*739 Adriana Villarreal and Anthony Casarez approached Dr. David Masel with a proposal: if you will set up businesses that provide intraoperative neuromonitoring procedures, we will manage them, and through our signature billing practices, we can make you a substantial profit. It did not work out that way; on the $190 million worth of services the entities allegedly provided, Villarreal and Casarez collected less than $11 million. Masel and his business partner sued, alleging Villarreal and Casarez induced them to join the enterprise with material misrepresentations and omissions in violation of the Securities Exchange Act of 1934. The district court granted defendants' motion to dismiss, and plaintiffs appealed. For the reasons set forth below, we REVERSE and REMAND IN PART and AFFIRM IN PART.

I.

We recount the allegations as they are pleaded in the complaint, taking them as true, as we are required to do at the motion-to-dismiss stage. Arias-Benn v. State Farm Fire & Cas. Ins. Co. , 495 F.3d 228 , 230 (5th Cir. 2007). Intraoperative monitoring ("IOM") is a method of monitoring a patient's nervous system during surgery. The administration of IOM has a so-called technical component and a professional component. The technical component is usually performed by a certified neurophysiological intraoperative monitoring professional ("CNIM"), who operates the IOM machinery. The professional component is performed by a licensed physician, who evaluates the IOM readings produced by the CNIM.

Plaintiff David Masel is a neurosurgeon with more than 30 years of experience in the field. Defendant Adriana Villarreal is the owner of Medical Practice Solutions, L.L.C. ("MPS"), a medical services billing company specializing in billing for IOM services. Masel and Villarreal met through defendant Anthony Casarez, a CNIM Masel had worked with.

In the spring of 2014, during a chance encounter at a hospital, Casarez informed Masel that the IOM business was very profitable and proposed a meeting between Masel, Casarez, and Villarreal to discuss investment opportunities within the industry. Shortly thereafter, Masel, Casarez, and Villarreal met at a bakery in Plano, Texas. There, Villarreal told Masel that MPS had superior billing practices and was capable of generating the highest payouts for IOM procedures. Villarreal explained that her ability to pinpoint how much a given claim will pay within a margin of error of about 10 to 20 percent gave MPS an advantage in the industry. She said she could achieve this feat using a special algorithm-or "secret sauce," in her words-that she had developed while working for two large insurance companies. Villarreal told Masel that her "secret sauce" enabled her to collect $50,000 or more for each out-of-network claim for IOM services. She also represented that the reimbursement cycle for such claims was around six months. According to the complaint "Ca[s]are[z] agreed with and confirmed the veracity of these statements for Dr. Masel, adopting them as his own *740 representations." In a later email, Masel emailed Villarreal to ask what percentage of accounts receivable MPS could be expected to recover. Villarreal replied, "I always say 50% but a lot of times its [sic] more."

Sometime after the Plano meeting, Masel took this proposal to plaintiff Dinesh Chandiramani, his business partner. Motivated by Villarreal's representations at the Plano meeting, the two agreed to Villarreal's proposal.

At this point in the narrative, the experienced reader of securities-law cases might expect the plaintiffs to allege that Masel and Chandiramani turned over large sums of cash to Villarreal and Casarez, who then squandered it all. If only it were that simple. Though the complaint is somewhat vague as to precisely how the transaction got off the ground, this much is clear: instead of just giving Villarreal and Casarez a check , Masel and Chandiramani (in reliance on Villarreal's pitch) established and invested in a large collection of business entities, each founded for the purpose of providing IOM services. Masel and Chandiramani then hired Villarreal's companies to operate these entities in exchange for a financial interest in the entities.

The first of these entities was Neuron Shield, LLC, organized and headquartered in Texas. Neuron Shield, LLC contracted with defendant CGR Investments, LLC ("CGR"), a Texas company solely owned by Villarreal, to grant CGR a 35% non-voting net-profits interest in the company in exchange for CGR's agreement to provide management services for the company. 1 According to the complaint, the profits-interest agreement "acknowledged the applicability of the Securities Act of 1933." Though this is not reflected in any formal agreement contained in the record, the complaint alleges that CGR's responsibilities included management of all day-to-day operations of Neuron Shield, LLC, including marketing the business, retaining and paying employees, booking and scheduling IOM services, locating new sources of business, and collecting and billing for Neuron Shield's services. Neuron Shield, LLC also contracted with MPS to provide billing services. 2

Plaintiffs then formed Neuron Shield Partners I, LP; Neuron Shield Partners 2, LP; Neuron Shield Partners 3, LP; and Neuron Shield Partners 4, LP. Each of these LPs designated one of the Neuron Shield LLCs 3 as the general partner. Limited partners were divided into two classes: Class A and Class B. In each agreement, CGR was a Class B limited partner. The agreements limit Class A partners to physicians and Class B partners to non-physicians. They provide that Class A partners shall have no more than 40% aggregate ownership in the partnership, but the provision explaining how a party's percentage interest in the partnership is determined is not in the record.

*741 However, the complaint alleges that plaintiffs sold CGR a 35% interest in each of the plaintiff entities "other than Neuron Shield[,] LLC." 4 Each agreement prohibited the general partner (Neuron Shield, LLC) from making certain enumerated major decisions (such as receiving a capital contribution or dissolving the partnership) without obtaining approval of each partner with an interest equal to or greater than 20% of the partnership, which would include CGR as a 35% interest-holder. According to the complaint, the Limited Partnership agreements each also "acknowledged the applicability of the Securities Act of 1933." CGR's interests in Neuron Shield Partners I, 2, and 4 were later transferred to defendant IOS Management Services, LLC ("IOS"), of which Villarreal is a principal member. Plaintiffs also allege that they agreed to pay an 8% fee to MPS on all collections, regardless of whether they made a profit, although it is unclear what agreement this arrangement was based on or whether this arrangement applied to each entity concerned in this case.

Plaintiffs also formed several other entities. Neither the complaint nor the exhibits provide any clarification as to how these entities were structured or what formal powers the parties had in them. These other entities were: Neuron Shield 2, LLC; Neuron Shield 3, LLC; Neuron Shield 4, LLC; Neuron Shield 7, LLC; Neuron Shield 9, LLC; Neuron Shield Monitoring Associates, PLLC; Neuron Integrity Texas, PLLC; Neuron Integrity 1, PLLC; Neuron Integrity Partners, LLC; and Neuron Shield Monitoring Associates, PC. Each entity is a plaintiff in this action. 5

The complaint also contains scant information regarding the scope and degree of plaintiffs' involvement in the Neuron Shield Enterprise. As discussed above, CGR, and in certain cases IOS, handled the day-to-day operations of the entities. It is clear that Masel was not entirely removed from the Neuron Shield Enterprise's operations, however. The complaint alleges that Masel's "established relationships with neurosurgeons and referral sources in the neurosurgical industry were critical to Neuron Shield's growth" and that these relationships "gave him an edge over other providers when competing to offer IOM services to these doctors." Masel's participation "paid off. In two and [a] half years since it was established, Neuron Shield provided over $190 Million in IOM services."

The Neuron Shield Enterprise's revenues ultimately fell far short of what Villarreal had promised. Of the $190 million the Neuron Shield Enterprise billed for IOM services, MPS collected just $11 million. MPS was unable to collect anything on half of the claims it submitted, and it collected less than one percent of the value on three-quarters. Based on this outcome, plaintiffs allege that no "secret sauce" algorithm existed. The reimbursement cycle also lasted significantly longer than Villarreal had represented. Instead of a six-month cycle, the complaint alleges that Villarreal later stated, "in sworn courtroom testimony" given during a parallel proceeding, that the cycle takes "anywhere from 12 to 18 months." It also turned out, unbeknown to plaintiffs, that defendants *742 had many other clients who provided IOM services. In a "recent court hearing," Villarreal testified that MPS provided billing services to "109 or 110" IOM providers. This is contrary to what Villarreal had told Masel previously: when asked if she had any other clients, Villarreal told Masel that she worked with only one other "small" IOM provider.

According to the complaint, defendants actively stole business from Neuron Shield entities for the benefit of their other clients. As an example, plaintiffs cite to an instance where a physician had been in discussions with Neuron Shield to use its IOM services but did not ultimately sign on. Sometime thereafter, the physician contacted Masel to ask him why, following a procedure performed by a CNIM wearing a Neuron Shield uniform, the bill for the procedure was sent to "an unknown and undisclosed entity." When Masel confronted Casarez about this development, Casarez admitted to Masel that he had redirected the physician from Neuron Shield to a competing business. When plaintiffs thereafter raised their concerns regarding defendants' conflicts of interest to Villarreal, she refused to proceed with any collections for the Neuron Shield entities "unless Neuron Shield signed over a vested interest in the uncollected accounts receivable to MPS and agreed to a one-sided confidentiality agreement." When plaintiffs refused, defendants terminated their contracts with all Neuron Shield entities.

Plaintiffs sued defendants in federal district court. They sought relief under the Securities Exchange Act of 1934, the Texas Securities Act, Texas common law, the Racketeer Influenced and Corrupt Organizations Act ("RICO"), and the Texas Theft Liability Act. With respect to their securities-fraud claims, plaintiffs cited several misrepresentations and omissions. The precise statements alleged are set forth in the discussion below. Broadly speaking, plaintiffs alleged that defendants (1) misrepresented that a "secret sauce" algorithm existed, when in fact it did not; (2) misrepresented how much MPS was capable of collecting with this supposed algorithm and how soon collections could be expected; and (3) made material omissions by failing to disclose their intention to set up competing businesses.

The district court dismissed each of these claims for failure to state a claim under Federal Rule of Civil Procedure 12(b)(6). With respect to plaintiffs' securities-fraud claims, the court found that plaintiffs failed to plead several of the alleged misrepresentations or omissions with sufficient particularity, that other statements were inactionable as either puffery or future predictions, and in all other cases plaintiffs had failed to adequately plead that defendants acted with scienter in making the alleged misrepresentations or omissions. The court proceeded directly to the merits of plaintiffs' claims and did not consider the threshold question (raised by the defendants) whether plaintiffs had adequately pleaded the existence of a security. The district court also dismissed plaintiffs' RICO claims on the merits and declined to exercise supplemental jurisdiction over the state-law claims. Plaintiffs appeal, arguing the district court erred in dismissing their securities-fraud claims. 6

II.

A.

We review a district court's grant of a motion to dismiss de novo, *743 applying the same standard applied by the district court. Boyd v. Driver , 579 F.3d 513 , 515 (5th Cir. 2009). To survive a motion to dismiss, "a complaint must contain sufficient factual matter, accepted as true, to 'state a claim to relief that is plausible on its face.' " Waller v. Hanlon , No. 18-10561, 2019 WL 1783558 , at *4 (5th Cir. Apr. 24, 2019) (quoting Ashcroft v. Iqbal , 556 U.S. 662 , 678, 129 S.Ct. 1937 , 173 L.Ed.2d 868 (2009) ). In conducting this analysis, we accept as true any well-pleaded factual allegations, but we do not accept as true legal conclusions or " 'naked assertion[s]' devoid of 'further factual enhancement.' " Iqbal , 556 U.S. at 678 , 129 S.Ct. 1937 (alteration in original) (quoting Bell Atl. Corp. v. Twombly , 550 U.S. 544 , 557, 127 S.Ct. 1955 , 167 L.Ed.2d 929 (2007) ).

B.

Before proceeding to the merits of plaintiffs' securities-fraud claims, we must first address the threshold question-not considered by the district court-whether plaintiffs have successfully pleaded the existence of a security. Plaintiffs seek relief under § 10(b) of the Securities Exchange Act of 1934 (the "1934 Act"). 15 U.S.C. § 78j(b). To successfully state a private cause of action under § 10(b), a plaintiff must plead "(1) a material misrepresentation or omission by the defendant; (2) scienter; (3) a connection between the misrepresentation or omission and the purchase or sale of a security; (4) reliance upon the misrepresentation or omission; (5) economic loss; and (6) loss causation." Affco Invs. 2001, L.L.C. v. Proskauer Rose, L.L.P. , 625 F.3d 185 , 192 (5th Cir. 2010) (quoting Stoneridge Inv. Partners, LLC v. Sci.-Atlanta, Inc. , 552 U.S. 148 , 157, 128 S.Ct. 761 , 169 L.Ed.2d 627 (2008) ). The third factor in this test-a connection between the misrepresentation or omission and the purchase or sale of a security-requires the existence of a security. The 1934 Act defines "security" broadly to include, among other things, an "investment contract." 15 U.S.C. § 78c(a)(10). The Supreme Court has in turn interpreted "investment contract" to mean "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 , 90 L.Ed. 1244 (1946). 7 We have gleaned from this statement a three-factor test, requiring plaintiffs to show "(1) an investment of money; (2) in a common enterprise; and (3) on an expectation of profits to be derived solely from the efforts of individuals other than the investor." Williamson v. Tucker , 645 F.2d 404 , 417 (5th Cir. May 1981). In conducting this analysis, "form should be disregarded for substance and the emphasis should be on economic reality." United Hous. Found., Inc. v. Forman , 421 U.S. 837 , 848, 95 S.Ct. 2051 , 44 L.Ed.2d 621 (1975) (quoting Tcherepnin v. Knight , 389 U.S. 332 , 336, 88 S.Ct. 548 , 19 L.Ed.2d 564 (1967) ). Accordingly, the label given by parties to a certain transaction will not be determinative of whether the parties have engaged in the purchase or sale of a security. See Williamson , 645 F.2d at 423 ("A scheme which sells investments to inexperienced and unknowledgeable members of the general public cannot escape the reach of the securities laws merely by labelling itself a general partnership or joint venture.").

The only Howey factor disputed in this case is the third one: whether investors *744 have been "led to expect profits solely from the efforts of the promoter or a third party." Howey , 328 U.S. at 299 , 66 S.Ct. 1100 . In applying this third factor, we have not understood the Supreme Court to use the word "solely" in its literal sense. Williamson , 645 F.2d at 418 . Instead, the question is "whether the efforts made by those other than the investor are the undeniably significant ones, those essential managerial efforts which affect the failure or success of the enterprise." Affco , 625 F.3d at 190 (quoting SEC v. Koscot Interplanetary, Inc. , 497 F.2d 473 , 483 (5th Cir. 1974) ).

Interests in a general partnership typically do not meet Howey 's third prong. Youmans v. Simon , 791 F.2d 341 , 346 (5th Cir. 1986). The interest of a limited partner, by contrast, "is analogous to that of a stockholder in a corporation," and thus limited partnership interests typically are securities under Howey 's third prong. Id. Nonetheless, even a limited partnership interest may not be a security when limited partners are given such managerial control that it can no longer be said that the limited partners are dependent on the entrepreneurial skills of the promoter or a third party. See Frazier v. Manson , 651 F.2d 1078 , 1080 (5th Cir. Unit A July 1981).

The unusual facts of this case make this analysis particularly vexing. Typically, Williamson cases (and securities-fraud cases generally) involve the investment of money by a plaintiff either directly with the defendant or in an entity controlled by the defendant. Here, by contrast, plaintiffs' investment went into entities they themselves set up. Although defendants exercised day-to-day managerial control of each of those entities, it was plaintiffs who held controlling interests. Plaintiffs also marketed the entities to potential customers using Masel's professional connections. Nonetheless, it is undeniable that the inducement for setting up these entities was the profit that might be generated through the collection of payments-profits only made possible by Villarreal's promised "secret sauce."

Further complicating matters is the fact that plaintiffs do not make clear precisely what it was in this case that they bought or sold that amounted to a security-i.e., which agreement in this case constituted an investment contract. By our lights, plaintiffs attempt to argue that defendants' interests in the Neuron Shield entities were securities and that the agreement conveying those interests to defendants was a sale of securities (in relation to which defendants made fraudulent representations). In this scenario, somewhat idiosyncratically, plaintiffs are the promoters, and defendants are the investors.

We note also that the structure of our analysis may appear highly counterintuitive at first blush; because plaintiffs were the promoters of the Neuron Shield entities, the question is whether defendants exercised sufficient managerial control such that the third Howey factor was satisfied. Thus, plaintiffs' case depends on their demonstrating that defendants were passive investors. This is unlike the typical case, wherein defendants claim plaintiffs exercised such control that their interests were not securities. Unusual as this mode of analysis may be, nothing about it runs contrary to federal securities law. The 1934 Act prohibits fraud "in connection with the purchase or sale" of a security. 15 U.S.C. § 78j(b). Accordingly, so long as the fraud alleged is connected with the purchase or sale of a security, it is within the bounds of the 1934 Act, and a 10(b) suit may proceed, regardless of which side of the caption an investor or promoter falls on in a given case. See *745 Chadbourne & Parke LLP v. Troice, 571 U.S. 377 , 387, 134 S.Ct. 1058 , 188 L.Ed.2d 88 (2014) ("A fraudulent misrepresentation or omission is not made 'in connection with' such a 'purchase or sale of a covered security' unless it is material to a decision by one or more individuals (other than the fraudster) to buy or to sell a 'covered security.' ").

In all, plaintiffs allege that they set up 15 entities, including LLCs, LPs, PLLCs, and a PC. Defendants only argue on appeal that plaintiffs have failed to allege that the limited partnership interests were securities; they do not discuss the LLCs, PLLCs, or the PC. As discussed, the complaint and record tell us little about any of these entities: neither explains what each entity did, why each additional entity was set up, or what powers the parties had within them. Nonetheless, we will evaluate the limited partnerships based on the documents provided and the facts alleged.

Each of the Neuron Shield LPs is structured in roughly the same manner. Neuron Shield, LLC served as the general partner, and CGR and others served as the limited partners. Limited partners were divided into Class A and B partners, with CGR listed as a Class B partner in each agreement. The agreements, together with the complaint, tell us that (1) Class A partners could not possess more than 40% of aggregate ownership of the partnership; (2) Neuron Shield, LLC could not make major decisions without the approval of each partner holding an interest of 20% or greater in the partnership; and (3) CGR held a 35% interest in each partnership.

Defendants argue that they had too much control of the LPs for their interests in them to be securities. For support, defendants cite the district court opinion in Frazier v. Manson , 484 F. Supp. 449 (N.D. Tex. 1980), aff'd , 651 F.2d 1078 , as well as this court's brief affirming opinion in that case. Defendants cite Frazier for the proposition that where a party maintains managerial rights in an investment, the investment is not a security, even where that investment takes the form of a limited partnership. Thus, defendants argue, their extensive participation in the day-to-day work of the LPs negates the possibility that a security existed in this case.

This misunderstands Frazier 's characterization of "managerial rights." In Frazier , the plaintiff's involvement was not merely that of a limited partner in the relevant enterprise. See 484 F. Supp. at 450-51 . Manson-Frazier Companies was the general partner in the underlying limited partnerships. Manson-Frazier Companies, in turn, was a general partnership in which the plaintiff served as a general partner. Id. In granting defendants' motion for summary judgment on the grounds that plaintiff had failed to prove the existence of a security, the district court found it could not "segregate Frazier's limited partnership interests from his status as a partner in [Manson-Frazier Companies]." Id. at 452 . It was this status that gave him the "managerial rights" to "participate actively in the daily business operations of both [Manson-Frazier Companies] and the limited partnerships," id. , and it was on this basis that we affirmed. Frazier , 651 F.2d at 1080 ("The District Court properly concluded that Frazier's managerial rights negated the possibility that his limited partnership interests were securities. An investor does not always need the extensive protection of the federal securities laws when he or she has partial control of an enterprise." (citation omitted)).

Here, by contrast, defendants have not entered into a general partnership agreement with plaintiffs. Instead, CGR's (or IOS's) involvement in the relevant entities is solely as a limited partner. At this stage, not enough is known about defendants' formal powers under the limited *746 partnership agreements to say whether their powers were akin to those of a general partner. True, defendants did in fact take on significant day-to-day responsibilities within these LPs, but defendants took this control pursuant to what was essentially a service agreement subservient to the plaintiffs' formal powers under the limited-partnership agreements. As a factual matter, no agreement is provided or alleged that grants them any ultimate power to control those entities. As a legal matter, defendants cite to no case where the conduct of investors, without any formal power to engage in that conduct, has sufficed to overcome the presumption that limited partnerships are securities. More by way of formal powers is needed to do so. Cf. Steinhardt Grp. Inc. v. Citicorp , 126 F.3d 144 , 153-54 (3d Cir. 1997) (finding limited partnership interest was not a security where limited partner retained powers "far afield of the typical limited partnership agreement," including power to remove general partner without notice, propose and approve new business plan, and veto the general partner's business plan). Here, the record only establishes that defendants had the power to block certain enumerated business decisions, including receiving capital contributions, admitting a new partner, and dissolving the partnership. We will not expand the law to say that such veto powers, standing alone, suffice to negate the existence of an investment contract when it comes to limited partnership interests.

We therefore conclude that the limited partnership interests in this case were securities. Defendants do not argue on appeal that the interests in the other Neuron Shield entities were not securities, and we therefore will not analyze those entities under Howey . Moreover, nothing in the record undermines the conclusion that the entities discussed were securities for the same reasons stated above. There is no pleading, for example, suggesting that defendants had legal powers under any of the LLC entities that either matched or went beyond their powers under the LPs. We acknowledge the possibility that, during the course of discovery, new facts may emerge that cast doubt on this conclusion. We are limited at this stage, however, to the complaint and documents attached as exhibits. On the basis of those sources alone, we cannot say plaintiffs have failed to plead the existence of a security.

C.

i.

Having determined that plaintiffs have adequately pleaded the existence of a security, we proceed to the merits of plaintiffs' securities-fraud claims. As discussed, the Securities Exchange Act of 1934 makes it unlawful "[t]o use or employ, in connection with the purchase or sale of any security ... any manipulative or deceptive device or contrivance in contravention of such rules and regulations as the [SEC] may prescribe as necessary or appropriate in the public interest or for the protection of investors." 15 U.S.C. § 78j(b). The SEC has interpreted this provision to prohibit the making of "any untrue statement of a material fact or [omission of] a material fact necessary in order to make the statements made, in the light of the circumstances under which they were made, not misleading" in connection with the sale of a security. 17 C.F.R. § 240 .10b-5(b). Courts have read the above statute and regulation to imply the existence of a private cause of action, requiring proof of six elements: "(1) [a] misstatement or omission (2) of material fact (3) in connection with the purchase or sale of a security, which was made (4) with scienter, and upon which (5) plaintiff justifiably relied, (6) proximately causing injury to the plaintiff."

*747 Rosenzweig v. Azurix Corp. , 332 F.3d 854 , 865 (5th Cir. 2003).

The Private Securities Litigation Reform Act ("PSLRA") requires plaintiffs in 10b-5 actions to satisfy a pleading requirement higher than the ordinary Rule 12(b)(6) standard. For the first element of the cause of action-a misstatement or omission-plaintiffs must "specify each statement alleged to have been misleading, the reason or reasons why the statement is misleading, and, if an allegation regarding the statement or omission is made on information and belief, the complaint shall state with particularity all facts on which that belief is formed." 15 U.S.C. § 78u-4(b)(1). This pleading standard requires plaintiffs to set forth the who (the speaker), the what (the statements alleged to have been misleading, as well as their content), the when (when the statement was made), the where (where the statement was made), and the why (what the defendant obtained by making the statement, and why it was misleading). See Rosenzweig , 332 F.3d at 866 . Failure to plead any of these facts with particularity warrants dismissal of the case. See id. On the fourth element-that the statements or omissions were made with scienter-plaintiff must plead "with particularity facts giving rise to a strong inference that the defendant acted with" scienter. § 78u-4(b)(2)(A). In 10b-5 actions, "scienter" ranges from intentional deception to severe recklessness, the latter being defined as "an extreme departure from the standards of ordinary care, [presenting] a danger of misleading buyers or sellers which is either known to the defendant or is so obvious that the defendant must have been aware of it." Rosenzweig , 332 F.3d at 866 (quoting Nathenson v. Zonagen Inc. , 267 F.3d 400 , 408 (5th Cir. 2001) ).

Plaintiffs alleged both misrepresentations and omissions as part of their case under 10b-5. The misrepresentations are as follows:

(1) Villarreal's representation to Masel that MPS was "superior at billing IOM procedures and could generate the highest payouts." ("Statement 1").

(2) Villarreal's representation that she developed a "secret sauce" while working at other insurance companies that allowed her to "pinpoint how much any given claim will pay within a 10 to 20 percent margin of error." ("Statement 2").

(3) Villarreal's representation that she could generate "$50,000 or more for each out-of-network IOM claim," and that she "could generate $20,000 in additional revenue for the technical component of monitoring, and that she typically got much more, up to $50,000 for the professional component." ("Statement 3").

(4) Villarreal's representation that "the reimbursement cycle for out-of-network claims is around six months." ("Statement 4").

(5) When Masel inquired about the percentage of accounts receivable that MPS could be expected to recover on a given claim, Villarreal responded "I always say 50% but a lot of times its more." ("Statement 5").

(6) Villarreal's repeating of Statement 4 at a later date. ("Statement 6").

(7) Villarreal's statement that MPS's "average out of network claim will reimburse around 50-75k." ("Statement 7").

Plaintiffs also allege the following omissions:

(1) Defendants' failure to disclose their investments in other IOM providers. ("Omission 1").

(2) Defendants' failure to disclose their intent to set up other IOM businesses to *748 compete with Neuron Shield. ("Omission 2").

(3) Defendants' failure to disclose that plaintiffs would not receive compensation for most IOM cases. ("Omission 3").

Plaintiffs allege that Casarez adopted as his own Statements 1-4 above and is therefore liable for them as well. They also allege that MPS, CGR, and IOS are liable for each of these misrepresentations because Villarreal and Casarez made them while acting as their agents. For the reasons set forth below, we conclude that Statements 1, 6, and 7, as well as all three omissions, were properly dismissed. But we conclude that plaintiffs adequately stated a 10b-5 claim with regard to Villarreal and the defendant entities for Statements 2-5. However, we also conclude that plaintiffs' case against Casarez fails with regard to these statements.

ii.

We address first the statements and omissions properly dismissed by the district court, each in turn. The district court dismissed Statement 1-that MPS had superior billing procedures and was capable of generating the highest payouts-because it found the statement to be inactionable puffery. See Emps.' Ret. Sys. v. Whole Foods Mkt., Inc. , 905 F.3d 892 , 901 (5th Cir. 2018) ("[A] company's 'generalized, positive statements' are immaterial because they do not alter a reasonable investor's assessment of the company's prospects.") We express no opinion as to whether such a statement constitutes puffery because we believe there is a sounder basis for affirmance. Whether or not reasonable investors would rely on such statements, there is no allegation in plaintiffs' complaint that the statements were false when made, i.e., that MPS's billing procedures were not in fact superior and that they were not capable of generating the "highest payouts." This factual scenario is akin to one addressed by this court in Whole Foods . In that case, plaintiffs alleged that Whole Foods had made an actionable misrepresentation when it stated that its prices were "competitive." Id. at 901 . While acknowledging that the complaint may have established that Whole Foods' prices were not " as competitive as advertised," we concluded that "it need not follow that they were not competitive." Id. Similarly here, although the payouts generated by MPS fell short of what Villarreal represented, this does not mean that these payouts were not "the highest" or that the billing procedures were not "superior." Thus, the district court properly dismissed Statement 1 as inactionable because of plaintiffs' failure to allege with particularity that such statements were false when made.

Statements 6 and 7 also fail under the heightened pleading standard imposed by the PSLRA. As the district court recognized, the complaint "fail[ed] to specify the place Villarreal allegedly made such statements." Plaintiffs maintain on appeal that these statements were "obvious[ly]" made in correspondence because of their inclusion of misspelled words, quotation marks, and brackets. First, we express doubt as to whether the existence of typos, quotation marks, and brackets in a court filing indicates that the filing is quoting from correspondence. Second, and more to the point, we are aware of no case supporting the claim that the PSLRA's heightened standard can be satisfied where the context in which a statement is made has been implied by spelling and punctuation. The statute requires such matters to be set forth "with particularity," and we insist on nothing less.

In the alternative, plaintiffs request that this court reverse the judgment of the district court with instructions to *749 grant them leave to amend so that they can specify that Statements 6 and 7 were made in written correspondence. Generally, leave to amend should be "freely give[n] ... when justice so requires." Fed. R. Civ. P. 15(a)(2). But for leave to be given, it must be requested. Here, plaintiffs failed to request leave to amend at any point following the filing of their first amended complaint. Our rule that arguments not raised below are waived on appeal applies with equal force to requests for leave to amend. See Cent. Sw. Tex. Dev., L.L.C. v. JPMorgan Chase Bank, Nat'l Ass'n , 780 F.3d 296 , 300 (5th Cir. 2015) (treating as waived argument not raised before the district court); Ransom v. Nat'l City Mortg. Co. , 595 F. App'x 304 , 306 (5th Cir. 2014) (unpublished) (finding plaintiffs could not seek leave to amend on appeal when they had failed to do so in the district court). And a party's obligation to seek leave to amend from the district court first is not discharged, as plaintiffs seem to imply, when the party requests "any alternative relief." A holding to the contrary would render our waiver rules meaningless. See F.D.I.C. v. Mijalis , 15 F.3d 1314 , 1327 (5th Cir. 1994) ("[I]f a litigant desires to preserve an argument for appeal, the litigant must press and not merely intimate the argument during the proceedings before the district court. If an argument is not raised to such a degree that the district court has an opportunity to rule on it, we will not address it on appeal.").

Turning to the omissions, we conclude that none states a claim under 10b-5 and the PSLRA. Omissions 2 and 3 (failure to disclose an intent to set up competing businesses and that plaintiffs would not receive compensation on most claims) fail under the PSLRA's particularity requirement. To plead an omission with sufficient particularity, plaintiff must specifically plead when a given disclosure should have been made. Carroll v. Fort James Corp. , 470 F.3d 1171 , 1174 (5th Cir. 2006). Here, plaintiffs contend on appeal that defendants should have made these disclosures at the initial meeting between Masel and defendants. However, plaintiffs do not point us to anywhere in the complaint where this is alleged, and we are unable to discern therein any specific allegations to this effect. Accordingly, plaintiffs failed to plead with sufficient particularity the existence of a fraudulent omission on the part of defendants for Omissions 2 and 3.

As for Omission 1-defendants' failure to disclose conflicts of interest-we will assume, as the district court did, that defendants should have disclosed their alleged conflict of interest when Masel asked Villarreal if she had any other clients. Even granting plaintiffs that fact, the complaint does not plausibly allege that this conflict existed when the disclosure should have been made. The only support plaintiffs offer for their claim that defendants operated under a conflict of interest is Villarreal's testimony in a recent proceeding that she had over 100 IOM clients. Nowhere in the complaint is it alleged that Villarreal had over 100 clients at the time Masel inquired about her other business. Thus, plaintiffs' fraudulent omission claims were properly dismissed.

iii.

Statements 2 through 5, however, were adequately pleaded and should not have been dismissed. Broadly characterized, each representation relates to MPS's capabilities to pinpoint how much it will collect on a given claim and how long MPS would take to collect. Statements 2 and 4 were Villarreal's representations that she had developed a "secret sauce" to pinpoint how much a given statement would pay within a 10-20% margin of error and that the *750 reimbursement cycle for out-of-network claims was six months, respectively. The district court dismissed both because the complaint lacked "any allegations demonstrating that such statements were false when they were made." We disagree.

Beginning with Statement 2, plaintiffs alleged that no "secret sauce" existed at the time of Villarreal's alleged representations. They supported this claim by pointing to the fact that MPS was ultimately unable to collect on the overwhelming majority of claims it billed. Defendants contend that plaintiffs are attempting to prove fraud by hindsight by pointing to later events in order to shed light on the truth or falsehood of earlier statements. However, evidence of later events can provide useful circumstantial evidence that a given representation was false when made. See Lormand v. US Unwired, Inc. , 565 F.3d 228 , 254 (5th Cir. 2009) (concluding scienter was adequately alleged, even when allegation was partially based on later admissions by defendants, because the admissions "directly and cogently tend to prove their state-of-mind at the time of their misleading statements and omissions, i.e., they are evidence that the defendants actually knew earlier that the course of action would turn out badly"). Moreover, as the foregoing parenthetical indicates, fraud-by-hindsight issues arise in the context of the scienter factor, not the misrepresentation factor. Id. ("This is not the classic fraud by hindsight case where a plaintiff alleges that the fact that something turned out badly must mean defendant knew earlier that it would turn out badly.") (quoting Miss. Pub. Emps.' Ret. Sys. v. Bos. Sci. Corp. , 523 F.3d 75 , 91 (1st Cir. 2008) (emphasis added)).

Where, as here, the representation in question concerned an asset or skill possessed by the defendant (here, an algorithm), the defendant's failure to perform as promised casts doubt on whether he possessed that skill in the first place. Plaintiffs draw a helpful analogy: suppose a pianist represents that he is well-trained and commits to perform Gershwin's Rhapsody in Blue at a concert some time in the future. If he later arrives unable to play even Chopsticks, it becomes highly unlikely that he was a talented piano player to begin with. Here, Villarreal claimed to possess an algorithm that enabled her to pinpoint the payout for a given procedure within a 10-20% margin of error. The complaint alleges that MPS collected nothing on more than half of all claims and less than 1% on 75% of all claims. Taking these allegations as true, they allow for the plausible inference that either (1) Villarreal had no algorithm and therefore misrepresented her capabilities when she pitched her investment to Masel, or (2) Villarreal had an algorithm and some intervening event prevented her algorithm from functioning as described. Since no intervening event is alleged, the former scenario is the more plausible of the two. Plaintiffs have therefore adequately pleaded a misrepresentation with respect to Statement 2.

The same reasoning applies to Statement 4. Villarreal's assertion that claims would be reimbursed within six months was directly contradicted by her later sworn testimony in another proceeding that reimbursement took 12 to 18 months. There is no allegation that some intervening event made the reimbursement cycle substantially longer. Perhaps discovery may help explain this alleged inconsistency, but we may not probe such depths at this early stage. Accordingly, plaintiffs adequately alleged a misrepresentation for Statements 2 and 4.

We turn next to Statements 3 and 5. These both concern the profits MPS could generate. The district court dismissed both of these, finding they constituted *751 nonactionable future predictions. The district court is correct that "projections of future performance not worded as guarantees are generally not actionable under the federal securities laws." Shushany v. Allwaste, Inc. , 992 F.2d 517 , 524 (5th Cir. 1993) (quoting Krim v. BancTexas Grp., Inc. , 989 F.2d 1435 , 1446 (5th Cir. 1993) ). 8 However, we do not read Statements 3 and 5 as projections of future performance. Rather, they relate to the present capabilities of the MPS algorithm. To understand this contrast, consider an alternative scenario in which Villarreal stated that plaintiffs could expect profits of $50,000, without referencing the algorithm. In that case, the statement would be predictive in nature. Here, by contrast, Villarreal is alleged to have stated that "she had the ability to generate $50,000," and that she typically collects 50% or more of accounts receivable. The latter statement concerns how the algorithm had previously performed. The former statement concerns how the algorithm could perform at the time the statement was made. Because, as discussed, the failure of Villarreal to come close to generating the profits represented casts doubt on the veracity of her statements as to the algorithm's existence, it is plausible her statements as to how well the supposed algorithm could perform are false as well. Accordingly, we find that Statements 3 and 5 also satisfied the first element of a 10b-5 claim.

The district court concluded in the alternative that, even if Statements 2 through 5 were misrepresentations, plaintiffs had not adequately pleaded that defendants possessed the required scienter in making them. As discussed, the PSLRA requires plaintiffs to allege with particularity facts giving rise to a "strong inference" of scienter. 15 U.S.C. § 78u-4(b)(2)(A). In this context, scienter means, at a minimum, severe recklessness. See Lormand , 565 F.3d at 251 . The PSLRA's requirement of a "strong inference" does not require an "irrefutable" inference "or even the most plausible of competing inferences." Id. (quoting Tellabs, Inc. v. Makor Issues & Rights, Ltd. , 551 U.S. 308 , 324, 127 S.Ct. 2499 , 168 L.Ed.2d 179 (2007) ). Rather, the inference of scienter must be "cogent and at least as compelling as any opposing inference of nonfraudulent intent." Id. at 252 (quoting Tellabs , 551 U.S. at 314 , 127 S.Ct. 2499 ). This inference may be made using not only direct evidence but circumstantial evidence as well. Id. at 251.

In their complaint, plaintiffs alleged that defendants knew such representations were false because they were based on metrics and information in their control. The district court deemed these allegations to be conclusory and determined that they did not give rise to a strong inference of scienter. For support, the district court cited Indiana Electrical Workers' Pension Trust Fund IBEW v. Shaw Group, Inc. , 537 F.3d 527 (5th Cir. 2008). There, we stated that " 'general allegations and conclusory statements, such as stating [defendants] knew ... adverse material' do not contribute to a strong inference of scienter." Id. at 538-39 (alteration and omission in original) (quoting Fin. Acquisition Partners LP v. Blackwell , 440 F.3d 278 , 289-90 (5th Cir. 2006) ). Shaw is distinguishable. In the quoted portion of Shaw , this court confronted the question whether the CEO and CFO of a corporation were aware at the time they made certain representations about the corporation's financial performance that the corporation's accounting software "was not functional." Id. at 538. Plaintiffs supported their scienter *752 claim with allegations that a letter was sent to certain senior officials in the corporation detailing the problems with the software, but no specific allegations suggesting the CEO and CFO were themselves made aware of such issues. Id. at 538-39.

Here, by contrast, plaintiffs allege that Villarreal knew her algorithm could not generate the expected returns because her representations on this subject were based on metrics and information within her own control. That Villarreal would possess such information is plausible: returning to the pianist analogy discussed above, if it is plausible that a self-purported pianist misrepresented his ability to play piano, it is similarly likely that he knew he lacked such capability. This does not amount, as defendants contend, to an allegation that defendants "must have known" that their statements were false, or an allegation of scienter based solely on defendant's position. Rather, plaintiffs' scienter allegations proceed from the premise, plausible on its face, that because Villarreal had developed this algorithm and used it previously, she knew how and whether it would work. Since plaintiffs have plausibly alleged that this algorithm did not exist, they have similarly alleged that Villarreal was aware of this fact. Indeed, the competing inference-that Villarreal mistakenly, but honestly, believed she had developed and successfully employed an algorithm to precisely predict how much an IOM claim would pay out-seems unlikely. The same goes for Villarreal's representation that the reimbursement cycle took about six months: this too was purportedly based on information gleaned from her work billing for medical procedures. Defendants do not contest whether Villarreal's misrepresentations are attributable to MPS, CGR, and IOS, and we therefore do not consider the issue. Based on the foregoing, we conclude that plaintiffs have successfully pleaded a case for securities fraud against Villarreal, MPS, CGR, and IOS with respect to Statements 2 through 5.

The reasoning above does not extend to Casarez, however. Although the complaint alleges that Villarreal had developed the secret-sauce algorithm and previously used it, nothing in the complaint suggests that Casarez understood the algorithm, how it worked, or whether it existed. Nor is there any allegation that Casarez was aware that the billing cycle would take longer than six months. Although the representations Casarez adopted turned out to be false, nothing in the complaint suggests that he had any basis to know that they were false. At most, the face of the complaint establishes that Casarez was negligent in agreeing with Villarreal's statements about her algorithm and billing cycle. Accordingly, the district court correctly dismissed the claims against Casarez.

III.

Because we conclude that the district court erred in dismissing plaintiffs' securities-fraud claims for Statements 2 through 5 in their suit against Villarreal, MPS, CGR, and IOS, we REVERSE and REMAND IN PART with regard to the court's dismissal of those claims. Because the district court did not err in dismissing plaintiffs' claims stemming from the remaining alleged misrepresentations and omissions as to each defendant, as well as each claim against Casarez, we AFFIRM IN PART with regard to the court's dismissal of those claims. Each party shall bear its own costs.