On Appeal from the United States District Court for the District of New Jersey (D.C. No. 1:19-cv-05296 & 1:19-cv-08105) District Judge: Honorable Christine P. O’Hearn

Submitted Pursuant to Third Circuit L.A.R. 34.1(a) October 29, 2024

Before: CHAGARES, Chief Judge, PORTER and CHUNG, Circuit Judges

(Filed: August 20, 2025)Page 2 Laurence M. Rosen Daniel I. Tyre-Karp The Rosen Law Firm 275 Madison Avenue, 40th Floor New York, NY 10016

Jacob A. Goldberg The Rosen Law Firm 101 Greenwood Avenue, Suite 440 Jenkintown, PA 19046

Steven F. Hubachek Tor Gronborg Trig R. Smith Harini P. Raghupathi Robbins Geller Rudman & Dowd 655 W Broadway, Suite 1900 San Diego, CA 92101

Counsel for Appellants

Michael B. Carlinsky Sanford I. Weisburst Jesse Bernstein Jacob J. Waldman Leigha Empson Quinn Emanuel Urquhart & Sullivan 295 5th Avenue, 9th Floor New York, NY 10016

Kevin H. Marino John D. Tortorella Marino Tortorella & BoylePage 3 437 Southern Boulevard Chatham, NJ 07928

Counsel for Appellees

CHAGARES, Chief Judge.

Maiden Holdings, Ltd. ('Maiden') is a reinsurance company whose common stock is publicly traded on the NASDAQ stock exchange. Over the course of roughly two years, adverse developments with Maiden's biggest client required Maiden to pay out claims in greater amounts than it had budgeted for, causing it to lose hundreds of millions of dollars while its stock price dropped more than 80%. Plaintiffappellants Boilermaker Blacksmith National Pension Trust and Taishin International Bank Co. Ltd. (collectively, “Boilermaker”), representing a class of Maiden common stock owners, filed a lawsuit claiming that Maiden committed securities fraud. Boilermaker asserted that Maiden’s announcements of the reserve funds it set aside to pay out future claims were misleading because Maiden omitted historical data suggesting that those reserves were deficient.

The District Court, after denying Boilermaker’s requests for discovery into the historical data Maiden had access to, granted summary judgment in favor of Maiden. The District Court held that Maiden’s reserve announcements were Page 4not misleading as a matter of law because (1) there was no dispute that Maiden knew of and considered the undisclosed historical data, and (2) the withheld data did not 'totally eclipse” other considerations that informed Maiden’s predicted losses. Joint Appendix ('J.A.') 18-19. Boilermaker now appeals the District Court’s discovery and summary judgment rulings.

We conclude that the District Court erred in granting summary judgment. In its decision in Omnicare, Inc. v. Laborers District Council Construction Industries Pension Fund, 575 U.S. 175, 188 (2015), the Supreme Court explained that a securities issuer’s statement of opinion is “misleadingly incomplete' and thus unlawful if the speaker omits known material facts about his “basis for holding that view.” Whether withheld information is material depends on its relative importance to the challenged opinion. Proving the materiality of one piece of data may therefore be difficult if the opinion was “based on a variety of complex assumptions and considerations.' City of Warren Police & Fire Ret. Sys. v. Prudential Fin., Inc., 70 F.4th 668, 684 (3d Cir. 2023). But materiality is an issue that 'always depends on context,' Omnicare, 575 U.S. at 190, and we conclude that the District Court misapplied this context-sensitive framework by holding Boilermaker to a higher standard of materiality than the law requires and denying it the opportunity to conduct discovery afforded by the Federal Rules of Civil Procedure. We will vacate the judgment of the District Court.

Page 5Reinsurance is the business of insuring insurance companies. Therefore, just like any other insurance company, reinsurers have to set aside funds to pay out future claims. These set-aside funds, known as 'loss reserves,' are the product of “an insurer’s actuarial judgment” and are generally calculated based on many factors. Prudential, 70 F.4th at 684. Because reserves represent predicted losses, they are effectively removed from an insurer’s operating income and treated as liabilities in financial reports. A company that sets its loss reserves too low effectively understates its liabilities, thus inflating its perceived financial strength.

Loss ratios are one input that actuaries often consider when setting reserves. These ratios are expressed as percentages that capture losses incurred from claims paid out to policyholders relative to revenue earned from premiums. Loss ratios are generally tracked by accident year ('AY'), which is the twelve-month period in which a claim was filed. Because it can take multiple years to pay out or settle any given claim, the loss ratio for an AY can change over time. For example, if, in the year 2025, an insurance company collects a $50 premium, receives a claim, and pays out $25 on that claim, then the loss ratio for AY 2025 would be 50%. If, two years later, the company must pay an additional $10 on that same claim, the loss ratio for AY 2025 would increase to 70%. Tracking historical loss ratios enables insurance companies to monitor trends of “adverse” or “favorable development[s]” and determine the ultimate cost of prior AYs. J.A. 3576. This historical data can help inform what loss ratio estimate (or “loss ratio pick”) should be used to set loss reserves.

Maiden, the reinsurer at the center of this dispute, had Page 6only two reportable operating segments.¹ The larger of the two was Maiden’s coverage of AmTrust Financial Services, Inc. ('AmTrust'), which grew to represent more than 70% of Maiden's net premiums earned. Pursuant to its reinsurance agreement, AmTrust ceded a portion of its premiums to Maiden, who in turn was obligated to compensate AmTrust for a portion of the claims paid by AmTrust to its customers. Maiden also paid AmTrust an additional 31% commission of the premiums Maiden received from AmTrust, which meant the reinsurance agreement would become unprofitable for Maiden if loss ratios for the AmTrust segment reached or exceeded roughly 69%.

Maiden employed a team of actuaries to estimate its loss reserve needs. This team analyzed many variables, such as historical data, actuarial and statistical projections, and potential economic, legislative, and social changes. Maiden provided details about its loss reserve process in publicly filed Securities Exchange Commission ('SEC') disclosure forms. Recognizing the inherent complexity of these predictive judgments, Maiden advised investors in these disclosures that loss reserves “do not represent an exact calculation of liability”Page 7 and that actual losses could deviate from Maiden’s estimates. J.A. 162. Maiden further stated that there is no 'precise method” to evaluate “the impact of any specific factor on the adequacy of reserves.' Id. But Maiden also advised its investors of its assumption 'that past experience . . . is an appropriate basis for predicting future events” and that “historic loss development and trend experience is assumed to be indicative of future loss development and trends.” J.A. 162, 199. Maiden repeatedly informed investors that this experience-driven premise was in fact one of the 'most significant assumptions used . . . to estimate the reserve for loss' because of the relative weight of historical data in Maiden's 'determination of initial expected loss ratios and expected loss reporting patterns.” J.A. 199.

Historical loss development for Maiden's AmTrust business indicated that losses increased over time. At the end of 2012, the loss ratios for AY 2008 through AY 2012 ranged from 48.6% to 71.7%. Over the next two years, those loss ratios shot up to 69.2% to 78.5%. Losses for these AYs continued increasing through year-end 2017, reaching a range of 75.1% to 82.2%. By that time, loss ratios for four of the five most recent AYs had already surpassed 60% as well.² Notwithstanding these historical trends, Maiden consistently relied on loss ratio picks between 50% and 60%, resulting in reserve amounts that were hundreds of millions of dollars less than the loss estimates recommended by Maiden’s actuaries. Across its SEC forms, earnings conference calls, and press releases between February 2014 and November 2018 (the “class period”), Maiden discussed its loss reserves, financial Page 8performance, and the profitability of its AmTrust segment. But Maiden did not disclose the several AYs with loss ratios exceeding the 69% profitability threshold when announcing loss reserves based on 50-to-60% loss ratio picks.

Maiden's loss ratio picks and corresponding loss reserves proved inadequate. From 2017 through 2018, Maiden experienced “significant adverse loss development within [its] AmTrust Reinsurance segment,' was forced to increase its deficient loss reserves, and lost hundreds of millions of dollars as a result. J.A. 2951; see J.A. 2950, 2902. After this period of “loss reserve strengthening and adverse prior year development of loss reserves,” Maiden could offer “no assurance that [it] w[ould] return to profitability.” J.A. 2902. The price of Maiden’s common stock plummeted from $16.50 per share in February 2017 to less than $2.50 per share in November 2018. Before the stock price dropped, Maiden executives collected millions of dollars by selling several thousands of shares of common stock at an average price range of $13.50 to $16.40 per share.

Boilermaker filed a lawsuit in the United States District Court for the District of New Jersey against Maiden and three of its executives. Boilermaker first claimed that Maiden and its executives violated section 10(b) of the Securities Exchange Act of 1934 (the 'Exchange Act'), 15 U.S.C. § 78j(b), and SEC Rule 10b-5, 17 C.F.R. § 240.10b-5, by making unlawfully misleading statements about its loss reserves. Boilermaker alleged that Maiden’s reserve announcements were misleading due to the omission of material information regarding adverse developments in the AmTrust book of business, of which Page 9Maiden knew, and which contradicted Maiden’s “artificially low loss ratio assumptions.' J.A. 3760. Boilermaker also claimed that Maiden and its executives were jointly and severally liable under section 20(a) of the Exchange Act, 15 U.S.C. § 78t(a).

Maiden moved to dismiss Boilermaker's operative amended complaint, and the District Court denied the motion in part. The District Court instructed Boilermaker to file a second amended complaint to 'eliminate[] surplusage,' but otherwise held that discovery was necessary to test Boilermaker's theory of liability that Maiden 'should have disclosed the historical loss ratios of the AmTrust book of business” because this information constituted “material, adverse historical data of which [Maiden] had actual knowledge.' J.A. 85, 97. The District Court then ordered “limited discovery” that would be followed by “[d]ispositive motions.' J.A. 3549-50. The District Court ruled that the discovery during this period would be limited to Boilermaker’s 'sole claim against Maiden,' focusing on the following question: 'was there an intentional decision made by the Defendants to omit AmTrust’s historical loss ratio information from the view of investors?” J.A. 97–98.

Boilermaker requested that Maiden produce documents to confirm whether its historical AmTrust loss ratios reached the 70% to 80% figures alleged and whether Maiden knew or had access to this information when it used loss ratio picks of 50% to 60% to set loss reserves. After Maiden disclosed some of the actuarial analyses relevant to its loss reserve calculations, Boilermaker argued to the Magistrate Judge that Maiden had not produced the documents necessary “to identify the historical loss ratios the defendant[s] had at the time they Page 10made their statements or the underlying data that would allow [Boilermaker] to calculate those.” J.A. 45. Maiden objected to Boilermaker’s requests as beyond the scope of discovery. In Maiden’s view, the District Court had limited discovery to “a single, simple question, whether [Maiden] considered historical loss ratios as part of [its] analysis.' J.A. 59. Boilermaker resisted that characterization of discovery, highlighting that Boilermaker itself had alleged from the start of the litigation “that defendants knew and had access to the historical data.” J.A. 65. The relevant questions that remained, according to Boilermaker, were whether the historical loss ratios were as severe as alleged and whether that data was available to Maiden during the class period.

The Magistrate Judge did not grant Boilermaker the discovery it sought. She concluded that Maiden’s disclosure of at least some historical loss information was satisfactory because it “indicate[d] that [Maiden] did consider this information.” J.A. 66. In response to Boilermaker’s request for the “underlying data” related to Maiden’s calculation and consideration of historical loss ratios, the Magistrate Judge emphasized that there was no need 'to get into an entire discovery process' because the District Court did not order “typical discovery under the rules.” J.A. 3497, 3501. This first phase of discovery was “simply for the purpose of determining whether . . . the defendants' [renewed] motion to dismiss should be granted.' J.A. 3501-02. Thus, according to the Magistrate Judge, Boilermaker’s “request to receive all the loss ratio [data amounted] to second-guessing the conclusions made by [Maiden]' and contravened the District Court's limited discovery order. J.A. 66.

Boilermaker appealed the Magistrate Judge’s ruling to Page 11the District Court, and the District Court denied the appeal as untimely under Federal Rule of Civil Procedure 72(a). Boilermaker filed a motion for reconsideration, which the District Court denied as well. The District Court reaffirmed its conclusion that the appeal was untimely and held that, even if the appeal were timely, there was “no clear error or an abuse of discretion” by the Magistrate Judge. J.A. 30.

At the close of limited discovery, Maiden moved for dismissal or, in the alternative, summary judgment, and Boilermaker filed a motion under Federal Rule of Civil Procedure 56(d) to deny or defer summary judgment pending further discovery. The District Court denied Boilermaker's Rule 56(d) motion and granted summary judgment to Maiden. The District Court concluded that Maiden's omission of historical data was immaterial because the record did not support “Plaintiffs’ allegation that Maiden did not consider any of this material in setting loss reserves.” J.A. 18. It further explained that the historical loss ratios were “only one of many factors' and would not 'totally 'eclipse the balance of the numerous other considerations used to set reserves’ if revealed to investors.” J.A. 19 (quoting Prudential, 70 F.4th at 687). The District Court therefore held that Maiden's statements were not misleading as a matter of law, which defeated Boilermaker’s section 10(b) and Rule 10b-5 claim as well as its “derivative” section 20(a) claim. J.A. 20.

Boilermaker timely appealed.

Page 12We review the District Court’s order granting summary judgment to Maiden and its executives de novo and “apply the same test the District Court should have used.” In re Processed Egg Prods. Antitrust Litig., 881 F.3d 262, 267-68 (3d Cir. 2018) (quoting Howard Hess Dental Lab’ys Inc. v. Dentsply Int’l, Inc., 602 F.3d 237, 246 (3d Cir. 2010)). We evaluate the record in the light most favorable to Boilermaker and draw all inferences in its favor. Id. at 268. We do not 'weigh the evidence or make credibility determinations' because those tasks are properly left to the factfinder. Pichler v. UNITE, 542 F.3d 380, 386 (3d Cir. 2008). And we “may affirm based on any ground supported by the record.” Watters v. Bd. of Sch. Dirs. of City of Scranton, 975 F.3d 406, 412 (3d Cir. 2020) (quotation marks omitted).

Boilermaker argues that several of the District Court’s orders were the product of one legal error: misapplication of the legal framework governing Boilermaker’s misleading-byomission theory of liability. Boilermaker claims that this error taints the District Court’s limitations on the scope of discovery, the denial of Boilermaker’s Rule 56(d) motion, and the entry of summary judgment for Maiden on all claims. Maiden counters that the District Court properly applied the Supreme Court's and this Court's securities fraud precedent, that Boilermaker failed to preserve its discovery objections, and Page 13that the record contains no triable issue of fact as to materiality or scienter. We will begin by clarifying the law governing Boilermaker’s misleading-by-omission claim and then address the merits of summary judgment on the current record.

Section 10(b) of the Exchange Act provides that securities issuers may not use “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 regulation known as Rule 10b-5, in turn, provides that it is unlawful for an issuer “[t]o make any untrue statement of a material fact or to omit to state a material fact necessary in order to make statements made, in light of the circumstances under which they were made, not misleading.' 17 C.F.R. § 240.10b-5(b). In other words, Rule 10b-5 “requires disclosure of information necessary to ensure that statements . . . are clear and complete.” Macquarie Infrastructure Corp. v. Moab Partners, L.P., 601 U.S. 257, 264 (2024).

While statements of opinion 'convey some lack of certainty as to the statement’s content,” they can still mislead investors if the speaker leaves out key information. Omnicare, 575 U.S. at 187.⁴ '[A] reasonable investor may, depending on Page 14the circumstances, understand an opinion statement to convey facts about how the speaker has formed the opinion . . . .” Id. at 188. Thus, “if the real facts are otherwise, but not provided, the opinion statement will mislead its audience.' Id. The Supreme Court identified two ways in which an opinion statement may mislead investors “about the speaker’s basis for holding that view.' Id. First, an opinion statement may mislead investors about what the speaker did by “omit[ting] material facts about the issuer’s inquiry into” the facts relevant to the opinion (“inquiry theory”). Id. at 189. The statement, '[w]e believe our conduct is lawful,' for example, may be misleading if the speaker concealed that he did not consult an attorney before forming the opinion. Id. at 188. Second, an opinion statement may mislead investors about what the speaker knew by “omit[ting] material facts about the issuer's . . . knowledge' of the evidence for and against the opinion (“knowledge theory”). Id. at 189. The same statement, '[w]e believe our conduct is lawful,' may therefore be misleading if the speaker withheld the fact that his own lawyers told him otherwise. Id.

This Court recently applied Omnicare’s misleading-byomission theory of liability to claims of misleading loss reserve statements by an insurance company. In City of Warren Police & Fire Retirement System v. Prudential Financial, Inc., we Page 15held that when loss reserves are challenged as predictive statements about the amount of funds needed for future claims, 'the stated reserve amount, as a manifestation of actuarial judgment, functions as an opinion.' 70 F.4th at 684. The plaintiff in Prudential alleged that an insurance company’s loss reserve statement was misleading because it omitted data about increased mortality (and thus increased claims) in one segment of the company's life insurance portfolio. To determine whether the plaintiff stated a claim of securities fraud, the Court analyzed whether the factual allegations plausibly established that the omitted data was material. We answered in the negative.

We determined in Prudential that the plaintiff failed to allege necessary facts about the omitted data’s importance to the setting of loss reserves. Because loss reserves are “based on a variety of complex assumptions and considerations,” id. (citing Shapiro v. UJB Fin. Corp., 964 F.2d 272, 281 (3d Cir. 1992)), a reasonable investor would understand that an inherently predictive loss reserve announcement “rest[s] on a weighing of competing facts,” id. at 687 (quoting Omnicare, 575 U.S. at 190). The failure to disclose “some fact cutting the other way' does not 'necessarily' render that loss reserve announcement misleading. Id. (quoting Omnicare, 575 U.S. at 189). We deemed the complaint in Prudential deficient because it alleged that one segment within a larger insurance portfolio 'had a consistently negative mortality experience' but alleged nothing about the significance of that adverse data relative to the “many factors Prudential considered in setting its reserves.” Id. at 686–87. The Court acknowledged that “the alleged negative mortality in the [segment] would tend to increase the amount of needed reserves,” but the Court could not identify any 'accompanying allegation that the negative Page 16mortality . . . was so great that it would, for a reasonable investor, eclipse the balance of the numerous other considerations used to set reserves' for the entire insurance portfolio. Id. at 687. We therefore concluded that the plaintiff had failed to plausibly allege that the omission was material.

We did not hold in Prudential, however, that omitting adverse historical data from a loss reserve statement is never misleading. The plaintiff in Prudential did not plausibly allege materiality because it provided no information on how the omitted data compared to other considerations. The complaint contained no 'allegations about the relative size' of the struggling segment nor “the magnitude of [its] problems.” Id. at 689. But our Prudential decision does not stand for the proposition that insurance companies may withhold all adverse data from the view of investors when announcing reserves simply because reserve determinations are complex. Reasonable investors do “not expect that every fact known to an issuer supports its opinion statement,” but they assume that the material ones do. Omnicare, 575 U.S. at 190 (emphasis omitted). Separating the material from the immaterial is a context-specific task, id., and our ruling in Prudential did not create a context-blind rule governing all claims of misleading reserve statements.

Setting aside (for now) the question of whether discovery was adequate, we conclude that there are genuine, material factual disputes in the current record that provide grounds to vacate the entry of summary judgment. The District Court granted summary judgment after holding that there was no genuine dispute of material fact on one element of Page 17Boilermaker’s section 10(b) and Rule 10b-5 claim: whether Maiden “made a misstatement or omission of a material fact” that rendered its loss reserve statements misleading. Semerenko v. Cendant Corp., 223 F.3d 165, 174 (3d Cir. 2000). We thus focus our de novo review of the record on whether Boilermaker “adduce[d] more than a mere scintilla of evidence in its favor” on this element. Williams v. Borough of West Chester, 891 F.2d 458, 460 (3d Cir. 1989). We conclude that it did.

The District Court held that Maiden’s statements were not misleading as a matter of law after making two findings. First, it found that 'the record indisputably shows Maiden engaged in a complex actuarial process that considered historical losses.” J.A. 18. Second, it found nothing “in the record to suggest that the revelation of [higher] historical loss ratios would totally ‘eclipse the balance of the numerous other considerations used to set reserves’ if revealed to investors.” J.A. 18 (quoting Prudential, 70 F.4th at 687). These statements do not fully address the claim Boilermaker was asserting or the legal standard applicable to that claim.

Maiden's mere consideration of historical loss ratios does not entitle it to judgment as a matter of law because Boilermaker relies on a knowledge theory, not an inquiry theory, in asserting its claim that Maiden's statements were misleading by omission. To understand why, consider once more the hypothetical opinion statement, '[w]e believe our conduct is lawful.” Omnicare, 575 U.S. at 188. If a plaintiff claimed that this statement was misleading because the speaker did not disclose that his legal counsel told him that the conduct was illegal, it would be no defense for the speaker to respond, “well, I considered the advice.” Rather, if the speaker knew of Page 18the advice, whether his statement is misleading depends on how significant that known contrary advice was. The answer to that question will “always depend[] on context.” Omnicare, 575 U.S. at 190. On the one hand, if the speaker’s full team of in-house attorneys told him the conduct was illegal, but the speaker did not reveal that, investors would have 'cause to complain.” Id. at 188. In contrast, if a “single junior attorney expressed doubts . . . when six of his more senior colleagues gave a stamp of approval,” the failure to disclose the junior attorney's comments 'would not make the statement of opinion misleading.” Id. at 190.

Boilermaker’s theory of securities fraud was that Maiden unlawfully omitted known, materially adverse historical loss ratios that conflicted with Maiden's loss ratio estimates and loss reserve statements as reported in SEC disclosure forms. Contrary to the District Court’s suggestion, Boilermaker never alleged that “Maiden did not consider any of this material in setting loss reserves.” J.A. 12. Boilermaker alleged the opposite, asserting that Maiden considered but concealed its historical loss ratio data. See J.A. 3581 (“Defendants knew of or recklessly disregarded data showing a pattern [of] incurred losses . . . [that] materially conflicted with the loss reserves Defendants reported.'). There is no dispute that Maiden had access to historical loss data that it omitted from its statements about loss reserves.

The critical question is whether the omitted historical loss data was material. Reading the record in the light most favorable to Boilermaker, a reasonable factfinder could find that it was. The Court appreciates that Maiden's actuarial process for loss reserve calculations was complex. The voluminous record in this case is a testament to that Page 19complexity. But there is sufficient evidence in the summary judgment record to permit a reasonable factfinder to conclude that the omitted historical loss data was indeed material. See In re Ikon Off. Sols., Inc., 277 F.3d 658, 666 (3d Cir. 2002). Three straightforward findings drawn from the record would support that conclusion.

First, the record contains evidence that negative developments in Maiden’s AmTrust segment would have an outsized impact on Maiden’s entire business. Maiden repeatedly emphasized how important AmTrust was to Maiden. In every SEC 10-K Form filed for 2013 through 2017, Maiden reminded investors that “AmTrust is Maiden’s largest client relationship' and that Maiden was 'dependent . . . on AmTrust . . . for a substantial portion of [Maiden’s] business.” J.A. 178 (2013); J.A. 532 (2014); J.A. 856-57 (2015); J.A. 1183 (2016); J.A. 1534 (2017); see also J.A. 2892 (2018 10-K form similarly providing that “AmTrust is [Maiden’s] largest client”). When Maiden’s “total capital resources decreased by $677.9 million, or 45.4%' from 2017 to 2018, Maiden identified only two causes: poor investment performance and “significant adverse loss development within [Maiden’s] AmTrust Reinsurance segment.” J.A. 2950–51. Even setting all that aside, a factfinder could still reasonably find that losses from Maiden’s business with AmTrust could pose a serious risk of material harm to Maiden because it was undisputed that AmTrust accounted for more than 70% of Maiden's net premiums earned for much of the class period.

Second, the record contains evidence that Maiden had access to historical loss data for AmTrust that was inconsistent with Maiden's loss ratio picks. Boilermaker derived the alleged loss ratios from Maiden's own financial documents, Page 20see J.A. 2140-41, record evidence suggests that this loss information would have been regularly “reported to” Maiden, e.g., J.A. 551, and Maiden even conceded that the alleged loss ratios were 'approximately accurate.' Those loss ratios indicate that Maiden’s losses with AmTrust increased consistently and substantially over several years. Most prior AYs exceeded 70% losses by year-end 2014. By year-end 2017, the average loss ratio for all prior AYs was 71.1%, and multiple AYs even surpassed 80% losses. Yet Maiden set and reported loss reserves based on much lower loss ratio picks ranging from 50% to 60%. In qualitative terms, the record suggests that Maiden possessed data showing that its largest segment became increasingly unprofitable year after year, but Maiden informed investors it expected continued profits without disclosing the adverse historical data suggesting otherwise.

Third, the record contains evidence that the negative historical data was (or should have been) an important part of Maiden's loss reserve estimation process. Maiden told investors time and again that “historic loss development . . . is assumed to be indicative of future loss development and trends.” J.A. 199, 550, 875, 1203. Along those lines, Maiden assured investors that it would “establish or adjust reserves . . . based upon loss data received from the ceding companies with which [it] do[es] business, including AmTrust.” E.g., J.A. 517. Granted, a factfinder might have to balance those statements against Maiden’s disclaimers to investors that “[a]ctual results could materially differ from [its] estimates,” e.g., J.A. 598, or that there is no “precise method . . . for evaluating the impact of any specific factor on the adequacy of reserves,” e.g., J.A. 517. But nothing in the record as it stands — particularly when read in the light most favorable to Boilermaker - would Page 21preclude a reasonable factfinder from crediting Maiden's repeated statements that historical data was one of the “most significant” components of its loss reserve calculation process. E.g., J.A. 550.

Viewed in isolation, any one of these three considerations would not establish that the omitted data was material to Maiden’s predicted losses. The relative importance of AmTrust's business to Maiden matters only insofar as Boilermaker can show the degree of AmTrust’s underperformance. And, even still, AmTrust's historical underperformance matters only insofar as Boilermaker can show that historical trends were a significant part of Maiden’s complex, multifactor loss-reserve process. If, for example, there were no dispute that historical data was a miniscule consideration in loss-reserve calculations, always overshadowed by other factors, then no reasonable investor would expect such data to be disclosed every time historical losses diverged from predicted losses. See Omnicare, 575 U.S. at 190 (“A reasonable investor does not expect that every fact known to an issuer supports its opinion statement.”).

Yet we do not deal with these factors in isolation: viewed holistically, the evidence in the current record provides the full “context” necessary to raise a genuine issue of material fact as to whether the omission of AmTrust's historical loss data was material. Id. There is evidence that Maiden's business depended on AmTrust and that historical trends were one of the most significant considerations when Maiden set reserves. Maiden stated both points expressly in its communications to investors, and a reasonable factfinder could thus conclude that investors would expect Maiden’s predicted losses to align with historical trends in its AmTrust business.

Page 22Add to this the evidence that Maiden's predicted losses diverged by roughly 10-20% from its historical losses with AmTrust — including several, increasingly unprofitable AYs - and that factfinder could fairly conclude that Maiden's omission of that discrepancy when announcing its optimistic predictions would mislead investors.

Maiden argues that the District Court was correct to hold that the omitted data was immaterial because it did not “totally ‘eclipse the balance of the numerous other considerations used to set reserves.'' J.A. 19 (quoting Prudential, 70 F.4th at 687). But Maiden's proposed 'total eclipse' rule, even if based on a (modified) phrase in our Prudential opinion, suggests a heightened standard of materiality that our Prudential decision cannot support. We have long held that materiality questions demand 'delicate assessments of the inferences a reasonable shareholder would draw from a given set of facts [that] are peculiarly for the trier of fact.” Shapiro, 964 F.2d at 280 n.11 (citing TSC Indus., Inc. v. Northway, Inc., 426 U.S. 438, 450 (1976)). That approach is consistent with the Supreme Court’s description of materiality as a “fact-specific inquiry” into whether there is a “substantial likelihood that the disclosure of the omitted fact would have been viewed by the reasonable investor as having significantly altered the 'total mix' of information made available.” Basic Inc. v. Levinson, 485 U.S. 224, 231–32, 240 (1988) (quoting TSC, 426 U.S. at 449). The Supreme Court’s Omnicare decision did not change this materiality inquiry, see 575 U.S. at 190, and we have reaffirmed our longstanding, context-based materiality standard since Omnicare was decided, see SEC v. Chappell, 107 F.4th 114, 130 n.24 (3d Cir. 2024) (quoting Shapiro, 964 F.2d at 280 n.11).

Page 23The complexity of Maiden’s loss reserve determinations did not give Maiden free rein to omit any “known contradictory evidence” so long as it did not hide the single most important evidence. Omnicare, 575 U.S. at 189 n.6. Maiden’s omission of historical loss data is material so long as it is “significant” enough that it would, if revealed to a reasonable investor, 'alter[] the 'total mix' of information' presented. Basic, Inc., 485 U.S. at 232. The total mix of information in the record here leads us to conclude that, at a minimum, “reasonable minds [could] differ on the question of materiality.' TSC, 426 U.S. at 450 (quoting Johns Hopkins Univ. v. Hutton, 422 F.2d 1124, 1128 (4th Cir. 1970)).

We disagree with Maiden’s assertion that our Prudential case involved “circumstances nearly identical to those here.” Maiden Br. 20. The summary judgment record in this case is a far cry from the vague factual assertions at issue in Prudential. The plaintiff in Prudential alleged that one segment of a larger insurance portfolio experienced fourteen months of “consistently negative mortality” but offered no description of the relative size of the segment nor the magnitude of the negative trends therein. 70 F.4th at 686; see id. 687, 689. In contrast, the record here contains various calculations, charts, and statements capturing Maiden’s many years of substantial and growing losses with AmTrust, Maiden’s dependence on AmTrust as its largest client, and the importance of historical trends to Maiden’s loss reserve process.

Maiden also claims, again invoking this Court’s Prudential ruling, that “a purported disparity between a single input and the ultimate loss-reserve estimate, without the single input having been disclosed, does not render the loss-reserve estimate misleading.” Maiden Br. 24. But no such categorical Page 24rule exists. We held in Prudential that these kinds of omissions are “not necessarily misleading.” 70 F.4th at 687 (emphasis added) (quoting Omnicare, 575 U.S. at 189). The omission of known contradictory data is not per se immaterial just because a statement of opinion was informed by multiple considerations, and we decline to adopt Maiden's proposed context-blind rule, which is inconsistent with our precedent and that of the Supreme Court.

In the alternative, Maiden asks this Court to affirm on the grounds that Boilermaker raised no genuine issue of material fact as to scienter — an element the District Court did not analyze. Scienter for purposes of Rule 10b-5 liability consists of 'a mental state embracing intent to deceive, manipulate, or defraud.” Tellabs, Inc. v. Makor Issues & Rts., Ltd., 551 U.S. 308, 319 (2007) (quoting Ernst & Ernst v. Hochfelder, 425 U.S. 185, 193 n.12 (1976)). Evidence of scienter may include knowledge of the underlying facts relevant to the false or misleading statement, Belmont v. MB Inv. Partners, Inc., 708 F.3d 470, 493 (3d Cir. 2013), unusual stock selling activity, Institutional Invs. Grp. v. Avaya, Inc., 564 F.3d 242, 279 (3d Cir. 2009), or suspicious “departure[s] of corporate executive defendants,” In re Hertz Glob. Holdings Inc., 905 F.3d 106, 118 (3d Cir. 2018). Yet “determination of whether a party acted with scienter, intertwined as it may be with an assessment of witness credibility, often cannot be undertaken appropriately on summary judgment proceedings.” Ikon Off. Sols., 277 F.3d at 668.

The Court declines Maiden's invitation to affirm on these alternative grounds. For one, Maiden's argument sits awkwardly alongside its insistence that 'Defendants proved they did consider historical loss information,” Maiden Br. 40, Page 25which implies knowledge of adverse data that Maiden did not disclose to investors. Most importantly, we are also mindful that, like the Supreme Court, we are “a court of review, not of first view.” Cutter v. Wilkinson, 544 U.S. 709, 718 n.7 (2005). While we may affirm on grounds not reached by the District Court, Watters, 975 F.3d at 412, we also possess the authority to remand for the District Court to resolve unaddressed issues in the first instance, see Howze v. Jones & Laughlin Steel Corp., 750 F.2d 1208, 1213 (3d Cir. 1984). We find it prudent to do so here.

At bottom, the Court holds that genuine issues of material fact require this Court to vacate the entry of summary judgment on Boilermaker's section 10(b) and Rule 10b-5 claim. Reviewing the record in the light most favorable to Boilermaker, the Court concludes that a reasonable factfinder could find that Maiden’s reserve announcements were misleading. The Court does not rule on whether the current record presents genuine issues of material fact as to scienter.

C.

The viability of Boilermaker's section 20(a) claim depends on whether there was 'an underlying violation of Section 10(b).” Rahman v. Kid Brands, Inc., 736 F.3d 237, 247 (3d Cir. 2013) (quoting Avaya, 564 F.3d at 252). The District Court granted summary judgment to Maiden on the section 20(a) claim because it granted summary judgment to Maiden on the section 10(b) and Rule 10b-5 claim. Maiden defends the District Court’s ruling on the same basis and offers no alternative grounds to affirm. Because the grant of summary judgment on the section 10(b) and Rule 10b-5 claim Page 26was improper, we hold that the grant of summary judgment on the section 20(a) claim was improper as well.

Having concluded that the summary judgment ruling must be vacated, we now address the parties’ separate dispute as to whether discovery was complete when the District Court granted summary judgment to Maiden. We hold that additional discovery is necessary upon remand.

The District Court concluded that Boilermaker was not entitled to additional discovery because Boilermaker’s earlier requests for similar evidence were denied by the Magistrate Judge. But those prior discovery limitations arose in a distinct context. The Magistrate Judge communicated multiple times that she limited discovery based on the “understanding” that the sole purpose of this limited discovery was for Maiden to have “the opportunity to renew its motion to dismiss.” J.A. 3376; see also J.A. 3501–02. It was on those grounds that the Magistrate Judge ruled, and the District Court affirmed, that 'typical discovery under the rules' was not required at this initial stage. J.A. 3501. This characterization of discovery as atypical is consistent with the Magistrate Judge advising Maiden that she was “allowing [only] a motion to dismiss” at the conclusion of limited discovery. J.A. 3378.⁵

Whatever limitations on discovery were imposed for purposes of Maiden’s renewed motion to dismiss, Boilermaker Page 27should have received 'typical discovery under the rules' before facing an adverse summary judgment ruling. J.A. 3501. That discovery has yet to occur. For example, Boilermaker sought (and continues to seek) 'complete documentation identifying what the historical loss ratios were for the AmTrust business for each of the relevant fiscal quarters.” J.A. 3255. Such evidence may be relevant to Boilermaker’s claim but was excluded as beyond the scope of the limited discovery phase set by the Magistrate Judge and District Court. The litigation has moved past this initial stage and now should proceed to typical discovery under the Federal Rules of Civil Procedure.

In sum, because discovery is incomplete, the Court will remand with instructions to permit full discovery. The Court trusts the District Court and Magistrate Judge to exercise their sound discretion to manage discovery in a manner consistent with the legal principles set forth in this opinion.

For the foregoing reasons, we will vacate the judgment of the District Court and remand for further proceedings consistent with this opinion.