In re ORACLE CORPORATION SECURITIES LITIGATION, Nursing Home Pension Fund, Local 144; Drifton Finance Corporation; Robert D. Sawyer; Ryan Kuehmichel; Dzung Chu, Plaintiffs-Appellants, v. Oracle Corporation; Lawrence J. Ellison; Jeffrey O. Henley; Edward J. Sanderson, Defendants-Appellees, 627 F.3d 376

OPINION TALLMAN, Circuit Judge: Oracle Corporation is the second-largest producer of software in the world. In the third quarter of its 2001 fiscal year, Oracle missed its forecasted earnings per share by two cents. Its stock price dropped. A legion of analysts blamed the miss on a late-quarter reaction by several key customers to the unfolding U.S. economic downturn that would become commonly known as the burst of the dot-com bubble.

Plaintiffs, several intra-quarter purchasers of Oracle common-stock, brought this securities litigation against Oracle and three of its officers alleging the miss was actually caused by an elaborate scheme to defraud the public about the quality of Oracle products and the revenue gained therefrom. Because Plaintiffs have not developed evidence sufficient to permit a reasonable jury to conclude that their losses were caused by the market’s reaction to Defendants’ alleged fraud, as opposed to Oracle’s poor financial health generally, we affirm the district court’s order granting

summary judgment in favor of Oracle. I Plaintiffs allege that:

(1) Defendants violated

Section 10(

b) of the Securities Exchange Act of 1934 (“Exchange Act”), 15 U.S.C. § 78j(b), and Securities Exchange Commission Rule 10b — 5;

(2) Oracle’s Chief Executive Officer Larry Ellison, Chief Financial Officer Jeffrey Henley, and Executive Vice President Edward Sanderson are liable as control persons under

Section 20(

a) of the Exchange Act, 15 U.S.C. § 78t(a); and

(3) Henley and Ellison are liable for contemporaneous trading under

Section 20A of the Exchange Act, 15 U.S.C. § 78U-l(a). To support these allegations, Plaintiffs claim Defendants made false and misleading statements about a new software product, issued a false and misleading forecast for the third quarter of Oracle’s 2001 fiscal year, made false and misleading intra-quarter statements, and fraudulently overstated earnings for the second quarter of the 2001 fiscal year. The following undisputed facts are relevant to those claims.

In May 2000, Oracle released integrated business software called Suite lli, a suite of programs designed to work together to manage various company activities such as manufacturing, customer relations, sales, and accounting. Traditionally, enterprise resource planning (“ERP”) applications performed functions such as accounting, human resources, and manufacturing. Customer relationship management applications (“CRM”) performed functions such as managing call centers. Through the late 1990s, businesses that used enterprise applications software could not obtain ERP and CRM applications from the same vendor.

Suite lli was marketed as an innovative product that would combine the two types of applications. The four quarters of Oracle’s 2001 fiscal year were: from June 1 to August 31, 2000 (“1Q01”); from September 1 to November 30, 2000 (“2Q01”); from December 1, 2000, to February 28, 2001 (“3Q01”); and from March 1 to May 31, 2001 (“4Q01”). In 1Q01 and 2Q01, Oracle earned a combined $435 million in revenue from applications licenses. On December 14, 2000, Oracle announced 2Q01 earnings per share (“EPS”) of 11 cents and 66% growth in sales of Suite lli applications.

The same day, Oracle issued public guidance for 3Q01 projecting EPS of 12 cents. This guidance was based upon an accounting process that had generated projections that Oracle had met or exceeded for seven consecutive quarters. Historically, the majority of Oracle’s sales were made in the final days of a quarter as customers waited for Oracle to significantly lower prices in an attempt to meet its quarterly projections.

This trend was commonly referred to as “the hockey-stick effect” because plotting the quarterly sales on a graph resembled the shape of a hockey-stick — the sharp upswing at the end representing the bulk of quarterly sales. Oracle generates numerous internal financial reports throughout any given quarter. Relevant to this litigation are intraquarter reports projecting quarterly EPS — called internal forecasts — and intraquarter reports aggregating the company’s sales revenue at a given point in time— called flash reports.

Internal forecasts are an aggregation of the revenue and growth data contained in flash reports. Whereas flash reports take a snapshot of revenue and growth at the time they are produced, internal forecasts are forward-looking and involve a degree of extrapolation based on the judgment and experience of corporate management. From the beginning of 3Q01 on December 1, 2000, until February 5, 2001, every internal forecast that Oracle produced indicated that the company could fulfill its 3Q01 guidance. However, from February 5, 2001, until the end of the quarter, Oracle’s internally projected EPS began to fluctuate.

A February 5 internal forecast projected a potential EPS of 11 cents. A February 12 internal forecast returned the quarter’s potential EPS to 12 cents. A February 19 internal forecast again lowered the potential EPS to 11 cents. Then, a February 26 internal forecast — prepared two days prior to the end of the quarter— again predicted a potential EPS of 11 cents. While Oracle had exceeded late-quarter internal forecasts in quarters past, the recurring surge in late-quarter sales historically resulting from the hockey-stick effect did not materialize this time.

On March 1, 2001, Oracle preliminarily announced a 3Q01 EPS of 10 cents — a two penny miss. The announcement quoted Ellison as saying, License growth was strong in the first two months of Q3, and our internal sales forecast looked good up until the last few days of the quarter. However, a substantial number of our customers decided to delay their IT spending based on the economic slow-down in the United States. Sales growth for Oracle products in Europe and Asia Pacific remained strong. The problem is the U.S. economy. On March 2, 2001, Oracle’s stock price dropped from $21.38 to $16.88.

Earlier in the quarter, Ellison and Henley had exercised millions of Oracle stock options. Henley sold one million Oracle shares for a total of $32 million on January 4, 2001. Ellison sold 2.09% of his Oracle holdings between January 22 and January 31, 2001. A substantial amount of the shares sold were acquired through options set to expire in August 2001. In total, Ellison sold a little over 29 million Oracle shares for approximately $895 million. Importantly, these sales predated the first internal forecast indicating a potential EPS lower than the quarterly guidance of 12 cents.

II The district court originally dismissed Plaintiffs’ revised second amended complaint for failure to state a claim under Federal Rule of Civil Procedure 12(b)(6), finding that the allegations did not create a strong inference that intra-quarter statements made by Defendants were known to be false when made. See In re Oracle Corp. Sec. Litig., No. 01-CV-988, 2003 WL 23208956 (N.D.Cal. March 24, 2003). In November 2004 we reversed, holding that the operative complaint met the heightened pleading requirements of the Private Securities Litigation Reform Act (“PSLRA”). See Nursing Home Pension Fund, Local 144 v.

Oracle Corp., 380 F.3d 1226 (9th Cir.2004). Copious discovery ensued. On September 2, 2008, as a result of Defendants’ failure to preserve all its potential evidence, the district court issued an order concluding that Plaintiffs were entitled to adverse inferences with regard to two categories of evidence: Ellison’s email files and materials created in the process of writing the book Softwar. As a sanction, the court ruled that Plaintiffs would be entitled to an inference that the spoliated evidence would demonstrate Ellison’s knowledge of any material facts that Plaintiffs could otherwise establish.

Despite the absence of the missing Ellison emails and tapes, discovery proceedings in this case produced over 134 deposition days, countless discovery requests and answers, and over 2.1 million pages of documents. As the experienced district judge quipped, the word “voluminous” does not do justice to the record in this case. On October 20, 2008, Defendants moved for

summary judgment. Prior to the hearing on

summary judgment, Defendants made over eighty evidentiary objections. Plaintiffs never responded to those objections — notwithstanding the passage of four months between the

summary judgment hearing and the district court’s final order. On June 19, 2009, the district court made several key evidentiary rulings and granted Defendants’ motion for

summary judgment. In re Oracle Corp. Sec. Litig., No. 01-CV-988, 2009 WL 1709050 (N.D. Cal. June 19, 2009). The court concluded that Plaintiffs had failed to create a genuine issue of material fact as to (1) whether Oracle’s 3Q01 forecast lacked a reasonable basis, (2) whether several intra-quarter statements were false or misleading, (3) whether Plaintiffs’ loss was caused by Defendants’ misrepresentations regarding Suite lli, and (4) whether Plaintiffs’ loss was caused by Defendants’ misrepresentation of 2Q01 earnings. Plaintiffs timely appealed.

Ill Not surprisingly, Plaintiffs pursue numerous evidentiary arguments on appeal. A We review the district court’s exclusion of evidence for an abuse of discretion. United States v. Mitchell, 502 F.3d 931, 964 (9th Cir.2007). A district court abuses its discretion if it reaches a result that is illogical, implausible, or without support in inferences that may be drawn from facts in the record. United States v. Hinkson, 585 F.3d 1247, 1251 (9th Cir.2009) (en banc). A district court’s ruling on a motion for

summary judgment may only be based on admissible evidence. Beyene v. Coleman Sec. Servs., Inc., 854 F.2d 1179, 1181 (9th Cir.1988) (citing Fed. R.Civ.P. 56(e)). In this case, prior to the district court’s ruling on Defendants’ motion for

summary judgment, Plaintiffs failed to respond to over eighty evidentiary objections made by Defendants. There is much disagreement between the parties as to whether or not each of these objections had been preemptively addressed by Plaintiffs. Irrespective of that dispute, we conclude the district court did not abuse its discretion on this record. Plaintiffs, as the parties seeking admission of most of the evidence at issue, bore the burden of proof to show its admissibility. Pfingston v. Ronan Eng’g Co., 284 F.3d 999, 1004 (9th Cir.2002).

Given the overwhelming volume of documents before the district court, once Defendants objected to the evidence Plaintiffs sought to be admitted, the onus was on Plaintiffs to direct the district court’s attention to authenticating documents, deposition testimony bearing on attribution, hearsay exceptions and exemptions, or other evidentiary principles under which the evidence in question could be deemed admissible by the district court. Plaintiffs failed to do so.

We cannot declare that the district court reached an illogical or implausible result by excluding apparent hearsay or documents without sufficient foundational support as the rules of evidence prescribe, or that the district court otherwise exceeded the permissible bounds of its discretion by failing to comb through the voluminous record searching for evidentiary bases to introduce the evidence at issue. That was Plaintiffs’ obligation. It behooves litigants, particularly in a case with a record of this magnitude, to resist the temptation to treat judges as if they were pigs sniffing for truffles. See Downs v.

Los Angeles Unified Sch. Dist., 228 F.3d 1003, 1007 n. 1 (9th Cir.2000) (citing United States v. Dunkel, 927 F.2d 955, 956 (7th Cir.1991)). Moreover, for Plaintiffs to fail to respond to Defendants’ objections, and to then challenge the district court’s evidentiary rulings on appeal, is to invite the district court to err and then complain of that very error. We cannot countenance such a tactic on appeal. Cf. United States v. Reyes-Alvarado, 963 F.2d 1184, 1187 (9th Cir.1992) (“The doctrine of invited error prevents a [party] from complaining of an error that was his own fault.”).

B The district court found that Defendants wilfully failed to preserve Ellison’s email files and taped discussions with the author of the book Softwar. As a result, the district court gave Plaintiffs the benefit of an adverse inference that the emails and tapes would have proved Ellison’s knowledge of any material facts that Plaintiffs were able to establish. Unsuccessful in their efforts to persuade the district court that material issues of fact actually existed, Plaintiffs now argue the adverse inferences should not have been limited to Ellison’s knowledge.

Instead, they urge the inferences should be sufficient to defeat a challenge to the insufficiency of their prima facie case. We review a district court’s imposition of spoliation sanctions for an abuse of discretion. Anheuser-Busch, Inc. v. Natural Beverage Distribs., 69 F.3d 337, 348 (9th Cir.1995). On this record, we see no abuse of discretion in the district court’s construction and application of the adverse-inferences. Over 2.1 million documents were produced during discovery.

Although Ellison’s email account files were not produced, the documents that were produced contained numerous email chains in which Ellison’s correspondence was contained. If there were material issues of fact supporting securities fraud, Plaintiffs should have been able to glean them from the documents actually produced, the extensive deposition testimony, and the written discovery between the parties. An adverse inference would then properly apply to establish that Ellison must have known of those damaging material facts. Plaintiffs’ problem here lies in the dearth of admissible evidence to show fraud.

A district court’s adverse inference sanction should be carefully fashioned to deny the wrongdoer the fruits of its misconduct yet not interfere with that party’s right to produce other relevant evidence. Campbell Indus, v. M/V Gemini, 619 F.2d 24, 27 (9th Cir.1980). In light of the enormous record developed in this case, the only conceivable benefit of Defendants’ spoliation was the possibility of disclaiming Ellison’s knowledge of any damaging facts underlying the purported fraud. The district court’s sanction was carefully fashioned to deny Defendants that benefit.

As a result, there was no abuse of discretion. See id. IV Turning to the merits of this dispute, we review the district court’s order granting Defendants’ motion for

summary judgment de novo. Buono v. Norton, 371 F.3d 543, 545 (9th Cir.2004). We are mindful of the shifting burden of proof governing motions for

summary judgment under Federal Rule of Civil Procedure 56. The moving party initially bears the burden of proving the absence of a genuine issue of material fact. Celotex Corp. v. Catrett, 477 U.S. 317, 323, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986). Where the non-moving party bears the burden of proof at trial, the moving party need only prove that there is an absence of evidence to support the non-moving party’s case. Id. at 325, 106 S.Ct. 2548.

Where the moving party meets that burden, the burden then shifts to the non-moving party to designate specific facts demonstrating the existence of genuine issues for trial. Id. at 324, 106 S.Ct. 2548. This burden is not a light one. The non-moving party must show more than the mere existence of a scintilla of evidence. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 252, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). The non-moving party must do more than show there is some “metaphysical doubt” as to the material facts at issue. Matsushita Elec. Indus. Co., Ltd. v.

Zenith Radio Corp., 475 U.S. 574, 586, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986). In fact, the non-moving party must come forth with evidence from which a jury could reasonably render a verdict in the non-moving party’s favor. Anderson, 477 U.S. at 252, 106 S.Ct. 2505. In determining whether a jury could reasonably render a verdict in the non-moving party’s favor, all justifiable inferences are to be drawn in its favor. Id. at 255, 106 S.Ct. 2505. Plaintiffs allege Defendants violated

Section 10(

b) of the Exchange Act and Securities Exchange Commission Rule 10b-5 promulgated thereunder.

Section 10(

b) makes it unlawful to “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 Commission may prescribe.” 15 U.S.C. § 78j(b). The scope of Rule 10b-5 is coextensive with that of

Section 10(b). SEC v. Zandford, 535 U.S. 813, 815 n. 1, 122 S.Ct. 1899, 153 L.Ed.2d 1 (2002). “In a typical § 10(

b) private action a plaintiff must prove (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.” Stoneridge Inv. Partners, LLC v. Scientific-Atlanta, Inc., 552 U.S. 148, 156, 128 S.Ct. 761, 169 L.Ed.2d 627 (2008) (citation omitted). Plaintiffs argue that Defendants violated

Section 10(

b) by (1) issuing a 3Q01 forecast that lacked a reasonable basis, (2) making intra-quarter statements repeating the 3Q01 forecast and denying the effects of a slowing economy with knowledge of facts tending seriously to undermine those statements, (3) misrepresenting the functionality and success of Suite lli, and (4) overstating 2Q01 earnings. Plaintiffs have not developed evidence that would allow a jury to reasonably conclude that Defendants’ 3Q01 forecast or intra-quarter statements constituted material misrepresentations or omissions.

As to their Suite lli and 2Q01 claims, Plaintiffs cannot prove loss causation because they have not developed evidence sufficient to allow a jury to reasonably conclude that Defendants’ purported misrepresentations were a substantial cause of the decline of Oracle’s stock price on March 2, 2001. A For a forward-looking statement such as Oracle’s 3Q01 public guidance to constitute a material misrepresentation giving rise to

Section 10(

b) or Rule 10b-5 liability, a plaintiff must prove either “(1) the statement is not actually believed [by the speaker], (2) there is no reasonable basis for the belief, or (3) the speaker is aware of undisclosed facts tending seriously to undermine the statement’s accuracy.” Provenz v. Miller, 102 F.3d 1478, 1487 (9th Cir.1996) (citation omitted).

Plaintiffs allege that Defendants lacked a reasonable basis for their belief in Oracle’s December 14, 2000, public guidance for 3Q01 because the guidance failed to account for the serious impact that purported Suite lli defects had on sales or the effects of a declining economy. Oracle’s 3Q01 forecast was based on a “bottom up” assessment of revenue that began with sales representatives and that had produced accurate but conservative forecasts for seven consecutive quarters. We describe it below.

The structure of this process in and of itself refutes Plaintiffs’ argument that the forecast did not account for negative sales trends resulting from alleged Suite lli problems or the declining economy. As a result, Plaintiffs have not established sufficient evidence from which a jury could reasonably render a verdict in their favor. The bottom-up process began with sales representatives entering potential sales into a computer database.

These potential sales were known as the “pipeline.” Based on the pipeline, sales representatives would submit a forecast indicating what deals were likely to close and for how much. These forecasts were then adjusted up or down by regional managers based on that manager’s assessment of the likelihood that those potential sales would close. Regional forecasts were aggregated by Oracle’s Senior Vice President of Global Finance and Operations, Jennifer Minton. Minton would aggregate a forecast that adjusted the underlying forecasts up or down depending on several factors.

Those factors included: a license pipeline conversion analysis; a review of the status of certain large transactions and at times a request for executive involvement to ensure that a deal progressed and closed; conversations with “key financial personnel” in the field to seek their “independent thinking” as to whether or not the forecast was realistic; corrections to errors made in the underlying forecast when it was originally submitted; and an adjustment for “sandbagging,” a label management applied to understated forecasts by certain salespersons and managers who consistently exceeded their forecasts in the hope of generating larger bonuses.

In light of this process, Plaintiffs’ argument that Minton’s forecast and Oracle’s guidance failed to take into account the serious impact that purported Suite lli defects had on sales finds no support in the record. Baseline sales information submitted by sales representatives necessarily included any alleged impact that Suite Hi problems had on sales. In addition, “the effects of the declining economy” were also necessarily included in the size of deals and the probability of closing that sales representatives estimated for the quarter.

Plaintiffs’ strongest argument is that Minton took the aggregate volume of potential sales revenue in the pipeline for 3Q01 and simply applied the 3Q00 “conversion ratio” to come up with her 3Q01 forecast. Oracle’s “conversion ratio” is calculated by dividing the potential sales revenue in the pipeline at a given point in a quarter by the actual amount of revenue closed at the end of a quarter.

Plaintiffs thus maintain that if Oracle’s 3Q01 guidance announced December 14, 2000, was solely established by applying the 3Q00 conversion ratio to 3Q01 pipeline information, then the guidance would have failed to have accounted for alleged Suite Hi problems and the effects of the declining economy that had developed between 3Q00 and 3Q01. While appealing in the abstract, this argument is unsupported by the record. Oracle did not solely apply the 3Q00 conversion ratio to the pipeline at the outset of 3Q01 to come up with the forecast.

While Minton indicated in deposition testimony that she heavily relied on the 3Q00 conversion ratio, Defendants introduced an expert report containing a chart comparing the actual conversion ratio at several points in 3Q00 to Minton’s 3Q01 forecast’s estimated conversion ratio for the same points in 3Q01. The two ratios are different at each interval.

This comparison indicates that Minton took other factors into account — such as information gathered from conversations with key financial analysts within Oracle’s various divisions as well as the likelihood of closing certain targeted large deals — when determining the 3Q01 forecast. This conclusion is corroborated by the deposition testimony of several Oracle employees. While we must draw justifiable inferences in Plaintiffs’ favor, Anderson, 477 U.S. at 255, 106 S.Ct. 2505, the undisputed chart renders unjustifiable any inference that Oracle’s guidance was solely based on the 3Q00 conversion ratio.

As a result of Oracle’s thorough forecasting process, Plaintiffs are unable to prove that Defendants lacked at least a reasonable basis for their belief in the 3Q01 forecast. Moreover, the fact that Oracle’s forecast turned out to be incorrect does not retroactively make it a misrepresentation. See In re VeriFone Sec. Litig., 11 F.3d 865, 871 (9th Cir.1993) (“The fact that the prediction proves to be wrong in hindsight does not render the statement untrue when made.”) (citing Marx v. Computer Seis. Corp., 507 F.2d 485, 489-90 (9th Cir.1974)); Paracor Fin., Inc. v. Gen. Elec.

Capital Corp., 96 F.3d 1151, 1158-59 (9th Cir.1996) (“[T]he Investors have not introduced evidence that GE Capital lacked at least a reasonable basis for their various representations, even though in hindsight they may now appear a little too rosy.”). Therefore, a jury could not reasonably conclude that Oracle’s 3Q01 forecast was a material misrepresentation giving rise to

Section 10(

b) or Rule 10b-5 liability. B A statement can constitute a material misrepresentation giving rise to

Section 10(

b) or Rule 10b-5 liability if there is no reasonable basis for the speaker’s belief in the statement’s accuracy or if the speaker is aware of undisclosed facts tending seriously to undermine the statement’s accuracy. Provenz, 102 F.3d at 1487. Plaintiffs argue that several intra-quarter statements in which Defendants expressed confidence in the 3Q01 forecast or denied the effects of a weakening economy were material misrepresentations. In light of the district court’s evidentiary rulings, however, Plaintiffs are only able to challenge the accuracy of two statements by Oracle management.

We conclude as a matter of law that neither of these statements constituted a material misrepresentation. Plaintiffs must “demonstrate that a particular statement, when read in light of all the information then available to the market, or a failure to disclose particular information, conveyed a false or misleading impression.” In re Convergent Techs. Sec. Litig., 948 F.2d 507, 512 (9th Cir.1991). They must also show that Defendants engaged in “knowing” or “intentional” conduct or acted with “deliberate recklessness.” South Ferry LP, No. 2 v. Killinger, 542 F.3d 776, 782 (9th Cir.2008).

In the securities context, “an actor is reckless if he had reasonable grounds to believe material facts existed that were misstated or omitted, but nonetheless failed to obtain and disclose such facts although he could have done so without extraordinary effort.” Howard v. Everex Sys., Inc., 228 F.3d 1057, 1064 (9th Cir.2000) (citation omitted). First, Plaintiffs allege that a February 13, 2001, statement made by Executive Vice President George Roberts was a material misrepresentation.

Roberts purportedly said, “our guidance remains the same that we indicated at the beginning of Q3.” As previously discussed, Oracle’s internal forecasts began to fluctuate on February 5, 2001. Plaintiffs thus argue that Roberts’ February 13 statement was a material misrepresentation. While the February 5, 2001, internal forecast indicated a potential EPS of 11 cents, the February 12 internal forecast returned the potential EPS to the original forecast of 12 cents. Roberts’ statement was therefore supported by the February 12 internal forecast.

The fact that the February 13 prediction proved incorrect in hindsight does not make it untrue when made. In re VeriFone Sec. Litig., 11 F.3d at 871. Rather, the question is whether Roberts lacked at least a reasonable basis for his representation when it was made, or whether he made the representation with knowledge of facts tending to seriously undermine its accuracy. Paracor Fin., Inc., 96 F.3d at 1158-59. Plaintiffs base a considerable amount of their argument on a single flash report produced on January 17, 2001.

This report indicates that, but for a single $60 million license deal with Covisint, the December FY01 growth rate would have been only six percent. With the Covisint deal, however, December growth was thirty-five percent — ten points better than the publicly forecasted twenty-five percent growth and twenty-five points better than the ten percent growth in December FYOO. We see no reason why a successful transaction should be excluded from Oracle’s forecasting model.

Moreover, we note that information contained in such flash reports and internal emails was necessarily included in Oracle’s internal forecasts throughout the quarter — including the February 12, 2001, internal forecast preceding Roberts’ statement. We have previously said that issuers need not reveal all internal projections. In re Stac Elecs. Sec. Litig., 89 F.3d 1399, 1411 (9th Cir.1996) (citation omitted). Companies generate numerous estimates internally, and they may reveal the projection they think best while withholding others, as long as the projection revealed had a reasonable basis. Id.

The most recent internal forecast to precede Roberts’ statement supported Oracle’s 3Q01 guidance, and his statement therefore had a reasonable basis. In addition, viewing the totality of the information available to Roberts at the time he made the statement, a jury could not reasonably conclude that he had knowledge of facts tending seriously to undermine its accuracy. Second, Plaintiffs allege that a February 21, 2001, public statement made by Henley, as repeated in a February 26, 2001, internal email, was a material misrepresentation.

As reflected in the email, Henley purportedly said, “we do not expect the slowing economy, barring a serious slide to a recession, to significantly impact near term guidance.” This statement is not unambiguous. The contingency, “barring a serious slide to a recession” and the modifier “significantly” attached to “impact near term guidance” leave a reader with a less-than-certain impression that Henley stood by Oracle’s 3Q01 guidance. Furthermore, there is no dispute over the fact that the United States economy slid to a recession beginning in March 2001, when the dot-com bubble burst.

See The Business-Cycle Peak of March 2001, National Bureau of Economic Research (Nov. 26, 2001), available at http://www.nber.org/ cycles/november2001/recessions.pdf. Henley’s contingency therefore arose. His confidence in Oracle’s guidance barring such a slide into a recession was not a material misrepresentation. C At the heart of Plaintiffs’ Suite lli argument is their fundamental disagreement with the district court’s theory of loss causation. In the end, if Plaintiffs cannot establish loss causation — the sixth element of a

Section 10(

b) violation — then a jury could not reasonably return a verdict in their favor on this claim. We see no error in the district court’s application of precedent on this point. Loss causation is the causal connection between a defendant’s material misrepresentation and a plaintiffs loss. Dura Pharms., Inc. v. Broudo, 544 U.S. 336, 344-45, 125 S.Ct. 1627, 161 L.Ed.2d 577 (2005). “A plaintiff bears the burden of proving that a defendant’s alleged unlawful act ‘caused the loss for which the plaintiff seeks to recover damages.’ ” In re Gilead Seis. Sec.

Litig., 536 F.3d 1049, 1055 (9th Cir.2008) (quoting 15 U.S.C. § 78u-4(b)(4)). In Dura, the Supreme Court held that a person who misrepresents the financial condition of a corporation in order to sell stock is only liable to a relying purchaser for the loss the purchaser sustains when the facts “become generally known” and “as a result” share value depreciates. 544 U.S. at 344-45, 125 S.Ct. 1627. To adequately plead loss causation, the Court held, a plaintiff must allege that the “share price fell significantly after the truth became known.” Id. at 347, 125 S.Ct. 1627.

The district court concluded that this case is analogous to Metzler Inv. GMBH v. Corinthian Colls., Inc., 540 F.3d 1049, 1063 (9th Cir.2008). We agree. In Metzler, the purported fraud was the manipulation of student enrollment figures at a college. Id. We applied Dura and indicated that loss causation is not adequately pled unless a plaintiff alleges that the market learned of and reacted to the practices the plaintiff contends are fraudulent, as opposed to merely reports of the defendant’s poor financial health generally. Id.

The market need not know at the time that the practices in question constitute a “fraud,” nor label them “fraudulent”, but in order to establish loss causation, the market must learn of and react to those particular practices themselves. Id. This reaction, in turn, must be the cause of a plaintiffs loss. Id. As the stock drop in Metzler was not adequately alleged to be a result of the market learning of and reacting to enrollment fraud, as opposed to the college’s poor financial health generally, the plaintiffs had not adequately pled loss causation. Id. at 1064.

While Metzler addressed loss causation on a motion to dismiss, our discussion of loss causation applies with even greater force to our consideration of Defendants’ motion for

summary judgment in the present case. Plaintiffs take issue with our opinion in Metzler. Specifically, they assert that they should be able to prove loss causation by showing that the market reacted to the purported “impact” of the alleged fraud — -the earnings miss— rather than to the fraudulent acts themselves. We reject that assertion. Loss causation requires more than an earnings miss. See Dura, 544 U.S. at 343, 125 S.Ct. 1627 (“To ‘touch upon’ a loss is not to cause a loss, and it is the latter that the law requires.”); Metzler, 540 F.3d at 1063 (“Loss causation requires more.”).

Loss causation is established if the market learns of a defendant’s fraudulent act or practice, the market reacts to the fraudulent act or practice, and a plaintiff suffers a loss as a result of the market’s reaction. See Metzler, 540 F.3d at 1063. Evaluating the totality of evidence in this case, we hold that Plaintiffs are unable to create a triable dispute that Oracle’s share price dropped as a result of the market learning of and reacting to Defendants’ purported fraud, as opposed to Oracle’s poor financial health generally.

Plaintiffs allege a fraud that consisted of statements misrepresenting the quality and success of Oracle’s Suite lli product. Even though Plaintiffs acknowledge that the market was already aware of initial rollout issues with Suite lli, Plaintiffs allege that Defendants’ misrepresentations downplayed defects in the product. The alleged fraud in this case, therefore, concealed a purported “truth” about Suite lli. According to the Plaintiffs, the March 1, 2001 earnings miss revealed this “truth”: Suite lli was defective and customers had not bought it as a result of the defects.

Thus, Plaintiffs argue, the truth regarding Suite lli “[became] generally known” on March 1 and the stock price dropped as a result. Dura, 544 U.S. at 345, 125 S.Ct. 1627. Plaintiffs’ theory is unsupported by the record. The overwhelming evidence produced during discovery indicates the market understood Oracle’s earnings miss to be a result of several deals lost in the final weeks of the quarter due to customer concern over the declining economy. Numerous analyst reports support this conclusion. To be sure, the miss was caused by customers not buying Oracle’s products.

But the market did not learn on March 1, 2001, that customers did not buy Suite lli as a result of defects. Instead, the market learned that customers did not buy Oracle’s products in the final weeks of the quarter as a result of uncertainty in an economy that would slide into a recession within the next month. In other words, the market reacted to reports of Oracle’s “poor financial health generally.” Metzler, 540 F.3d at 1063; cf. In re Daou Sys., Inc., Sec.

Litig., 411 F.3d 1006, 1026 (9th Cir.2005) (holding loss causation sufficiently pled where analyst reports specifically noted, “[y]ou have got to question whether they are manufacturing earnings”); see also In re Williams Secs, litig. — WCG Subclass, 558 F.3d 1130, 1139-40 (10th Cir.2009) (holding that loss causation was not established when a plaintiff failed to show that losses following public disclosures could be “reliably attributed to the revelation of fraud rather than to other factors”).

While Plaintiffs did find two analyst reports purportedly questioning this explanation, these reports would not allow a jury reasonably to render a verdict in their favor in light of the agglomeration of evidence supporting a contrary conclusion. Moreover, “the market” must react to the fraudulent acts or practices. Metzler, 540 F.3d at 1063. Plaintiffs’ two reports are not indicative of what “the market” learned of and reacted to.

Over 130 pages of the record before this court consist of analyst reports — many based on independent interviews with Oracle’s customers— confirming that late-quarter sales were lost as a result of customer concerns over a faltering economy. This overwhelming evidence renders unjustifiable an inference that Plaintiffs’ two reports reflect the market’s consensus. Plaintiffs’ reliance on a single March 13, 2001, email written by an employee in Oracle’s European division is equally unavailing. The email is not evidence of what the market learned of and reacted to on March 1, 2001.

Finally, Oracle made $338 million in revenue from Suite lli in 4Q01 alone. If “the market” learned of and reacted to fraudulently concealed Suite lli defects at the end of 3Q01, surely Oracle’s customers would have reacted as well. Plaintiffs cannot establish loss causation. Their

Section 10(

b) claim alleging misrepresentations of Suite lli’s quality and success fails as a result. D Plaintiffs also argue that their losses were caused by fraudulently overstated 2Q01 earnings. This claim fails for the same reasons as Plaintiffs’ Suite lli claim. There is simply no evidence to support the conclusion that Oracle’s stock price dropped as a result of the market’s reaction to learning of fraudulently overstated 2Q01 earnings on March 1, 2001, as opposed to Oracle’s poor financial health generally. Metzler, 540 F.3d at 1063. Oracle’s 2Q01 earnings have never been revised downward.

To the extent Plaintiffs argue that the purportedly fraudulent 2Q01 earnings statement caused the 3Q01 earnings miss, we reiterate that an earnings miss alone is insufficient to establish loss causation. See id.; Dura, 544 U.S. at 343, 125 S.Ct. 1627. As control-person liability under

Section 20(

a) requires a prerequisite

Section 10(

b) violation, Plaintiffs’

Section 20(

a) claim fails as well. See 15 U.S.C. § 78t(a). V Plaintiffs separately allege that Ellison and Henley are liable for contemporaneous trading under

Section 20A of the Exchange Act. See 15 U.S.C. § 78t-1(a). Under that section, “Any person who violates any provision of this

chapter or the rules or regulations thereunder by purchasing or selling a security while in possession of material, nonpublic information shall be liable ... to any person who ... has [contemporaneously] purchased ... securities of the same class.” Id. “Claims under

Section 20A are derivative and therefore require an independent violation of the Exchange Act.” Johnson v. Aljian, 490 F.3d 778, 781 (9th Cir.2007). Plaintiffs “must demonstrate that the alleged fraud occurred ‘in connection with the purchase or sale of a security.’ ” Id. at 782 (quoting Ambassador Hotel Co., Ltd. v. Wei-Chuan Inv., 189 F.3d 1017, 1025 (9th Cir.1999)). As with their

Section 10(

b) claim, Plaintiffs “must prove both actual cause (‘transaction causation’) and proximate cause (‘loss causation’).” Id. Plaintiffs’ inability to establish a triable issue on loss causation for their Suite 11i and 2Q01 claims ends their

Section 20A claim. See Johnson, 490 F.3d at 782. Moreover, as all of Ellison’s trades were completed prior to the first internal forecast indicating that Oracle might miss its 3Q0I guidance, the district court’s adverse inference merely imputes to Ellison the knowledge that Oracle was on track to fulfill its forecast. Plaintiffs’ contemporaneous-trading claims fail as a result. VI The district court properly granted Defendants’ motion for

summary judgment. AFFIRMED. . Plaintiffs now urge us to take judicial notice of certain deposition testimony that they claim bears on the propriety of the district court’s hearsay rulings. We decline Plaintiffs’ invitation. The proper place to call attention to any such testimony was in front of the district court, and furthermore, the content of a deposition is not a clearly established "fact” of which this panel can take notice. See Bias v. Moynihan, 508 F.3d 1212, 1225 (9th Cir.2007). .

On appeal, Defendants have disclaimed invocation of the PSLRA safe harbor exempting from liability forward-looking statements “containing a projection of revenues, income (including income loss), earnings (including earnings loss) per share, capital expenditures, dividends, capital structure, or other financial items” that are also accompanied by meaningful cautionary statements. 15 U.S.C. § 78u-5(i)(1)(A). . Plaintiffs also argue that the 3Q01 forecast lacked a reasonable basis as a result of two purported accounting frauds that resulted in a fraudulently overstated EPS of 11 cents for 2Q01.

First, we reject Plaintiffs' representation that the district court "found” a triable issue as to whether 2Q01 earnings were overstated. That representation is not supported by the district court’s order. Second, in light of the forecasting process discussed supra, we reject Plaintiffs’ argument that the 3Q01 forecast was rendered unreliable based upon fraudulently overstated 2Q01 earnings. .

Given our conclusion that Oracle’s December 14, 2000, guidance had a reasonable basis, we do not address Plaintiffs' challenges to December 14, 2000, and December 15, 2000, statements reaffirming that guidance, which, unlike the district court, we view as part and parcel of the guidance itself. . Unlike the district court, after de novo review, we view this statement as a forward-looking projection bearing upon Oracle's 3Q01 guidance. It is thus arguably subject to the PSLRA safe harbor. See 15 U.S.C. § 78u-5(i)(1)(A). However, Defendants have not invoked that provision on appeal. .

We note that this conclusion would also apply to Henley’s December 14, 2000, statement in which he hedged, "So, you know, if we were to have a — I guess, a hard landing, a recession, depression, I mean certainly, that could have some impact,” as well as his December 15, 2000, statement in which he hedged, "If the economy got really really bad then obviously that would probably have some effect on all of us.” .

Plaintiffs argue that Oracle’s March 1 earnings announcement served as the final piece of the puzzle regarding Suite lli from investors’ point of view, such that previous disclosures about lli's functionality do not eliminate a triable issue as to loss causation. In certain circumstances, it may be possible to prove loss causation even where the market reaction and loss occur some time after an initial public disclosure of the fraud. See Gilead, 536 F.3d at 1058.

We merely conclude, as did the district court, that nobody could reasonably find, based on the evidence admitted into the record in this case, that Plaintiffs’ losses were caused substantially by misstatements separate and apart from other independent factors. See id. at 1055. . We agree with the district court that Plaintiffs do not accurately describe the evidence they cite in support of their argument on this point. While Plaintiffs purport to have four analyst reports supporting "an lli link," only two could fairly be construed to imply that product problems caused the March 1, 2001, earnings miss. .

We also note that the email, written by Sergio Giacoletto, is insufficient to allow a jury reasonably to conclude that North American customers did not buy Suite lli as a result of defects. Giacoletto opines that salespeople in Oracle's European division pushed "technology” sales over "apps” sales in 3Q01 as a result of Suite lli rollout "problems” in 1Q01 and 2Q01. This email has no bearing on whether these "problems” were present in Oracle's North American division or whether they persisted throughout 3Q01.

Despite serving over one hundred subpoenas on Oracle's customers, Plaintiffs have not developed evidence to support a conclusion that North American customers declined to buy Suite 1 li in 3Q01 as a result of defects. Giacoletto’s email is inadequate to support such a conclusion. . Plaintiffs argue that a fraudulent transaction with Hewlett Packard was part of the 2Q01 accounting fraud, and that the district court erred by denying Plaintiffs an opportunity to amend their complaint to include this transaction.

We do not need to address this argument, however, because even including this transaction with the purported 2Q01 "debit memo accounting fraud” that the district court considered, Plaintiffs have failed to establish that the market learned of and reacted to a 2Q01 fraud when Oracle announced its 3Q01 earnings miss.