422. But consistent with the substance over form principle of Howey, “[a] scheme which sells investments to inexperienced and unknowledgeable members of the general public cannot escape the reach of the securities laws merely by labeling itself a general partnership or joint venture.” Williamson, 645 F.2d at 423. A general partnership interest may qualify as an investment contract if the general partner in fact retains little ability to control the profitability of the investment. Williamson recognized three situations where this would be the case: (1) “[A]n agreement among the parties leaves so little power in the hands of the partner or venturer that the arrangement in fact distributes power as would a limited partnership,” id. at 424; (2) “[T]he partner or venturer is so inexperienced and unknowledgeable in business affairs that he is incapable of intelligently exercising his partnership or venture powers,” id.; or (3)“[T]he partner or venturer is so dependent on some unique entrepreneurial or managerial ability of the promoter or manager that he cannot replace the manager of the enterprise or otherwise exercise meaningful partnership or venture powers,” id. Under Williamson, the presence of any one of these factors renders a general partnership interest an investment contract. Id. The three factors also are not exhaustive. Id. at 424 n. 15. Williamson is ultimately simply a guide to determining whether the partners expected to depend solely on the efforts of others, thus satisfying the Howey test. The SEC argues that the defendants should not receive the benefit of the Williamson presumption against investment contract status because the RLLP interests are more akin to limited partnership interests, which are routinely treated as investment contracts. See Williamson, 645 F.2d at 423. It is true that an RLLP bears some similarity to a limited partnership. An RLLP partner is liable only for the amount of his or her capital contribution, plus the partner’s personal acts, and is not exposed to vicarious liability for the acts of other partners or the acts of the partnership as a whole. See Colo.Rev. Stat. § 7-60-115(2)(a), (b). This limitation on liability means that RLLP partners have less of an incentive to preserve control than general partners do. While general partners normally wish to preserve control because their personal assets are at risk, RLLP partners have only their investment at risk if they remain passive, and risk personal liability only if they become active. On the other hand, it is not invariably true that partners in an RLLP, limited liability company (LLC), or limited liability partnership (LLP) lack the ability to control the profitability of their investments. The powers of partners or members in these forms of business can be altered by agreement, and may assume virtually any shape, despite the limitation on liability. As these business forms represent a hybrid between general and limited partnerships, it is unsurprising that courts, even in jurisdictions that apply Williamson to general partnership interests, have reached mixed results concerning whether the Williamson presumption against investment contract status applies to RLLP, LLP, and LLC interests. See, e.g., Robinson v. Glynn, 349 F.3d 166, 174 (4th Cir.2003); SEC v. Shiner, 268 F.Supp.2d 1333, 1340 (S.D.Fla.2003); Keith v. Black Diamond Advisors, Inc., 48 F.Supp.2d 326, 333 (S.D.N.Y.1999). It is clear in this circuit, however, that an RLLP interest is an investment contract if one of the Williamson factors is present. That is because the powers available in an RLLP cannot exceed the powers available in a general partnership. If anything, an RLLP is somewhat more likely to be an investment contract because of the incentive against exercising control that is produced by the limited liability shield. Therefore, because the presence of one of the three Williamson factors renders even a general partnership interest an investment contract, a fortiori the presence of one such factor would render an RLLP interest an investment contract. Because we need not decide the general applicability of the Williamson presumption to limited liability interests if one of the Williamson factors is present, we begin by analyzing those. In addition to defining the conditions under which a general partnership interest may qualify as an investment contract, Williamson also defines the scope of the investment contract analysis. We analyze the expectations of control at the time the interest is sold, rather than at some later time after the expectations of control have developed or evolved. Williamson, 645 F.2d at 424 n. 14. A post-sale delegation cannot, for example, convert a general partnership into an investment contract, if the partners had control at the beginning. Id. As an evidentiary matter, however, we may look at how the RLLPs actually operated to answer the question of how control was allocated at the outset. See Albanese v. Fla. Nat’l Bank, 823 F.2d 408, 412 (11th Cir.1987) (looking to “reality” of partners’ control over placement of ice machines as evidence of amount of control present at inception); Rivanna Trawlers Unltd. v. Thompson Trawlers, Inc., 840 F.2d 236, 242 (4th Cir.1988) (noting, as evidence of control at inception, that the managers were in fact replaced on two later occasions). Williamson also defines the kind of evidence that is to be considered in determining the expectations of control. Consistent with Howey’s focus on substance over form, we look at all the representations made by the promoter in marketing the interests, not just at the legal agreements underlying the sale of the interest. SEC v. C.M. Joiner Leasing Corp., 320 U.S. 344, 353, 64 S.Ct. 120, 124, 88 L.Ed. 88 (1943) (“In the enforcement of an act such as this it is not inappropriate that promoters’ offerings be judged as being what they were represented to be.”); Gordon, 684 F.2d at 742 (“Williamson requires an examination of the representations and promises made by promoters ... to induce reliance upon their entrepreneurial abilities.”); Koch v. Hankins, 928 F.2d 1471, 1478 (9th Cir.1991). The ultimate issue under Howey is whether the partners expected to rely solely on the efforts of others, and we may rely on the totality of the circumstances surrounding the offering in making this determination. A. Did the arrangement in fact distribute power as would a limited partnership? The first Williamson factor requires us to analyze whether “an agreement among the parties leaves so little power in the hands of the partner or ven-turer that the arrangement in fact distributes power as would a limited partnership.” Williamson, 645 F.2d at 424. In arguing that the partners did not function as limited partners, Merchant relies primarily on the allegedly substantial powers reserved to the partners through the partnership agreement. The partnership materials informed partners that they were expected to take an active role in the business, and the agreement gave partners certain rights and powers. Partners had the ability to call meetings and hold regular quarterly meetings; the ability to participate in committees; the ability to elect the MGP; the ability to remove the MGP for cause upon a certain vote; the ability to inspect books and records; the ability to approve additional funding; the ability to amend the agreement or to dissolve the partnership upon a two-thirds vote; and the exclusive authority to approve obligations exceeding $5000. In the first place, the power to name the MGP was not a significant one in this case. Partners were required to turn in their ballots with their capital contribution, before their partnerships had even been formed. The power therefore reveals nothing about the partners’ ability to control the business after their initial investment. Moreover, Merchant was the only option for MGP. The investors had no independent experience in the debt purchasing industry and no way of knowing about alternative MGPs. And, as a result, Merchant was named on all ballots. This power was not significant. The partners also did not have the practical ability to remove Merchant once it was installed as MGP. First, the agreement provided for removal only for cause. This meant that Merchant could not be removed readily, and even in the case of gross incompetence, the partners would have had to litigate any unconsented removal. We have previously found that where removal is only for cause, and the investors have no other ability to impact management, the interest is an investment contract as a matter of law. Albanese, 828 F.2d at 411. Here there were further barriers to removing Merchant. The partnership agreement required a unanimous vote of the partners. The district court found that Merchant could be removed somewhat more easily, with a two-thirds vote of the partners in an individual partnership. Even if a two-thirds vote were enough, we would likely conclude that removal was practically impossible, when combined with the other factors. But we find that the district court clearly erred. The written materials are somewhat ambiguous: the partnership agreement seems to say that a two-thirds vote is enough, while the partnership application says that a unanimous vote is required. Wyer admitted at trial, however, that he told partners a unanimous vote is required. Williamson makes clear that its test applies to the representations made by promoters, not to the strict legal terms of the partnership agreement. See Williamson, 645 F.2d at 424 n. 14. That approach is necessary because the ultimate test under Howey is whether a person “is led to expect profits solely from the efforts of the promoter or a third party.” Howey, 328 U.S. at 299, 66 S.Ct. at 1103 (emphasis added). For the purposes of Williamson, then, we hold Wyer’s representation against him and conclude that a unanimous vote was required. And it is clear that such a vote, combined with removal only for cause and the factors discussed below, rendered Merchant effectively unremovable. Cf. Gordon, 684 F.2d at 741 (in the absence of other Williamson factors, partners in standard joint venture who control by majority vote do not hold investment contracts); Holden v. Hagopian, 978 F.2d 1115, 1120 (9th Cir.1992) (no investment contract where manager could be removed with simple majority vote). Compounding the legal difficulty in removing Merchant, the investors in an individual partnership were geographically dispersed, with no preexisting relationships. Howey found it significant that the interests in that ease were offered “to persons who reside in distant localities.” Howey, 328 U.S. at 299, 66 S.Ct. at 1103. Similarly, in this case, the lack of face-to-face contact among the partners exacerbated the other difficulties and rendered the supposed power to remove Merchant illusory. The district court, in finding that the partners in an individual partnership could remove Merchant, relied in part on the fact that RLLP-19 did succeed in removing Merchant as MGP. That event does not shake our confidence that, at the time of the original investment, the partners did not have the practical ability to remove Merchant. By the time RLLP-19 replaced Merchant, Merchant had no interest in opposing removal. It had already recommended liquidation of the RLLPs, which meant that there would be no more debt transactions and thus no further fees for Merchant. Merchant also at that time had an active interest in encouraging removal; the SEC investigation was in progress, and Merchant’s defense hinged upon showing that the partners were in control. The removal of Merchant at that stage therefore shows nothing about whether partners could have overcome the cause, unanimity, and dispersion barriers and removed Merchant as manager of a viable going concern. We conclude that the partners in fact lacked the power to remove Merchant. The next power reserved to the partners was the ability to approve all obligations over $5,000. If this power was real, it was a substantial one. The primary business of each partnership was purchasing frac-tionalized interests in pools of debt that generally exceeded $5,000 in value. However, as shown by Merchant’s tenure as MGP, the ballot right also did not give the partners meaningful control over their investment. First, Merchant controlled how much information appeared in the ballots, and did not submit sufficient information for the partners to be able to make meaningful decisions to approve or disapprove debt purchases. Each ballot indicated only the face value of the debt pool, the price per dollar of debt, and the name of the issuer. This information at most gave the partner the price of the pool and the name of the issuer. But unrebutted testimony at trial established that much more data is necessary to make an informed decision about how much a debt pool is really worth. The SEC’s expert testified that many factors go into the price that should be paid for a debt pool, in addition to the name of the issuer: the reputation of the issuer, how the accounts were selected from the issuer’s pool, the geographic distribution, the terms of the cardholder agreements, how old the debt is, when the last payment was, and whether a payment was ever made, among many others. A former associate of Wyer’s, Mitch Bonilla, who also operates a debt purchasing company, testified to the same effect. Finally, the defendants’ own witness, Fred Howard, testified that when he buys debt, he considers the last payment date, when the credit card was issued, the history of the account, and the location of the account. The only testimony supporting the sufficiency of the ballot information came from Wyer, who testified that the information contained in the ballots was enough to allow a partner to make an informed decision. But Wyer, besides being an interested witness, was not an authority on the debt purchasing business. He admitted that New Vision had sole responsibility for purchasing debt pools. New Vision only passed on the name of the issuer and the price of the pool to Merchant. Wyer therefore had limited experience in purchasing debt pools, and his testimony does not call into question the convincing testimony of the industry participants, including that of Howard, head of New Vision. Wyer’s testimony is also contradicted by the partnership application itself. In an informational section, the application informed partners that “[a]mong the many criteria that may be analyzed before a pool of debt is purchased are the following: the number of times the debt has been placed, the state of residency of the customer, and when the debt was incurred. These factors and others should be taken into account in the RLLP’s analysis of the potential future recovery of a debt portfolio.” The ballots sent to partners contained none of this information, and therefore did not permit partners to make an informed decision about debt purchases. The defendants nevertheless argue (and the district court found) that the ballot gave partners control over the business because Merchant passed on all the information it received to the partners. That does not, however, establish that the ballot process was meaningful. It may have made business sense for Merchant to outsource the debt purchase decisions to New Vision. New Vision certainly had more experience in the business than any of Merchant’s principals or any RLLP partners did. But Merchant, having delegated the power to make debt buying decisions, cannot claim that the partners also possessed that power. Because New Vision was in possession of all the relevant information for making debt buying decisions, it and not the partners was responsible for deciding what debt to purchase. The partners had no information with which to make a meaningful decision. Second, besides the fact that the ballots were completely devoid of meaningful information, the partners had no way to force Merchant to heed the results of the process. The SEC’s expert testified, based on a sample of balloting data for five partnerships, that Merchant repeatedly abused the balloting process. Merchant purchased more debt than the ballots authorized thirty times; purchased debt before ballots were sent six times; and purchased before the ten-day return period expired seventy-three times. In part because the partners had no ability to remove Merchant, they also lacked the power to force Merchant to abide by the results of the ballots. Finally, the voting process was tilted in Merchant’s favor from the very start. The partnership agreement provided that unre-turned and unvoted ballots were voted in favor of management. Until October 2002, the ballots had only a signature line, and no “no” box. Even after that, the ballots always contained insufficient information for investors to make informed decisions. The levers of the voting process were thus in Merchant’s hands from the very beginning. It had control over the information in the ballots, did not have any incentive to heed the results of ballots, and was assured of victory in any balloting anyway because of the default rules. Unsurprisingly, no ballot ever went against Merchant’s decisions. We therefore conclude that the voting process was a sham and did not give partners meaningful control over their investment. Merchant insists that we are restricted to the terms of the partnership agreement in applying the first Williamson factor, and argues that because the approval authority was included in the terms, the district court had no choice but to conclude that it was meaningful. Merchant relies on Rivanna Trawlers Unltd. v. Thompson Trawlers, Inc., 840 F.2d 236 (4th Cir.1988). That case followed Williamson in concluding that the relevant scope of the investment contract analysis is the expectation at the time of the agreement, rather than post-investment developments that may constitute a later delegation of power. See id. at 240-41 (“[T]he mere choice by a partner to remain passive is not sufficient to create a security interest.”). Merchant argues that the way the ballot process turned out in practice is no indication of how much power the partners retained at the inception: the partners might simply have decided that Merchant was doing a good job, and had no desire to rectify the problems with the balloting. Our analysis is, however, fully consistent with Williamson. It is true that we are limited to assessing the expectations of control at the inception of the investment. Williamson, 645 F.2d at 424 n. 14. But we are not limited to the terms of the partnership agreement in assessing those expectations of control. Post-investment events can serve as evidence of how much power partners reserved at the inception. See Albanese, 823 F.2d at 412 (11th Cir.1987) (looking to “reality” of partners’ control over placement of ice machines as evidence of amount of control present at inception); Rivanna Trawlers, 840 F.2d at 242 (using actual post-investment replacement of managers as evidence of degree of control at inception). It is difficult to imagine how a court could determine how much power the partners “in fact” retained under the agreement without looking to some extent at post-investment events. A focus on the bare terms of the legal agreement would also be inconsistent with the substance over form principle of Howey, and would be an invitation to artful manipulation of business forms to avoid investment contract status. See Williamson, 645 F.2d at 418. Here, the operation of the balloting process revealed that the partners from the very beginning lacked power to control Merchant’s actions. The balloting process was a sham because the partnership agreement contained no way to force Merchant to conduct a meaningful ballot. The agreement did not require a particular form of ballot, which meant that Merchant could control the information partners received. And the partners had no other influence over Merchant because they lacked a practical removal power. They therefore could not force Merchant to give them a meaningful amount of information or to respect the results of the ballots. Rivanna Trawlers presented an entirely different case. It dealt with a simple joint venture arrangement ruled by majority vote where, “[significantly, on two separate occasions the external managers were replaced” and on one occasion “one of the promoters ... was removed as managing partner ... and replaced with a management committee of partners.” Rivanna Trawlers, 840 F.2d at 242. The general partners in that case always had the power to remove the manager, had substantial control over the business, and made a decision to temporarily delegate some duties to a manager. Here the RLLP partners lacked the ability to remove or control the manager from the inception of the enterprise. Nor did the RLLP partners ever expect to take an active role in managing the business, as the recruiters represented that the partners’ duties would be limited to checking a box on the ballots that were periodically sent to them. The remaining powers — the ability to inspect books and records, participate in committees, and hold meetings — did not on them own give the partners the potential to control Merchant’s management of the business, and thus do not contribute to the Williamson analysis. We have previously held that “the opportunity to inspect ... records” and an “obligation] to give periodic accountings” of income and expenses did not bear on an investor’s ability to control the profitability of the investment, and were therefore irrelevant from the perspective of investment contract analysis. Albanese, 823 F.2d at 411 n. 4. Here, the inspection and meeting rights might have helped the partners make more informed decisions, if the partners also had the ability to effectuate those decisions. But because the appointment, approval, and removal powers were illusory, these other powers had no independent salience. The limited extent of the partners’ powers makes this case similar to Albanese v. Fla. Nat'l Bank, 823 F.2d 408 (11th Cir.1987). There the promoter sold ice machines and service contracts to individual investors. Id. at 411. Even though some of the contracts allowed the investors to specify the location of the ice machine or refuse to consent to the ice machine being moved, we concluded that the instruments were investment contracts. Id. at 412. We found that the ability to affect the placement of the ice machines was too insubstantial to preclude investment contract status because the investors still depended on the promoter for most of the crucial business tasks, including “finding the locations, contracting with the hotels and other institutions, servicing the machines, and accounting for the profits.” Id. In the case of the RLLPs, the partners’ control was even less substantial. There was no analogue to the Albanese partners’ limited ability to control the location of the ice machines. Merchant retained responsibility for every business decision of consequence; could not be removed; and was not subject to any other form of control by the partners. In Albanese, it was also easier to remove the manager than it was in this case. Though removal was only for cause, the service contract was between a single investor and the promoter. See id. at 412. In addition to the cause requirement, the RLLP partners also faced the dispersion and unanimity barriers. Because the RLLP partners could not remove or otherwise control Merchant, the arrangement in fact distributed power as would a limited partnership, and the first Williamson factor was satisfied. B. Were the RLLP partners so inexperienced and unknowledgeable in business affairs that they were incapable of intelligently exercising partnership or venture powers? The second Williamson factor asks whether “the partner or venturer is so inexperienced and unknowledgeable in business affairs that he is incapable of intelligently exercising his partnership or venture powers.” Williamson, 645 F.2d at 424. If the partner is inexperienced in “business affairs,” we will find a relationship of dependency on the promoter supporting a finding of investment contract, even if the partner possesses some powers under the arrangement. The district court erroneously applied this factor by looking to the general business experience of the partners. Howey itself focused on the experience of investors in the particular business, not their general business experience. In finding that the orange grove plus service contract was an investment contract, the Court said, “[the investors] are predominantly business and professional people who lack the knowledge, skill and equipment necessary for the care and cultivation of citrus trees.” Howey, 328 U.S. at 296, 66 S.Ct. at 1102. Similarly, in Albanese, we found it significant that “[e]ven though some plaintiffs were experienced businessmen, none of them had any experience in placing, managing, or servicing ice machines.” Albanese, 823 F.2d at 412. Williamson is not to the contrary. There the court held that it was “clear that the plaintiffs had the business experience and knowledge adequate to the exercise of partnership powers in a real estate joint venture.” Williamson, 645 F.2d at 425. The plaintiffs there not only were high corporate executives in a large business, but also had been involved in other similar real estate ventures. A focus on experience in the particular business is also more consistent with the Howey test. The ultimate question is whether the investors were led to expect profits solely from the efforts of others. Regardless of investors’ general business experience, where they are inexperienced in the particular business, they are likely to be relying solely on the efforts of the promoters to obtain their profits. See also Long v. Shultz Cattle Co., 881 F.2d 129, 134 n. 3 (5th Cir.1989) (interpreting Williamson to focus on the experience in the particular business, not business in general, and noting that “any holding to the contrary would be inconsistent with Howey itself’). In this case, the SEC presented uncon-tradicted evidence that the individual partners had no experience in the debt purchasing business. They were members of the general public, and included a railroad retiree, a housewife, and a nurse. Their possible general business experience is not significant in this case. They were relying solely on Merchant to operate the business, as evidenced by the fact that one-hundred percent of the partners chose Merchant as MGP. In Howey, the Supreme Court found an investment contract upon much less evidence of dependence: where only 85% of the orange grove acreage sold was managed by the promoter. Howey, 328 U.S. at 295, 66 S.Ct. at 1101. Merchant contends that the partners’ lack of preexisting experience is irrelevant because anyone with general business experience can easily learn to be successful in the debt purchasing business. That argument is not supported by the record, and to the extent the district court relied on it, the court clearly erred. Bonilla, a former associate of Wyer’s who operates a debt purchasing company, testified that someone needs two years of experience to be successful in the business. He and the SEC’s expert said that a great deal of sophistication is necessary to participate effectively in the business. Their testimony is supported by the indisputably complicated nature of the business. A successful debt purchaser must make sensitive pricing determinations of financial instruments, based on a multitude of fluctuating factors, and also must cultivate relationships with sophisticated financial institutions. The only person who testified that investors could easily learn the business was Wyer, and his testimony is unavailing for two reasons. First, his experience is atypical. He had extensive prior experience and relationships in the financial services industry. He admitted that his prior business involved building relationships with financial services players, which meant that he was both more likely to understand the business and better suited to cultivate the relationships necessary to operating a successful debt purchasing business. Merchant presented no evidence that RLLP partners had similar experience. Moreover, even with Wyer’s head start, he was not a successful participant in the debt purchasing industry. The history of Merchant’s operation of a debt purchasing business is one of unmitigated failure. Wyer’s experience thus showed nothing about how much time is necessary to develop actual expertise in the debt purchasing industry. The second Williamson factor was also present here. C. Were the partners so dependent on Merchant’s entrepreneurial or managerial ability that they could not replace it or otherwise exercise meaningful powers? The third factor asks whether “the partner or venturer is so dependent on some unique entrepreneurial or managerial ability of the promoter or manager that he cannot replace the manager of the enterprise or otherwise exercise meaningful partnership or venture powers.” Williamson, 645 F.2d at 424. The first Williamson factor analyzes the powers the partners practically retain under the arrangement with the promoter. The third factor provides that, even if the arrangement gives the partners some practical control, the instrument is an investment contract if the investors have no realistic alternative to the manager. See Albanese, 823 F.2d at 412. The SEC admits that Merchant had no special skill in the debt purchasing business. Instead it argues that each individual RLLP partnership had no reasonable alternative to Merchant because the power to remove Merchant as MGP was illusory, and because of the fact that the twenty-eight partnerships’ money was tied up in a common pool. We have already discussed the former, so we now focus on the latter. Merchant took the capital contributed by the twenty-eight partnerships, pooled it, and then pooled it further with capital raised by New Vision. The district court rejected the SEC’s position, finding that an RLLP that could break away would have reasonable alternatives for new management. Assuming (counterfactually) that an RLLP could break away, we conclude that the district court did not clearly err with respect to the availability of alternatives for new management. Evidence at trial showed that alternative managers approached Merchant expressing a willingness to manage an RLLP. Merchant also showed that New Vision had pooled RLLP funds of its own. It is not beyond the realm of possibility that a breakaway RLLP could have joined another pool that would have allowed it to recoup the returns that were expected at the investment’s inception. For a different reason, however, the RLLP partners did not have any realistic alternative to management by Merchant (in addition to having no practical ability to remove Merchant). That is because Merchant effectively had permanent control over each partnership’s assets. Merchant pooled the partnerships’ assets and invested them in pools of accounts owned by New Vision. Merchant had a service contract with New Vision that gave Merchant a right to the return of debt accounts only in certain limited circumstances, or upon termination of the entire contract. Beasley admitted at trial that the partnerships had no contractual right to demand the return of the debtor accounts. Thus, even if an individual partnership managed to replace Merchant, it would find that its major assets were tied up in fractional share form in a New Vision debt pool. Merchant asserts that the n-select process developed by Merchant with New Vision and EAM allowed partnerships to obtain their shares in large debt pools, and points to RLLP-19 as an example of the n-select process working. The district court agreed, but clearly erred. In the first place, the n-select process did not exist until well after the SEC’s investigation had begun, and long after all of the offerings of RLLP interests. Then, once developed, the n-select process was an available means for partitioning debt pools. But it did not provide the partnerships with the right to obtain their debt pools from New Vision. That right was Merchant’s alone, and it was circumscribed by the terms of the service contract. RLLP-19 is not a counterexample. By the time RLLP-19 withdrew in 2004, years after the events relevant here, Merchant’s sister company, Merchant Management, had taken over New Vision’s role in managing the debt pools. In addition, by that point, Merchant had no interest in resisting repossession of the debt pools: the withdrawal of one of the RLLPs both helped its prospects in the SEC investigation and did not hurt it financially, since it had already recommended liquidation and there were no more transaction fees to earn. To show that the partners could repossess their debt portfolios, Merchant (and the district court) relied on testimony from New Vision’s head, Fred Howard, to the effect that he would have been willing to allow an individual RLLP to withdraw its debtor accounts using the n-select process, even though the RLLP might not have had a contractual right to do so. Howard’s testimony about withdrawal was, however, relevant only to the period after Merchant, along with New Vision and EAM, developed the n-select process. The n-select process was developed after the SEC investigation began, with the apparent purpose of demonstrating to the SEC that the individual partnerships could remove then-assets from the pool (thus bolstering the argument that the partnership interests were not securities). Before that time, it was not even possible for an individual RLLP to withdraw its accounts, because there was no way to convert the RLLP’s fractional share in the pool into concrete and portable debtor accounts. And all of the RLLP offerings took place before the development of the n-select process. Thus, even if Howard had been willing to remit the debtor accounts, he had no means by which to do so. Howard’s testimony therefore did not establish that the partners had a reasonable alternative to Merchant at the time of the initial investment, and the district court clearly erred in concluding otherwise. Another practical limitation on the partners’ ability to find an alternative to Merchant was the fact that EAM was in actual possession of the debtor accounts, and had a contract with New Vision that limited the circumstances under which New Vision could repossess them. This means that, even crediting Howard’s testimony that he would have been willing to remit debtor accounts in his possession to an individual RLLP, the defendants produced no evidence that EAM would have been willing to return debtor accounts to New Vision that would have enabled Howard to send them to an RLLP. As a result, Howard’s testimony did not establish that individual RLLPs had a practical means of obtaining their accounts from EAM, and thus did not establish that they had a reasonable alternative to management by Merchant. The RLLPs’ lack of a realistic alternative to Merchant was present from the inception of the arrangement between Merchant and the partners. See Williamson, 645 F.2d at 424 n. 14. Wyer admitted that he and Beasley intended from the beginning to pool capital from multiple partnerships. The partnership agreement also expressly gave Merchant the authority to contract with a third-party service. Therefore, from the beginning, RLLP partners had no realistic alternative to management by Merchant, and the third Williamson factor was also present. D. Conclusion: RLLP interests were investment contracts For all of these reasons, the RLLP interests were investment contracts covered by the federal securities laws. The partners had the powers of limited partners, since they had no ability to remove Merchant and the purported authority to approve purchases was illusory. They were completely inexperienced in the debt purchasing industry. Finally, even if they could have removed Merchant (which they could not), they had no realistic alternative to Merchant as manager because their debt pools were in fractional form with a company whose only contractual relationship was with Merchant. Because the RLLP interests were investment contracts, and the defendants sold the interests without filing a registration statement, the defendants violated the registration provisions of the securities laws. See SEC v. Continental Tobacco Co., 463 F.2d 137, 155 (5th Cir.1972). IV. Securities fraud We turn next to the SEC’s contention that the defendants also committed securities fraud in the marketing of those interests. The SEC alleged that the defendants violated section 10(b) of the Exchange Act and section 17(a) of the Securities Act. 15 U.S.C. § 77q(a), 15 U.S.C. § 78j(b). To prove a 10(b) violation, the SEC must show (1) material misrepresentations or materially misleading omissions, (2) in connection with the purchase or sale of securities, (3) made with scienter. Aaron v. SEC, 446 U.S. 680, 695, 100 S.Ct. 1945, 1955, 64 L.Ed.2d 611 (1980). To show a violation of section 17(a)(1), the SEC must prove (1) material misrepresentations or materially misleading omissions, (2) in the offer or sale of securities, (3) made with scienter. Id. at 697, 100 S.Ct. at 1956. Finally, to show that the defendants violated section 17(a)(2) or 17(a)(3), the SEC need only show (1) material misrepresentations or materially misleading omissions, (2) in the offer or sale of securities, (3) made with negligence. Id. at 702, 100 S.Ct. at 1958. The district court concluded that the defendants made no material misrepresentations or omissions in marketing the RLLP interests, and also concluded that the defendants did not act with scienter. The district court did not address whether or not the defendants had acted with negligence. “Mixed questions of law and fact, such as questions of materiality, scienter, and reliance, involve assessments peculiarly within the province of the trier of fact and hence are renewable under the clearly erroneous rule.” Lucas v. Fla. Power & Light Co., 765 F.2d 1039, 1040-41 (11th Cir.1985). A finding will be held clearly erroneous only when “the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” United States v. U.S. Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948). In this opinion we address only issues involving whether there were misrepresentations or omissions of material matters, leaving the district court to reconsider scienter issues on remand. A. Projections and omission of performance information The test for materiality in the securities fraud context is “whether a reasonable man would attach importance to the fact misrepresented or omitted in determining his course of action.” SEC v. Carriba Air, 681 F.2d 1318, 1323 (11th Cir.1982) (citing TSC Indus. v. Northway, 426 U.S. 438, 449, 96 S.Ct. 2126, 2132, 48 L.Ed.2d 757 (1976)). The SEC first claims that Merchant made projections for the performance of the partnerships that were materially misleading. “Statements regarding projections of future performance may be actionable under Section 10(b) or Rule 10b-5 if they are worded as guarantees or are supported by specific statements of fact ... or if the speaker does not genuinely or reasonably believe them.” Kowal v. IBM Corp. (In re IBM Corp. Sec. Litig.), 163 F.3d 102, 107 (2d Cir.1998); see also Rubinstein v. Collins, 20 F.3d 160, 166, 168 (5th Cir.1994); Kowal v. MCI Communications Corp., 16 F.3d 1271, 1277 (D.C.Cir.1994). The SEC argues that Merchant promised investors a 3.2% quarterly or 16.5% annual return on capital, and that those returns were unreasonable given the high fees charged by Merchant and other service providers. The district court found that these projections were not materially misleading. We divide our analysis, and address first the projections made before June 2002. We conclude that the district court did not clearly err with respect to the partnership interests offered prior to June 2002. First, the defendants did not guarantee a rate of return. The partnership agreement explicitly advised partners that the returns were not guaranteed. The agreement also recognized that the return might be greater or worse: it provided for a 50-50 split of profits left after those distributions, but also disclosed that the return might be less, and that partners might in fact lose their entire capital contribution. Second, the district court also did not clearly err in determining that the projections were initially made in good faith and had a reasonable basis. Evidence in the record supports Merchant’s contention that the projections were not unreasonable when made. New Vision provided Merchant with the models it used to calculate the projected return on the investment, and Merchant used those models to calculate its internal projections. Merchant claims that the projected returns failed to materialize in part because the partnerships began operating soon after September 11, 2001. Further, Wyer testified that the projected return was so high precisely because the business was risky, and the fact that the investments underperformed did not necessarily demonstrate that the expected return at the inception was unreasonable. Thus, viewing the evidence as a whole, we do not believe that the district court clearly erred in determining that the projections were not initially made in good faith and with a reasonable basis. The initial projections were also rendered immaterial by the accompanying cautionary language. In this circuit we adhere to the bespeaks caution doctrine in assessing the materiality of forward-looking statements. “When an offering document’s projections are accompanied by meaningful cautionary statements and specific warnings of the risks involved, that language may be sufficient to render the alleged omissions or misrepresentations immaterial as a matter of law.” Saltzberg v. TM Sterling/Austin Assocs., 45 F.3d 399, 399 (11th Cir.1995). The cautionary language must be meaningful: boilerplate will not suffice. Id. A disclaimer does not provide per se immunity, precisely because the disclaimer must be meaningful and tailored to the risks the business faces. The bespeaks caution doctrine is ultimately simply “shorthand for the well-established principle that a statement or omission must be considered in context, so that accompanying statements may render it immaterial as a matter of law.” Kaufman v. Trump’s Castle Funding (In re Donald J. Trump Casino Sec. Litig.), 7 F.3d 357, 364 (3d Cir.1993) (cited in Saltzberg, 45 F.3d at 399). Here, the offering documents included meaningful cautionary language that informed investors of the risk inherent in the investment and rendered the pre-June 2002 projections immaterial. A bolded section titled “Risk Factors” warned that “becoming a General Partner is only appropriate for those persons capable of withstanding the risk of losing their entire capital contribution.” It advised that there are “no assurances” that any amount of debt “can actually be recovered.” It noted the riskiness of the market, noting that “[t]he debt collection business is intensely competitive” and that “[a]n economic downturn of any serious proportion or a national crisis could adversely affect the ability to collect debt in [sic] a timely basi