Todd Fener and other plaintiffs filed a class action against Belo Corporation and some of its officers (collectively “Belo”) alleging violations under the Securities Exchange Act of 1934. Fener moved for class certification, which the district court denied. He appeals, and we affirm.
I.
Belo is a media company that owns television stations, websites, and newspapers, including the Dallas Morning News (“DMN”). Revenue from the DMN makes up about 60% of Belo’s publishing revenue and 30% of its total revenue; 90% of the DMN’s revenue comes from advertising sales, which are priced based on circulation.
The plaintiffs allege that Belo engaged in a fraudulent scheme designed to inflate DMN’s circulation artificially in the face of a nationwide downward trend in newspaper circulation. Belo allegedly paid bonuses for achieving circulation targets, rigged audits of DMN’s circulation, and imple*405merited a no-return policy that eliminated any incentive for distributors to return unsold newspapers. Those actions, the plaintiffs claim, allegedly artificially increased recorded circulation, which led to higher advertising revenues for DMN and larger profits for Belo.
On March 9, 2004, Belo announced that DMN’s future circulation would be down 2.5% on daily papers and 3.5% on the Sunday paper. On August 5, after the New York Stock Exchange (“NYSE”) closed, Belo issued a press release (“the press release”) that admitted that an internal investigation had revealed questionable circulation practices.
According to the press release, the allegedly fraudulent practices resulted in a 1.5% daily paper circulation decline and a 5% Sunday decline. The press release noted that the declines were “coupled with” the circulation declines announced in March and with lower anticipated circulation for the next six months; the total circulation decline from all of these announcements was predicted to be 5% for the daily paper and 11.5% for Sunday. The press release also stated that Belo would begin exercising more stringent control over possible improper manipulation of circulation.
When the NYSE opened the next day, Belo’s stock, which had closed the previous day at $23.21, dropped to as low as $18.00. It finished the day at $21.55, down $1.66 from the previous day’s close. Several securities analysts lowered their earning estimates for Belo and downgraded its stock.
On August 16, Belo announced that it would compensate advertisers by approximately $23 million, with an additional $3 million to cover costs related to an ongoing internal investigation. On September 29, Belo revised its initial circulation figures, projecting a decrease of 5.1% in daily circulation and 11.9% for Sunday. It said that most of the declines were related to the overstatements.
Plaintiffs1 sued on behalf of those who held Belo’s common stock between May 12, 2003, and August 6, 2004, alleging that Belo and five of its senior officers and directors had violated § 10(b) of the Securities Exchange Act of 1934, 15 U.S.C. § 78j(b), and SEC Rule 10b-5, 17 C.F.R. § 240.10b-5. They claimed the class members had bought Belo stock when its price was artificially (and fraudulently) inflated as a result of the manipulation of DMN’s reported circulation and were injured when Belo revealed the fraud and its stock price fell. Plaintiffs eventually2 moved for class certification under Federal Rule of Civil Procedure 23.
Belo opposed class certification and presented expert Dr. Paul Gompers, who testified that class certification was inappropriate because plaintiffs could not show that the fraudulent disclosure in the press release was the primary cause of the stock price decline. Plaintiffs responded with a declaration from expert Dr. Scott Hakala, who rejected Gompers’s testimony and stated that the decline was “entirely or almost entirely attributable to the revelation of the relevant truth in this case.”3 *406After hearing from the experts and examining the other evidence, the district court denied class certification.
II.
“We review class certification decisions for abuse of discretion in recognition of the essentially factual basis of the certification inquiry and of the district court’s inherent power to manage and control pending litigation. Whether the district court applied the correct legal standard ..., however, is a legal question that we review de novo.” Regents of the Univ. of Cal. v. Credit Suisse First Boston (USA), Inc., 482 F.3d 372, 380 (5th Cir. 2007) (citation, internal quotation marks, and ellipses omitted). ‘Where a district court premises its legal analysis on an erroneous understanding of governing law, it has abused its discretion.” Id. (citing Unger v. Amedisys Inc., 401 F.3d 316, 320 (5th Cir.2005)). “A district court must conduct a rigorous analysis of the rule 23 prerequisites before certifying a class .... The party seeking certification bears the burden of proof.” Castano v. Am. Tobacco Co., 84 F.3d 734, 740 (5th Cir.1996).
“[C]lass certification creates insurmountable pressure on defendants to settle, whereas individual trials would not.” Id. at 746 (citation omitted). “The risk of facing an all-or-nothing verdict presents too high a risk, even when the probability of an adverse judgment is low.” Id. (citation omitted). This risk is particularly high in securities-fraud class actions, in which the current “class-based compensatory damages regime in theory imposes remedies that are so catastrophically large that defendants are unwilling to go to trial even if they believe the chance of being found liable is small.” Janet Cooper Alexander, Rethinking Damages in Securities Class Actions, 48 Stan. L.Rev. 1487, 1511 (1996). Some have observed that seeking class certification to force favorable settlements does not benefit small investors4 but instead resembles a shakedown5 or “judicial blackmail.”6
III.
In securities fraud cases, the plaintiff must prove
(1) a material misrepresentation (or omission); (2) scienter, i.e., a wrongful state of mind; (3) a connection with the purchase or sale of a security; (4) reliance, often referred to in cases involving public securities markets (fraud-on-the-market cases) as “transaction causation”; (5) economic loss; and (6) “loss causation,” i.e., a causal connection between the material misrepresentation and the loss.
Dura Pharms., Inc. v. Broudo, 544 U.S. 336, 341-42, 125 S.Ct. 1627, 161 L.Ed.2d 577 (2005) (citations omitted). In Basic v. Levinson, 485 U.S. 224, 108 S.Ct. 978, 99 L.Ed.2d 194 (1988), the Court held that requiring proof of “actual reliance” was unduly burdensome to plaintiffs, and instead it “recognized the securities fraud theory of fraud-on-the-market.” Greenberg v. Crossroads Sys., Inc., 364 F.3d 657, 661 (5th Cir.2004) (citation omitted). Under that theory, plaintiffs can create a rebuttable presumption of reliance if they *407can show “(1) the defendant made public material misrepresentations, (2) the defendant’s shares were traded in an efficient market, and (3) the plaintiffs traded shares between the time the misrepresentations were made and the time the truth was revealed.” Id. (citation omitted). When proving reliance, “plaintiffs cannot trigger the presumption ... by simply offering evidence of any decrease in price following the release of negative information .... [They instead] must show that the false statement causing the increase was related to the statement causing the decrease.” Id. at 665. This last requirement for fraud-on-the-market reliance is known as loss causation, a concept that is at the heart of the instant ease.7
Proving loss causation “require[s] a plaintiff additionally to ‘prove that the defendant’s non-disclosure materially affected the market price of the security.’ ” Alaska Elec. Pension Fund v. Flowserve Corp., 572 F.3d 221, 227 (5th Cir.2009) (per curiam) (citing Nathenson v. Zonagen Inc., 267 F.3d 400, 414 (5th Cir. 2001)). A plaintiff must show “(1) that the negative truthful information causing the decrease in price is related to an allegedly false, non-confirmatory positive statement made earlier and (2) that it was more probable than not that it was this negative statement, and not other unrelated negative statements, that caused a significant amount of the decline.” 572 F.3d at 227 (internal quotation marks and brackets omitted) (citing Greenberg, 364 F.3d at 666). “[L]oss causation as an issue of predominance must be established at the class certification stage by a preponderance of all admissible evidence.” 572 F.3d at 228 (citations omitted).
A court can examine loss causation at the pleadings stage,8 the class certification stage,9 on summary judgment,10 or at trial. The proof needed for loss causation at the pleadings stage should not be conflated with the requirements needed at the class certification stage.11 We must examine whether these plaintiffs have presented enough information to show loss causation under rule 23.
*408IY.
A.
As a threshold issue, we address plaintiffs’ argument that Stoneridge Investment Partners, LLC v. Scientific-Atlanta, 552 U.S. 148, 128 S.Ct. 761, 169 L.Ed.2d 627 (2008), overruled some of our earlier opinions, specifically Greenberg and Oscar Private Equity Investments v. Allegiance Telecom, Inc., 487 F.3d 261 (5th Cir.2007). Plaintiffs cannot direct us to any mention of either case in Stoneridge or to any discussion of this circuit’s fraud-on-the-market theory. They point only to Stoneridge’s general summary of the fraud-on-the-market theory, urging that that discussion somehow indicates that the Supreme Court meant to strike down our recent securities fraud caselaw.
Plaintiffs are mistaken. First, when the Supreme Court discusses a general legal standard and cites its earlier caselaw on point, it does not necessarily overrule intervening decisions of the lower courts. Moreover, Basic “allows each of the circuits room to develop its own fraud-on-the-market rules.” Oscar, 487 F.3d at 264 (quoting Abell v. Potomac Ins. Co., 858 F.2d 1104, 1117-18 (5th Cir.1988), vacated on other grounds, 492 U.S. 914, 109 S.Ct. 3236, 106 L.Ed.2d 584 (1989)). In Green-berg and Oscar, we used this “room” to develop specific rules. Stoneridge and its general discussion of the fraud-on-the-market theory do not affect those rules.
B.
Recognizing that Stoneridge does not change our fraud-on-the-market analysis, we turn to whether plaintiffs met our requirements for proving loss causation at the class certification stage. Plaintiffs filed their certification motion with about one hundred pages of support.12 The relevant information consisted of excerpts from Belo’s SEC Form 10-K for two years, Belo’s historical stock prices, Belo’s SEC S-3 forms from 1996 to 2006, financial data from Yahoo! finance, and a chart of Belo’s daily share price. Plaintiffs submitted no expert testimony.
In response, Belo presented Gompers’s testimony and an event study that he had conducted. Gompers argued that the press release contained not one piece of information but three separate items of news: DMN’s circulation decrease resulted from (1) fraudulent overstatements;13 (2) changes in DMN’s methodology; and (3) industry-wide decline in newspaper circulation.14 Gompers’s event study examined 132 analyst reports and found that the stock price decline was primarily related to the non-fraudulent disclosures instead of the fraudulent one.
Plaintiffs responded with Hakala’s testimony and event study. That study concluded that Belo’s stock was efficiently traded during the class period, that revenue at DMN was closely tied to circulation, that Belo’s stock moved in close connection with the industry’s other stocks, and that its share price had moved significantly when it issued the press release. More importantly, he disputed Gompers’s conclusion that the press release could be separated into three parts, claiming instead that the fraudulent and non-fraudulent parts of the press release had to be examined together as one disclosure. Fi*409nally, he alleged that the market had already absorbed the non-fraudulent disclosures’ information, and thus only the fraudulent disclosure affected Belo’s stock price.15
Whether we view the press release as one complete disclosure or three separate ones is important. In Oscar, 487 F.3d at 266, we specified how parties must prove loss causation in cases in which “multiple items of negative information were released together with the corrective disclosure.” “In such multi-layered loss-causation inquiries,” we require that the plaintiff shows that “it is more probable than not that it was this negative statement, and not other unrelated negative statements, that caused a significant amount of the decline.”16 Id. (citing Greenberg, 364 F.3d at 666). Thus, if we view the press release as multiple pieces of information, plaintiffs must prove that the fraudulent disclosure caused a significant amount of the decline.
By its plain language, the press release consists of three separate pieces of information, and — contrary to plaintiffs and Hakala’s belief — Gompers did not invent that three-part classification.17 The press release first discusses the fraudulent “overstatement” and the estimated “decline in circulation related to this matter.” It then recognizes that the disclosure is “coupled with” the earlier reduction announcement and the “anticipated lower circulation” over a six-month period. Thus, the release divides the news into fraudulent and non-fraudulent information related to possible future circulation declines.
Plaintiffs assert, however, that even if the news can be divided into three parts, they still meet Oscar’s requirement of showing that the fraudulent disclosure caused a significant reduction in Belo’s stock price. As an initial point, plaintiffs’ original motion for certification does not meet our standards for proving loss causation. They submitted only SEC reports, stock-price charts, analyst reports, and other similar information; they did not include expert testimony.
As we stated in Oscar, these items contain “little more than well-informed speculation.” Id. at 271. Although analyst reports and stock prices are helpful in any inquiry, the testimony of an expert — along with some kind of analytical research or event study — is required to show loss causation. See id. For all of the reasons stated in Greenberg and Oscar, stock prices and analyst reports, without more, are insufficient at the class certification stage.
*410Even considering the plaintiffs’ analyst commentary and stock price information together with Hakala’s testimony and event study, the motion for class certification still falls short. As the district court correctly held, Hakala’s testimony was fatally flawed; he wedded himself to the idea that the press release was only one piece of news and conducted his event study based on that belief. We reject any event study that shows only how a “stock reacted to the entire bundle of negative information,” rather than examining the “evidence linking the culpable disclosure to the stock-price movement.” Id. Because Hakala based his study on that incorrect assumption, it cannot be used to support a finding of loss causation.
Without the event study, the rest of Hakala’s testimony relates to analyst opinions about Belo’s stock decline. Again, this “well-informed speculation” cannot support a finding of loss causation “without reference to any post-mortem data [the analysts] have reviewed or conducted.” Id. Thus, once we disregard Hakala’s flawed event study, the rest of his testimony is insufficient to prove loss causation.
Plaintiffs’ other arguments are also flawed. First, they argue that the press release recognizes that the fraudulent disclosure would result in a 1.5% total daily paper decline and a 5% Sunday decline. This, they claim, is nearly one-third of the total daily decline and over 40% of the total Sunday decline from all three disclosures. Assuming arguendo that one-third of a given stock decline is a “significant amount” of the total, we do not need to answer that question,18 because plaintiffs fail to understand which decline we are examining. A court should examine the decline in the stock price related to the disclosure, not the decline in the circulation.
Conceivably, DMN’s fraudulent practices could have resulted in 90% of the circulation decline, but if the stock price fell because the market was concerned only with the reason for the other 10%, loss causation could not be proven. Belo’s fraud regarding DMN was significant, but for long-term investors, news about the substantial and continuing decline in nationwide newspaper circulation could be much more disconcerting than were the fraudulent practices. If investors sold Belo’s stock' because of that long-term trend, and not because of the fraud, there is no loss causation.
Plaintiffs’ only remaining argument is premised on Hakala’s allegation that the market already knew about and had absorbed the impact of the non-fraudulent disclosures. If that is so, the plaintiffs argue, Oscar does not apply, the stock drop after the press release is related to only the one fraudulent disclosure, and the drop in stock price alone proves loss causation. We disagree. “There is no reason why the concerns stated in Oscar do not equally apply to cases in which only one negative disclosure is at issue.” Luskin v. Intervoice-Brite Inc., 261 Fed.Appx. 697, 702 (5th Cir.2008). Plaintiffs still must prove loss causation through the same rigorous process that we established in Oscar, even if there is only one negative disclosure. “[P]laintiffs cannot trigger the presumption of reliance by simply offering evidence of any decrease in price following the release of negative information.” Greenberg, 364 F.3d at 665.
Under an alternative system, were a defendant to release a corrective disclo*411sure on a particularly volatile market day, its stock could plummet regardless of whether the market cared about the disclosure. Such a drop, even coupled with the disclosure, is insufficient, however, unless there is a showing that the disclosure actually caused the decline. A class action may not proceed against a defendant whose only fault is releasing a disclosure on a volatile trading day.
Securities fraud litigation is not “a scheme of investor’s insurance”19 but instead is designed to protect those who buy stock at fraudulently inflated prices. If the fraud did not cause the price of the stock to increase, and its disclosure does not cause the price to go down, no injury has occurred. Thus, regardless of the number of disclosures, plaintiffs must establish the connection between the disclosure and the decline in price.20
V.
In summary, the district court did not abuse its discretion in denying class certification. Plaintiffs’ failure to present an expert witness with an event study or other analytical evidence of loss causation runs afoul of Oscar. The order denying class certification is AFFIRMED.
. The district court designated Operating Engineers Construction Industry and Miscellaneous Pension Fund as the lead plaintiff.
. Other developments, including the district court's dismissal of some claims against the individual defendants, are described in district court opinions on this dispute. See Fener v. Belo Corp., 425 F.Supp.2d 788 (N.D.Tex. 2006) (Fener I); Fener v. Belo Corp., 513 F.Supp.2d 733 (N.D.Tex.2007) (Fener II); Fener v. Belo Corp., Nos. 3:04-CV-1836-D, 3:04-CV-I869-D, 3:04-CV-2156-D, 2007 WL 4165709 (N.D.Tex. Nov. 26, 2007) {Fener III).
. We discuss Gompers’s and Hakala’s testimony more thoroughly infra.
. See Janet Cooper Alexander, Do the Merits Matter? A Study of Settlements in Securities Class Actions, 43 Stan. L.Rev. 497, 501 (1991) ("[T]he actual distribution of the settlement funds suggests that the present system does not really benefit the small investors who are presumed to be the beneficiaries of class actions, and it may actually foreclose more efficient client-controlled litigation by large investors.”).
. Vincent E. O'Brien, The Class-Action Shakedown Racket, Wall St. J., Sept. 10, 1991, at A20.
. Castano, 84 F.3d at 746.
. In Oscar Private Equity Investments v. Allegiance Telecom, Inc., 487 F.3d 261 (5th Cir. 2007), we further detailed our requirements for loss causation:
We now require more than proof of a material misstatement; we require proof that the misstatement actually moved the market. That is, the plaintiff may recover under the fraud on the market theory if he can prove that the defendant’s non-disclosure materially affected the market price of the security. Essentially, we require plaintiffs to establish loss causation in order to trigger the fraud-on-the-market presumption. Our most recent statement of this rule was in Greenberg, which held that to trigger the presumption of reliance plaintiffs must demonstrate that the cause of the decline in price is due to the revelation of the truth and not the release of the unrelated negative information.
Id. at 265 (footnotes, internal quotation marks, ellipses, and brackets omitted).
. See Lormand v. U.S. Unwired, Inc., 565 F.3d 228, 255-58 (5th Cir.2009) (discussing pleading standards for loss causation).
. See Oscar, 487 F.3d at 266-70 (establishing that this examination is appropriate as part of the class certification inquiry).
. See id. at 269 n. 40 ("This is not to say that loss causation, as an element of a 10b-5 claim, cannot be reexamined at summary judgment.”).
. The parties submitted extensive letters regarding the impact of Lormand, which, however, examines what is required under Federal Rule of Civil Procedure 8 to survive the pleadings stage. To the degree those standards are easier to meet than are our requirements for class certification, Lormand does not inform our decision.
. Of those hundred pages, eighty were the resumes of the lawyers at Fener's law firm. That information is not helpful for the class certification motion.
. We refer to this part of the press release as “the fraudulent disclosure.”
. We refer to parts 2 and 3 of the press release collectively as "the non-fraudulent disclosures.”
. Gompers submitted a response to Hakala’s study challenging his conclusions and analysis. Ultimately, however, we do not need to examine Gompers’s reply, because his initial study — rather than Hakala’s — discusses the proper way in which to examine Belo’s disclosures.
. We also clarified that this inquiry is appropriate at the class certification stage. See Oscar, 487 F.3d at 266-70.
. The relevant part of the press release reads as follows:
Belo Corp. announced today that [DMN], a wholly-owned subsidiary, will report a greater than expected decline in its September 2004 circulation. An internal investigation, which is ongoing, has disclosed practices and procedures that led to an overstatement in circulation, primarily in single copy sales. Belo, estimates the decline in circulation related to this matter to be approximately 1.5 percent daily and five percent Sunday. This decline, coupled with a reduction in state circulation that was first communicated publicly on March 9, 2004, of approximately 2.5 percent daily and 2.5 percent Sunday, and anticipated lower circulation volumes for the six month period ending September 30, 2004, will result in a total decline in circulation of approximately five percent daily and 11.5 percent Sunday ....
. Like the court in Oscar, "[w]e will not attempt to quantify what fraction of a decline is ‘significant.’ ” Oscar, 487 F.3d at 270.
. Basic, 485 U.S. at 252, 108 S.Ct. 978 (White, J., concurring in part and dissenting in part) (citation omitted).
. Hakala also failed to submit enough evidence to prove that the stock market absorbed the other pieces of information. He admitted that his analysis was incomplete, and the conclusions he did reach lack analytical support.