After a two-week trial, Mitchell Stein, a lawyer, was convicted of mail, wire, and securities fraud based on evidence that he fabricated press releases and purchase orders to inflate the stock price of his client Signalife, Inc., a publicly-traded manufacturer of medical devices. The district court sentenced Mr. Stein to 204 months’ imprisonment, over $5 million in forfeiture, and over $13 million in restitution. Mr. Stein appeals his conviction and sentence.
Regarding his conviction, Mr. Stein argues, among other points, that the government failed to disclose Brady material1 to the defense before trial and knowingly relied on false testimony to make its case. As regards his sentence, Mr. Stein argues that the district court erred in calculating actual loss for the purposes of the Mandatory Victims Restitution Act of 1996 (“MVRA”), 18 U.S.C. § 3663A, and § 2B1.1 of the United States Sentencing Guidelines. In particular, he argues that in estimating actual loss the district court *1140erroneously presumed that all purchasers of Signalife stock during the period when the fraud was ongoing relied on false information Mr, Stein promulgated. He also argues that the district court failed to take into account other market forces that likely contributed to the investors’ losses. After careful consideration of the parties’ briefs and with the benefit of oral argument, we affirm Mr. Stein’s conviction but vacate his sentence.
This opinion proceeds in three parts. We first provide background regarding Mr. Stein’s fraudulent scheme, his subsequent indictment, his pretrial and post-trial motions, and his sentencing. Second, we address and reject Mr. Stein’s challenges to his conviction. Mr. Stein identified only one potential Brady document, and it contained no information favorable to him and was accessible through reasonable diligence before trial. And, he failed to identify any suppressed material or any materially false testimony on which the government relied, purportedly in violation of Giglio.2
Third, with respect to sentencing, we review the district court’s actual loss calculation. We agree with Mr. Stein that to establish an actual loss figure under the guidelines or the MVRA based on investors’ losses, the government must prove that, in deciding to purchase Signalife stock, investors relied on the fraudulent information Mr. Stein disseminated. The district court found that more than 2,000 investors relied on Mr. Stein’s fraudulent information, but the only evidence supporting this finding was the testimony of two individuals that they relied on Mr. Stein’s false press releases and generalized evidence that some investors may rely on some public information. This evidence was insufficient to permit reliance to be inferred for over 2,000 investors. Accordingly, the district court erred in calculating an actual loss figure based on the losses of all these investors. The district court also failed to determine whether intervening events caused the Signalife stock price to drop and, if so, whether these events were unforeseeable such that their effects should be subtracted from the actual loss figure. We remand so that the district court can remedy these errors.
I. BACKGROUND
A. The Fraudulent Scheme
The evidence adduced at trial—including the testimony of Mr. Stein’s two co-eon-spirators, Martin Carter and Ajay Anand—supported the following facts. In an effort to inflate artificially the value of Signalife stock, Mr. Stein drafted three press releases and three corresponding purchase orders touting more than $5 million in bogus Signalife sales.3 The fraudulent period began on September 20, 2007, when Mr. Stein sent the first false press release to John Woodbury, Signalife’s securities lawyer, with instructions to publish it. The press release reported that Signal-ife had sold $1.98 million worth of its products. Mr. Stein represented that the press release was “backed up by a purchase order.” Trial Tr., Doc. 240 at 59.4 Mr. *1141Woodbury lacked any independent knowledge of the truth of the statements in the press release. He published it that day anyway, though, because Mr. Stein had told him that he and Signalife’s Chief Executive Officer, Lowell T. Harmison, were traveling together visiting potential clients, and Mr. Woodbury believed that this sale was the fruit of those efforts.
A few days later, Mr. Stein emailed Mr. Woodbury a second press release about an additional $3.3 million in sales and represented that Mr. Harmison had approved the press release. Mr, Woodbury published the release the next day despite lacking any supporting documentation.
Mr. Stein emailed Mr. Woodbury a third press release about two weeks later. The press release reported an additional $551,500 in sales orders. Mr. Woodbury issued the release early the next morning, again without supporting documentation.
Mr. Woodbury later asked Mr. Stein for additional information regarding the sales that were described in the press releases. In response, Mr. Stein sent Mr. Woodbury three purchase orders. None of these purchase orders provided an address for shipment. Tracey Jones, Mr. Harmison’s assistant, maintained that she “never received any backup or anything on” the purchase orders, and thus she considered them “phantom purchase orders.” Doc. 241 at 117.
The first purchase order, dated September 14, 2007, reflected an order by a company called Cardiac Hospital Management (“CHM”). The order reflected a sale of $1.93 million worth of product and noted a $50,000 deposit. The signature block showed “Cardiac Hospital Management” and an illegible signature without a name. A week after the date of the purchase order, Thomas Tribou, a consultant who had worked with Signalife, paid Signalife $50,000 for goods he expected to receive.
The second and third purchase orders, dated September 24, 2007 and October 4, 2007, respectively, reflected sales to a company called IT Healthcare. One order reflected a sale of products at a cost of $3.3 million and noted a $30,000 deposit. The other reflected a sale with a “net due” amount of $551,500.
The facts of these purchase orders resurfaced several times. Mr. Harmison incorporated information about them in a March 2008 memorandum to Signalife’s auditors. Likewise, Signalife filed reports with the Securities and Exchange Commission (“SEC”) that detailed these orders. According to Mr. Woodbury, who oversaw the drafting of the SEC filings, Mr. Stein was the sole source of information about the purchase orders and was intimately involved in the drafting process.
Mr. Stein used the help of his personal assistant, Mr. Carter, and a Signalife contractor, Mr. Anand, to make the fake purchase orders appear legitimate. For example, Mr. Stein gave Mr. Carter a template to create bogus letters requesting a change of shipment address, one for IT Healthcare and another for CHM. Mr. Carter drafted a letter ostensibly from a man named Yossie Keret of IT Healthcare requesting that products be delivered to an address in Israel that Mr. Carter made up. Mr. Carter also prepared a letter appearing to come from CHM.that asked for products to be delivered to an address in Tokyo, Japan. This letter purportedly was signed by “Toni Nonoy.” Mr. Carter never spoke with Yossie Keret, Toni Nonoy, or anyone at IT Healthcare or CHM; indeed, he had no idea whether the companies or the individuals actually existed. He believed, however, that Mr. Stein had fabricated these names.
Mr. Stein directed Mr. Carter to help him with the fraud in other ways as well. *1142Mr. Stein asked Mr. Carter for two numbers he could use as fax numbers for purchase confirmation letters from Yossie Keret and Toni Nonoy. Mr. Carter provided Mr. Stein with two numbers unaffiliated with either company or person. Then, in June 2008, Mr. Stein told Mr. Carter to fabricate a letter from Yossie Keret purporting to cancel IT Healthcare’s orders. Mr. Carter did as he was told and sent the letter to Mr. Woodbury. At one point, Mr. Stein arranged for Mr. Carter to travel to Israel ostensibly to find customers for Sig-nalife even though Mr. Carter had no business contacts there. On another occasion, Mr. Stein sent Mr. Carter to Japan with a sealed envelope in a plastic bag, instructing him to mail the envelope back to the United States while wearing gloves and then return home the same day.
Mr. Stein similarly relied on Mr. Anand for help in perpetrating the fraud. Once Mr. Stein asked Mr. Anand to travel to Texas to mail two IT Healthcare purchase orders to Signalife. When Mr. Anand asked whether the purchase orders were real, Mr. Stein responded that it did not matter. Mr. Anand declined to help, but later, on Mr. Stein’s request, he agreed to draft two letters that would appear to come from Yossie Keret on behalf of IT Healthcare. The first letter requested a shipping address change to an Israeli address. The second letter cancelled the Sig-nalife order. Mr. Anand sent these letters to Mr. Stein and Mr. Harmison.
Mr. Stein also used Carter and Anand to take money or stock from Signalife. At Mr. Stein’s direction, in January 2008, Mr. Carter executed an agreement with Signal-ife to provide consulting services, none of which he actually provided or was capable of providing. Pursuant to this agreement, Mr. Stein funneled money and Signalife stock from Signalife through Mr. Carter to himself. Mr. Stein also directed Mr. Carter to buy and sell Signalife stock and transfer most of the proceeds to him. Likewise, at Mr. Stein’s direction, Mr. Anand established “The Silve Group,” ostensibly to sell Signalife products in India. But Mr. Anand sold only one unit (in Mexico). Mr. Stein nonetheless arranged for Signalife to pay Mr. Anand more than one million shares for his work. Mr. Anand then gave Mr. Stein a “kickback ... [f]or the sweet deal [he] got from Mr. Stein.” Doc. 243 at 71.
On August 15, 2008, Signalife filed a Form 10-Q for the second quarter of 2008, which described the cancellation of an IT Healthcare purchase order. (GX159 at 22.) This was the first public disclosure arguably signaling to stock market participants that Signalife’s stock was overvalued based on the IT Healthcare purchase order, and thus, as the district court found, marked the end of the fraudulent period.
B. Procedural Background
1. The Investigation and Indictment
The SEC began investigating Signalife in 2009. During its investigation, the SEC amassed a database of about 200 million records produced by Signalife. In 2010, the United States Department of Justice (“DOJ”) began a criminal investigation of Mr. Stein. As a result of the DOJ’s investigation, a grand jury indicted Mr. Stein on charges of money laundering; mail, wire and securities fraud; conspiracy to commit mail and wire fraud; and conspiracy to obstruct justice. The indictment also charged that Mr. Stein obstructed justice by giving false testimony to SEC investigators. Mr. Stein’s two co-conspirators, Mr. Carter and Mr. Anand, also were indicted. Both pled guilty to conspiracy charges and testified against Mr. Stein at trial.
*11432.The Motion to Compel
Before trial, Mr. Stein sent the government nine letters requesting over 100 categories of documents, including documents in the SEC’s files. The DOJ refused to produce information that was “not in the possession of or known to the prosecution,” which included the documents in the SEC’s files. Mot. Compel Ex. B, Doc. 41-2 at 3. Mr. Stein responded with a motion to compel. The government opposed the motion, arguing that the DOJ lacked control over the SEC and that the DOJ and the SEC conducted no joint investigation. The magistrate judge denied the motion to compel as to documents “in the sole custody of the SEC, and which the DOJ is unaware of.” Doc. 63 at 2.
3.The Pretrial Motion to Dismiss the Indictment
About two months before trial, at Mr. Stem’s direction, his attorney filed a motion to withdraw as counsel, which was granted. Mr. Stein then filed a motion to proceed pro se. The court held a Faretta5 hearing and then granted Mr. Stein’s motion. During the hearing, Mr. Stein learned that in the course of its investigation the DOJ had accessed a “very small subset” of documents in the SEC’s database, which the DOJ had then provided to him. Tr. of Faretta Hrg. Proceedings, Doc. 146 at 41. Based on this revelation, Mr. Stein promptly filed a pro se motion to dismiss the indictment, alleging the suppression of unidentified “Brady material” in the SEC database. Mot. to Dismiss, Doc. 150 at 17-22. Mr. Stein also requested an evidentiary hearing. The district court denied the motion, concluding, among other things, that the motion was untimely and failed to identify any exculpatory Brady material.
4.The Trial and Post-Trial Motions
The trial lasted two weeks. The jury returned guilty verdicts against Mr. Stein on all charges.
Mr. Stein filed several post-trial motions, including two motions for new trial based on newly discovered evidence. The newly discovered evidence included, among other documents, a publicly-filed SEC Form 8-K (“Exhibit X”) regarding an unrelated company whose Chief Financial Officer was named “Yossi Keret.” Mot. for New Trial Ex. J, Doc. 264-10. Mr. Stein alleged that Exhibit X was on the “SEC website.” See Mot. for New Trial, Doc. 264 at 9. Mr. Stein argued this document proved that Yossie (with an “e”) Keret, the man who purportedly signed the IT Healthcare purchase orders, was a real person, contrary to the government’s representation at trial. He contended that his conviction thus “was based on the perjured testimony of key Government witnesses and exclusion of crucial exculpatory and impeachment evidence as a result of prose-cutorial misconduct.” Id. at 1; see also 2d Mot. for New Trial, Doc. 312 at 2, 8-9. Mr. Stein also filed a motion for an evidentiary hearing on his motions for new trial and a motion to compel documents from the SEC database. The district court summarily denied these motions.
A little more than a year after the trial, in an SEC enforcement action against Sig-nalife’s successor company, the SEC produced about two million documents from its database. Within this collection, Mr. Stein found a copy of Exhibit X, the publicly-available SEC document containing the name ‘Yossi Keret.” Based on this document, Mr. Stein filed a third motion for new trial and accompanying motion for a hearing, arguing that the document was *1144exculpatory and had been withheld in violation of Brady.
The district court denied the third motion for a new trial and the corresponding motion for an evidentiary hearing. The court found that there had been “no showing that the person named Yossi Keret’ in [Exhibit X was] the same person connected to the [IT Healthcare purchase order confirmation and purchase order cancellation] upon which [Defendant’s convictions] ... are based.” Doc. 388 at 2. The court further found there was no evidence showing that the prosecution team possessed this document and knowingly withheld it.
5. The Sentencing
Before Mr. Stein’s sentencing, the probation office issued a presentence investigation report (“PSI”). Under the applicable Sentencing Guidelines, the PSI calculated a base offense level of 7 and recommended several enhancements and one reduction. Relevant to this appeal, the PSI recommended a 24-level increase under U.S.S.G. § 2Bl.l(b)(l)(M) based on a loss calculation of more than $50 million but less than $100 million. Mr. Stein objected to this proposed calculation of loss, contending that there was no actual loss to any investor.
The government proposed a method for calculating actual loss coined the “buyer’s only” method, which was based on actual purchase and sales data. Tr. of Sentencing Proceedings, Doc. 429 at 30. Under this method, the court would consider only “those customers who only purchased Sig-nalife shares during the fraudulent period,” defined as September 20, 2007 (the date of the first false press release) through August 15, 2008 (the date of Sig-nalife’s SEC filing noting that IT Healthcare had cancelled its purchase order). Tr. of Sentencing Proceedings, Doc. 428 at 25. The court would then “value the amount of those purchases ... [and] subsequently subtract the value of those shares as of the end of the fraudulent period.” Id. at 42. The government identified 2,415 unique investors who bought Signalife stock during the fraudulent period and subsequently lost a total of $13,186,025.85.6
Mr. Stein objected to this method, contending that the government needed to show both “but for” causation (reliance) and proximate causation (“that the causal connection between the conduct and the loss is not too attenuated”). Doc. 428 at 220. As regards “but for” causation, Mr. Stein argued there was no evidence that the 2,415 investors actually relied on false press releases or other fraudulent information promulgated by Mr. Stein. He noted that only one investor testified at trial that he had relied on one of Mr. Stein’s false press releases and only one investor provided a victim impact statement to the same effect. Although Mr. Stein acknowledged that a number of other investors provided victim impact statements, he emphasized that none of these investors specified that he or she relied on the false information he released.
The government responded that many of the victims’ impact statements showed they relied on press releases generally (albeit not necessarily the specific press releases Mr. Stein disseminated) in purchasing Signalife stock. The government urged that this evidence was enough to infer reliance for all 2,415 investors identified. The government also relied on testimony that the only source for information about Signalife stock was press releases and *1145public filings, and at least some investors probably relied on this type of information.
Regarding proximate cause, Mr. Stein argued that the government needed to “take into account ... extrinsic market factors.” Doc. 428 at 221. He noted that other circuits require this and that the Sentencing Guidelines specifically contemplate it. He identified specific events unrelated to the fraud that he contended caused the stock price to decline during the fraudulent period, including the 2008 financial crisis and the rampant short selling of Signalife stock. Mr. Stein urged the district court to reject the government’s actual loss calculation because it failed to tease out these external market factors. The government responded simply, “The offense [Mr. Stein committed] was luring people in to invest in this stock.... Did they then lose money? Of course. Was that reasonably foreseeable to Mr. Stein? Of course, it was. That’s the Government’s position here, Your Honor.” Id. at 242.
The district court adopted the buyer’s only method over Mr. Stem’s objections. It concluded that there was “sufficient evidence to demonstrate both reliance and causation of damage to the shareholders.” Doc. 429 at 30. Based on over $13 million in actual loss, the court applied a 20-level increase to Mr. Stein’s base offense level. See U.S.S.G. § 2B1.1(b)(1)(K).7 The court also imposed a 6-level enhancement because there were more than 250 victims. See id. § 2B1.1(b)(2)(C). With other enhancements and reductions not at issue here, Mr. Stein’s total offense level was 45, resulting in an advisory guidelines sentence of life imprisonment. The district court found that this range was “certainly way above what would be sufficient but not greater than necessary to comply with the requirements of [18 U.S.C. § ] 3553,” Doc. 429 at 70, and varied downward, sentencing Mr. Stein to 204 months’ imprisonment.
The government then filed a motion for judgment of restitution, asking the district court to use the same actual loss figure to award $13,186,025.85 to 2,415 Signalife investors. Mr. Stein waived his right to a hearing but filed a response arguing, again, that the government failed to prove reliance and proximate cause. The district court rejected this argument and granted the government’s motion. This appeal followed.
II. DISCUSSION
A. The Conviction Issues
Mr. Stein argues that the government violated Brady and Giglio, and thus the district court erred in denying his motions for a new trial. We review de novo alleged Brady or Giglio violations. United States v. Brester, 786 F.3d 1335, 1339 (11th Cir. 2015); United States v. Jones, 601 F.3d 1247, 1266 (11th Cir. 2010). We review the district court’s denial of a motion for new trial for an abuse of discretion. United States v. Vallejo, 297 F.3d 1154, 1163 (11th Cir. 2002). As explained below, we find no basis for vacating Mr. Stein’s convictions,
1. The Brady Claims
Mr. Stein first argues that the government’s failure to produce material, exculpatory evidence contained in the SEC’s database violated- Brady. “[T]he burden to show a Brady violation lies with the defendant, not the government....” United States v. Esquenazi, 752 F.3d 912, 933 (11th Cir. 2014). To establish a Brady violation, Mr. Stein must show that:
*1146(1) the government possessed favorable evidence to the defendant; (2) the defendant does not possess the evidence and could not obtain the evidence with any reasonable diligence; (3) the prosecution suppressed the favorable evidence; and (4) had the evidence been disclosed to the defendant, there is a reasonable probability that the outcome would have been different.
Vallejo, 297 F.3d at 1164.
Mr. Stein argues that the government violated Brady by failing to disclose Exhibit X, a document filed with the SEC showing that a person named “Yossi Ker-et” (not Yossie with an “e”) was an officer of a company unrelated to any of the players in this case. According to Mr. Stein, this document suggests that Yossie Keret, the man who purportedly signed the IT Healthcare purchase orders, was a real person.8
Mr. Stein’s argument fails for two reasons. First, Exhibit X contains no information favorable to Mr. Stein. Evidence is favorable to the accused for Brady purposes if “ ‘it is either exculpatory or impeaching.’ ” United States v. Naranjo, 634 F.3d 1198, 1212 (11th Cir. 2011) (quoting Stephens v. Hall, 407 F.3d 1196, 1203 (11th Cir. 2005)). Exhibit X is neither. Contrary to Mr. Stein’s contention, Exhibit X does not contradict Mr. Carter’s testimony that Yossie Keret was a fabricated name and not an officer of IT Healthcare. Not only is the name Yossi Keret on Exhibit X spelled differently from the name Yossie Keret on some of Mr. Stein’s fabricated documents, but also Exhibit X indicates that Yossi Keret is affiliated with a different company, not IT Healthcare. Thus, the district court’s conclusion that Mr. Stein had made “no showing that the person [referenced in Exhibit X was] the same person connected to the wires upon which [Defendant’s convictions] ... are based,” Doc. 388 at 2, was not erroneous. Mr. Stein failed to prove that Exhibit X was exculpatory or impeaching; thus, this document cannot be the basis of a Brady violation.
Second, even if Exhibit X were favorable to Mr. Stein, he failed to show that he was unable to locate it with reasonable diligence. “‘[T]he government is not obliged under Brady to furnish a defendant with information which he already has or, with any reasonable diligence, he can obtain himself.’” United States v. Valera, 846 F.2d 923, 928 (11th Cir. 1988) (quoting United States v. Prior, 646 F.2d 1254, 1259 (5th Cir. 1977)); see, e.g., United States v. Hansen, 262 F.3d 1217, 1235 (11th Cir. 2001) (holding that the government’s failure to disclose court opinions, which “were all available through legal research,” does not violate Brady). Mr. Stein conceded that Exhibit X was a publicly available document filed with a public agency. Although in some cases a publicly available document practically may be unobtainable with reasonable diligence, see, e.g., Milke v. Ryan, 711 F.3d 998, 1017-18 (9th Cir. 2013),9 Mr. Stein made no effort to estáb-*1147lish that this is such a casé. In fact, Mr. Stein represented that he located the document on the “SEC website.” See Mot. for New Trial, Doc. 264 at 9. For these reasons, Mr. Stein failed to satisfy his burden of proving a Brady violation based on Exhibit X.10
2. The Giglio Claims
Mr. Stein next argues that the government violated Giglio by knowingly relying on false testimony. “Giglio error, a species of Brady error, occurs when the undisclosed evidence demonstrates that the prosecution’s case included perjured testimony and that the prosecution knew, or should have known, of the perjury.” Ford v. Hall, 546 F.3d 1326, 1331 (11th Cir. 2008) (interpal quotation marks omitted). Giglio also applies where the prosecutor herself made “explicit factual representations” to the court or “implicit factual representations to the jury,” knowing that those representations were false. United States v. Alzate, 47 F.3d 1103, 1110 (11th Cir. 1995).
“To prevail on a Giglio claim, a [defendant] must establish that (1) the prosecutor knowingly used perjured testimony or failed to correct what he subsequently learned was false testimony; and (2) such use was material i.e., that there is any reasonable likelihood that the false testimony could have affected the judgment.” Ford, 546 F.3d at 1331-32 (internal quotation marks and ellipses omitted); accord Guzman v. Sec’y, Dep’t of Corr., 663 F.3d 1336, 1348 (11th Cir. 2011). “‘The could have standard requires a new trial unless the prosecution persuades the court that the false testimony was harmless beyond a reasonable doubt.’ ” Guzman, 663 F.3d at 1348 (quoting Smith v. Sec’y, Dep’t of Corr., 572 F.3d 1327, 1333-34 (11th Cir. 2009)). Thus, “Giglio’s materiality standard is more defense-friendly than Brady’s.” Id. (internal quotation marks omitted).
In addition, because Giglio error is a type of Brady violation, the defendant generally must identify evidence the government withheld that would have revealed the falsity of the testimony. See, e.g., Ford, 546 F.3d at 1331 (emphasizing that Giglio error “occurs when the undisclosed evidence demonstrates' that the prosecutor’s case included perjured testimony” (emphasis added) (internal quotation marks omitted)). In other words, “[t]here is no violation of due process resulting from prosecutorial non-disclosure of false testimony if. defense counsel is aware of it and fails to object.” Routly v. Singletary, 33 F.3d 1279, 1286 (11th Cir. 1994) (holding that because defense counsel was aware that a false statement was subject to impeachment and yet failed to object to the statement, there was no due process violation under Giglio). But where the government not only fails to correct materially false testimony but also affirmatively capitalizes on it, the defendant’s due process rights are violated despite the government’s timely disclosure of evidence showing the falsity. See DeMarco v. United States, 928 F.2d 1074, 1076-77 (11th Cir. 1991) (finding prosecutorial misconduct warranting a new trial despite no suppression of evidence where the prosecutor not only failed to correct false testi*1148mony, but also capitalized on the false testimony in closing argument); United States v. Sanfilippo, 564 F.2d 176, 178-79 (5th Cir. 1977) (same).
Mr. Stein identifies several categories of statements he contends were false, but none of them supports a Giglio violation, and only two merit discussion: (1) statements the prosecutor made to the court and during his closing argument regarding Thomas Tribou and (2) testimony of Ms. Jones and Mr. Woodbury about the bogus purchase orders.
a. Thomas Tribou
Mr. Stein first argues that the government knowingly made false representations to the court about Thomas Tribou—a Signalife consultant who paid the company $50,000 shortly after the date of the CHM purchase order—and then relied on that false representation in its closing argument in violation of Giglio. Specifically, Mr. Stein points us to two allegedly false representations the government made to the district court and one made to the jury. This argument fails because the government made no material false representations.
Mr. Stein’s argument as it pertains to all three representations arises out of his attempt near the end of trial to admit into evidence a copy of an October 24, 2007 email from Signalife’s CEO’s administrative assistant, Ms. Jones, to Signalife’s certified public accountant, Norma Provencio, which was forwarded to Signalife’s corporate counsel, Mr. Woodbury. The subject line of the email said, “[Fwd: Emailing: Tribou Payment],” and in the body, Ms. Provencio noted, “Attached is the $50K deposit on the 9-14 purchase order.” Am. Resp. in Opp. to Def.’s Mots, for New Trial Ex. 1, Doc. 298-1 at 37. The exhibit also included a copy of the referenced September 27, 2007 cheek for $50,000 to Signalife, apparently signed by Delores Tribou out of an account shared with her husband, Thomas. The check displayed the CHM purchase order number on the memo line, along with the words “Tribou & Assoc.” Doc. 298-1 at 38.
Mr. Stein sought to use this exhibit to support the inference that the September 14, 2007 CHM purchase order, which called for a $50,000 deposit, was legitimate. The government objected on the ground that the email’s contents were hearsay. The district court sustained the objection and noted that Mr. Stein failed to authenticate the document. The court ultimately brokered the following stipulation: “On or about September 27th, 2007, an individual named Thomas Tribou paid Signalife $50,000 for goods he expected to receive.” Mr. Stein, through counsel, accepted this stipulation, which was presented to the jury. Mr. Stein did not call Mr. Tribou as a witness.
After the district court sustained the government’s hearsay objection, the government made two representations to the court that Mr. Stein argues were false. First, the government represented that, based on interviews Mr. Tribou previously had given to SEC investigators, if Mr. Tribou were called to testify he would say that although he paid $50,000 to Signalife, he never received any product and was not a Signalife reseller.11 Mr. Stein argues that this representation is inconsistent with statements Mr. Tribou made to SEC investigators admitting that he signed the CHM purchase order.
*1149We reject this argument. Mr. Tribou’s statement to SEC investigators that he signed the CHM purchase order in no way indicates he would have testified that he actually received Signalife products. Nor does it show that Mi'. Tribou considered himself a Signalife reseller. And, in any case, Mr. Tribou’s SEC testimony was, as Mr. Stein himself characterized it, “extremely inconsistent.” Doc. 247 at 55. On this record, we cannot conclude that the prosecutor spoke falsely when he told the district court how he believed Mr. Tribou would testify at trial.
Second, on the district court’s request, the government privately telephoned Mr. Tribou and then relayed to the court and the defense the contents of that telephone call, which, according to Mr. Stein, included a false statement. The government told the court that during the call Mr. Tribou never denied giving Signalife a $50,000 check, but he said that he was unfamiliar with Tribou & Associates and that he doubted he wrote the purchase order number on the check. Mr. Tribou previously had told an SEC investigator that Tribou & Associates was his name “for consulting and everything on [his] personal taxes.” 2d Mot. for New Trial Ex. A, Doc. 312-1 at 8. Thus, Mr. Stein argues, the government knew or should have known that Mr. Tribou was lying about his unfamiliarity with Tribou & Associates and yet relayed the lie to the court nonetheless.
We reject Mr. Stein’s argument about the second representation for two reasons. First, Mr. Stein contends not that the prosecutor misrepresented what Mr. Tribou told him on the call, but rather that the prosecutor should have flagged for the court the inconsistency between what Mr. Tribou said on the call and what he had said to SEC investigators in the past. But it is well-established that “a prior statement that is merely inconsistent .with a government witness’s testimony is insufficient to establish prosecutorial misconduct.” United States v. McNair, 605 F.3d 1152, 1208 (11th Cir. 2010) (collecting cases); accord Hays v. Alabama, 85 F.3d 1492, 1499 (11th Cir. 1996) (holding that there was no due process violation arising out of a witness’s inconsistent testimony where there was “no showing that [the witness’s] later, rather than earlier, testimony was false”).
Second, even if false, the government’s representation regarding Mr. Tribou was immaterial. A material misrepresentation occurs when there is any reasonable likelihood that the false testimony could have affected the judgment. Guzman, 663 F.3d at 1348. Mr. Stein argues that the representation influenced the court’s decision to sustain the government’s objection on hearsay grounds to the admission of the check and the email. We disagree. The court sustained the objection before the government made the representations about. Mr. Tribou. Moreover, the court based its ruling on hearsay grounds and Mr. Stein’s failure to authenticate the documents rather than anything Mr. Tribou might say if called to testify. Mr. Stein fails to explain how the government’s statements had any bearing on this eviden-tiary decision, which Mr. Stein expressly does not challenge on appeal.
The third allegedly false statement occurred during the government’s closing argument. The prosecutor told the jury that the CHM purchase order was “all made up” and “fake,” statements Mr. Stein argues constituted misrepresentations because Mr. Tribou signed the purchase order and paid Signalife $50,000. Doc. 248 at 34. But the prosecutor’s statement and these facts are not mutually exclusive. The fact that Mr. Stein obtained Mr. Tribou’s signature and check *1150does not rule out the possibility that he also fabricated the purchase order. Indeed, the government made this argument in its rebuttal, stating that regardless of any signatures Mr. Stein obtained, the purchase orders were fake. Moreover, the record contained overwhelming evidence that Mr. Stein fabricated supporting documentation for the purchase orders and used arbitrary names for companies and individuals supposedly purchasing Signal-ife products. On this record, we cannot conclude that the government violated Giglio with its characterization of evidence about the CHM purchase order.12 See Maharaj v. Sec’y for Dep’t of Corr., 432 F.3d 1292, 1313 (11th Cir. 2005) (“In the Giglio context, the suggestion that a statement may have been false is simply insufficient; the defendant must conclusively show that the statement was actually false.”).
In sum, Mr. Tribou’s previous inconsistent statements to SEC investigators and the ambiguity regarding his role in signing the CHM purchase order and paying $50,000 to Signalife provide an insufficient basis for us to conclude that the government knowingly relied on materially false testimony.
b. Jones and Woodbury
Mr. Stein next argues that (1) Mr. Harmison’s assistant, Ms. Jones, lied when she characterized the three purchase orders as “phantom purchase orders” simply because she lacked supporting documentation, and (2) Signalife’s securities lawyer, Mr. Woodbury, lied when he said he got all his information about the purchase orders from Mr. Stein. Again, Mr. Stein relies on the October 24, 2007 email and the copy of the $50,000 Tribou check, which was received by Ms. Jones and Mr. Woodbury, as demonstrating these lies. But Mr. Stein offers no argument that the prosecutor capitalized on the allegedly false testimony that contradicts this evidence, which he needed to show because none of this evidence was suppressed.13 In fact, the record shows that Mr. Stein located the email and the check before trial and even produced them to the government. In the absence of government suppression of the evidence, then, there can be no Giglio violation. See Ford, 546 F.3d at 1331; DeMarco, 928 F.2d at 1076. Accordingly, we reject Mr. Stein’s Giglio argument.14
*11513. Mr. Stein’s Remaining Arguments
Mr. Stein argues that the district court erred when it denied (1) the third motion for new trial without considering the alleged prosecutorial misconduct cumulatively and (2) the motions to compel discovery and for an evidentiary hearing regarding the alleged Brady and Giglio violations. We review these denials for an abuse of discretion. See Vallejo, 297 F.3d at 1163 (motion for new trial); United States v. Schlei, 122 F.3d 944, 990 (11th Cir. 1997) (evidentiary hearing); Holloman v. Mail-Well Corp., 443 F.3d 832, 837 (11th Cir. 2006) (motion to compel discovery). Because there were no Brady or Giglio violations, there was no cumulative reversible error. See United States v. Carter, 776 F.3d 1309, 1330 (11th Cir. 2015). And Mr. Stein has failed to show how the district court’s decision not to hold a hearing and compel discovery was an abuse of discretion.15 We find no basis for vacating his conviction in Mr. Stein’s remaining arguments. Accordingly, we affirm his conviction and move on to his sentence.
B. The Sentencing Issues
Mr. Stein raises several challenges to his sentence, only one of which warrants discussion. Mr. Stein asserts that the district court erred in calculating actual loss for purposes of U.S.S.G. § 2Bl.l(b)(l) and for restitution under the MVRA. The district court’s actual loss calculation was premised on an estimate of losses suffered by 2,415 investors in Signalife stock during the fraudulent period. Mr. Stein argues that the actual loss calculation was too high because the court (1) presumed, without an adequate factual basis, that each investor relied on fraudulent information he disseminated and (2) failed to take into account intervening events that led to a decline in the price of Signalife stock.16
“We review a district court’s interpretation of the Sentencing Guidelines de novo, and the determination of the amount of loss involved in the offense for clear error.” United States v. Maxwell, 579 F.3d 1282, 1305 (11th Cir. 2009). A district court’s determination that a person or entity was a victim for purposes of loss calculation is an interpretation of the guidelines, so we review it de novo. United States v. Martin, 803 F.3d 581, 593 (11th Cir. 2015). A district court’s determination of proximate cause, however, is part of the court’s determination of the amount of loss involved in the offense and, thus, is reviewed only for clear error. Id. “We will overturn a court’s loss calculation under the dear-error standard where we are left with a definite and firm conviction that a mistake has been committed.” United States v. Campbell, 765 F.3d 1291, 1302 (11th Cir. 2014) (internal quotation marks omitted).
First, we provide an overview of loss calculation principles for purposes of the Sentencing Guidelines and restitution under the MVRA. Then we consider Mr. Stein’s arguments regarding reliance (fac*1152tual causation) and intervening events (legal causation).
1. Loss Calculation under the Guidelines and the MYRA
Section 2Bl.l(b)(l) of the Sentencing Guidelines provides a table for determining the level of enhancement based on the loss attributable to the offense. This loss calculation “serves as a proxy for ‘the seriousness of the offense and the defendant’s relative culpability.’ ” Campbell, 765 F.3d at 1301 (quoting U.S.S.G. § 2B1.1 cmt. background). In financial fraud cases, the loss calculation often drives the sentence. See, e.g., United States v. Olis, 429 F.3d 540, 545 (5th Cir. 2005) (“The most significant determinant of [the defendant’s] sentence is the guidelines loss calculation.”); United States v. Robles, No. CR 04-1594(B)SVW, 2015 WL 1383756, at *5 (C.D. Cal. Mar. 19, 2015) (“[T]he loss calculation in this case is the primary driver behind the Guidelines range—more than doubling the offense level and tripling the suggested sentence.... ”); United States v. Faulkenberry, 759 F.Supp.2d 915, 928 (S.D. Ohio 2010) (“[T]he harsh sentence recommended by the Guidelines is primarily driven by the loss calculation, which increases [the defendant’s] Base Offense Level by 30 points.”).
There are two ways to measure loss under U.S.S.G. § 2B1.1, actual and intended loss, and we are instructed to take the greater of the two. U.S.S.G. § 2B1.1, cmt. n.3(A). Here, however, the government did not argue for an intended loss calculation; we thus focus on the calculation of actual loss.
The government bears the burden of proving by a preponderance of the evidence actual loss attributable to the defendant’s conduct. United States v. Rodriguez, 751 F.3d 1244, 1255 (11th Cir. 2014), “[A] sentencing court is not generally required to make detailed findings of individualized losses to each victim.” United States v. Orton, 73 F.3d 331, 335 (11th Cir. 1996) (considering the similar predecessor guideline, U.S.S.G. § 2F1.1). Instead, the court may employ a variety of methods to derive a “reasonable estimate of the loss” to the victims based on the information available to the district court. United States v. Snyder, 291 F.3d 1291, 1295 (11th Cir. 2002); accord United States v. Ford, 784 F.3d 1386, 1396 (11th Cir. 2015); see also U.S.S.G. § 2B1.1 cmt. n.3(C)(iv) (providing that district courts should “tak[e] into account, as appropriate and practical under the circumstances,” a variety of factors including the “approximate number of victims multiplied by the average loss to each victim”). Although the district court may estimate the amount of loss, it cannot “speculate about the existence of facts and must base its estimate on reliable and specific evidence.” Ford, 784 F.3d at 1396; accord United States v. Sepulveda, 115 F.3d 882, 890-91 (11th Cir. 1997).
Under the guidelines, “[a]ctual loss ... is defined as the ‘reasonably foreseeable pecuniary harm that resulted from the offense.’ ” Campbell, 765 F.3d at 1302 (quoting U.S.S.G. § 2B1.1 cmt. n.3(A)(i)). This definition “incorporates [a] causation standard that, at a minimum, requires factual causation (often called ‘but for’ causation) and provides a rule for legal causation (ie., guidance to courts regarding how to draw the line as to what losses should be included and excluded from the loss determination).” U.S.S.G. App. C, Vol. II at 178, Amend. 617 (Nov. 1, 2001); see United States v. Evans, 744 F.3d 1192, 1196 (10th Cir. 2014) (“[Section] 2B1.1 incorporates and requires both factual or ‘but for’ causation and legal or foreseeable causation.”); United States v. Peppel, 707 F.3d 627, 643-44 (6th Cir. 2013) (recognizing *1153that, to establish actual loss under § 2B1.1, the government must “establish both cause in fact and legal causation by a preponderance of the evidence”); see also Burrage v. United States, — U.S. -, 134 S.Ct. 881, 887-91, 187 L.Ed.2d 715 (2014) (holding that the ordinary meaning of the term “results from” in a criminal statute requires “but-for causality”).
The MVRA requires the district court to calculate actual loss “to identifiable victims of certain crimes, including crimes of fraud.” Martin, 803 F.3d at 592. Under the MVRA, the district court must award restitution to such victims “without regard to the defendant’s ability to pay.” Id. The method for calculating actual loss, as opposed to intended loss, under the Sentencing Guidelines is “largely the same” as the method for establishing actual loss to identifiable victims under the MVRA. United States v. Cavallo, 790 F.3d 1202, 1239 (11th Cir. 2015). In most cases, the amount of actual loss under the guidelines will be the same as the restitution figure. Id. Thus, it is unsurprising that to prove a victim suffered an actual loss under the MVRA, the government must establish both factual and legal causation in essentially the same manner as it must show causation under the guidelines—by proving but for and proximate causation. See, e.g., Martin, 803 F.3d at 594; United States v. Robertson, 493 F.3d 1322, 1334-35 (11th Cir. 2007). Here the district court used the same figure for actual loss under the guidelines and the MVRA. Thus, we analyze the two calculations together, considering first factual and then legal causation.
2. Reliance (Factual Causation)
The parties agree that the government must show that the investors relied on Mr. Stein’s fraudulent information to satisfy the “but for” causation requirement under U.S.S.G. § 2B1.1. See also Currie v. Cayman Res. Corp., 835 F.2d 780, 785 (11th Cir. 1988) (“Reliance is ... a type of ‘but for’ requirement.” (quoting Huddleston v. Herman & MacLean, 640 F.2d 534, 549 (5th Cir. 1981), aff'd in part and rev’d in part, 459 U.S. 375, 103 S.Ct. 683, 74 L.Ed.2d 548 (1983))). The government also must show reliance to prove “but for” causation for restitution purposes. See Martin, 803 F.3d at 594. The parties disagree on what this showing must entail.
As we see it, the government may show reliance in a securities fraud case either through direct evidence or specific circumstantial evidence. The government may of course introduce individualized evidence of reliance—that is, direct evidence that each individual investor read the false information and relied on it when deciding to purchase stock. See United States v. Ebbers, 458 F.3d 110, 126-27 (2d Cir. 2006) (recognizing that reliance can be shown for loss calculation purposes under § 2B1.1 by offering evidence to demonstrate “express reliance on the accuracy of the [fraudulent] financial statements”). But, as the district court aptly recognized, requiring individualized proof of reliance for each investor is often infeasible or impossible. See Basic Inc. v. Levinson, 485 U.S. 224, 245, 108 S.Ct. 978, 99 L.Ed.2d 194 (1988) (recognizing in civil securities fraud context that requiring direct proof of reliance may be “an unnecessarily unrealistic evidentiary burden on the Rule 10b-5 plaintiff who has traded on an impersonal market”); Local 703, I.B. of T. Grocery & Food Emps. Welfare Fund v. Regions Fin. Corp., 762 F.3d 1248, 1253 (11th Cir. 2014) (same). Thus, in cases such as this one involving numerous investors, the government may instead offer specific circumstantial evidence from which the district court may reasonably conclude that all of the inves*1154tors relied on the defendant’s fraudulent information.
Here, though, the government failed to satisfy either of these options. As a result, the district court’s statement that “from the record that there [was] sufficient evidence to demonstrate ... reliance” for 2,415 investors was erroneous. Tr. of Sentencing Proceedings, Doc. 429 at 30. The record contains no direct, individualized evidence of reliance for each investor. And the circumstantial evidence in the record is far too limited to support a finding that 2,415 investors relied on the fraudulent information Mr. Stein disseminated. The only evidence arguably supporting the reliance finding was: (1) trial testimony from one investor that he relied on one of Mr. Stein’s false press releases; (2) a victim impact statement from another investor to the same effect; (3) a number of victim impact statements suggesting that the investors relied on press releases and other publicly available information generally, but not specifically the fraudulent information Mr. Stein disseminated; and (4) testimony that, because the only place to get information about Signalife stock was from press releases and public filings, at least some investors likely relied on this type of information. This evidence standing alone is insufficient to support the inference that all 2,415 investors relied on Mr. Stein’s fraudulent information when deciding to purchase Signalife stock. On this thin record, the district court “engage[d] in the kind of speculation forbidden by the Sentencing Guidelines.” United States v. Bradley, 644 F.3d 1213, 1292 (11th Cir. 2011); see Sepulveda, 115 F.3d at 890-91. Accordingly, the district court’s actual loss calculation was in error.
We therefore vacate Mr. Stein’s sentence, which was based on a guidelines calculation founded on the erroneous actual loss figure, and remand for a recalculation of actual losses. On remand, the government may again seek to prove actual loss by showing losses suffered by Signal-ife investors. Alternatively, the government may also seek to prove actual loss through direct losses to the company resulting from, for example, Mr. Stein’s theft of Signalife stock. See U.S.S.G. § 2B1.1 cmt. n.3(C)(i). And if the district court determines that the loss “reasonably cannot be determined,” the court may use instead “the gain that resulted from the offense.” Id. § 2B1.1 cmt. n.3(B).17
3. Intervening Events (Legal Causation)
We next turn to the requirement of legal causation, and, in particular, whether the district court erred in failing to take into account intervening events that may have contributed to investors’ losses. The standard for legal causation for purposes of the actual loss calculation is essentially the same under the guidelines and the MVRA. See Cavallo, 790 F.3d at 1239. Under the guidelines, “ ‘[a]ctual loss’ means the reasonably foreseeable pecuniary harm that resulted from the offense.” U.S.S.G. § 2B1.1 cmt. n.3(A)(i). A reasonably foreseeable pecuniary harm is one “that the defendant knew or, under the circumstances, reasonably should have known, was a potential result of the of*1155fense.” Id. § 2B1.1 cmt. n.3(A)(iv). Thus, the legal cause standard we use under § 2B1.1(b) is reasonable foreseeability.
We also consider reasonable foreseeability when assessing proximate cause for purposes of actqal loss under the MVRA. See, e.g., Martin, 803 F.3d at 594; Robertson, 493 F.3d at 1334-35. In Martin, the defendant fraudulently obtained loans that later were sold to successor lenders. Martin, 803 F.3d at 586-87. The district court relied on losses suffered by these successor lenders when estimating actual loss for restitution purposes. Id. at 592-93. We upheld the district court’s loss calculation, holding that the successor lenders could recover restitution under the MVRA because it “was entirely foreseeable to [the defendant] not only that the original lenders would rely on the fraudulent applications, but that the mortgages would be resold to other lenders that would rely on the applications as well.” Id. at 594. Put differently, because the intervening event—the sale of the loan to a successor lender—was reasonably foreseeable, it did not “break the chain of causation.” Id. (citing Robertson, 493 F.3d at 1334-35).18
In Robertson, in contrast, we vacated a restitution award because there was inadequate evidence to find that intervening events between the fraud and the loss were reasonably foreseeable. 493 F.3d at 1334-35. The defendant fraudulently obtained computer software from Novell, Inc. and then sold the software to Network Systems Technology, Inc. Id. at 1327-28. Network Systems resold the software at a profit. Id. at 1328. At some later point, Novell sued Network Systems in a case involving the software purchased from the defendant. Id. The record did not indicate the precise ground for the lawsuit. Id. Network Systems settled the lawsuit by agreeing to pay Novell $125,000. Id.
The district court determined that Network Systems was a victim for purposes of the MVRA, but we reversed. Id. at 1334-35. “Whether the lawsuit and settlement were reasonably foreseeable consequences of [the defendant’s] fraud on Novell,” we explained, “depends on the nature of the litigation.” Id. at 1335. All the government had established at sentencing, we noted, was “that the litigation was ‘related to’ the units of software” the defendant sold, and this “vague description” was insufficient to support the district court’s finding that the lawsuit and settlement were reasonably foreseeable. Id. Thus, we held that the district court erred in finding that Network Systems was a victim under the MVRA, and we vacated the $125,000 restitution award. Id. at 1335-36.
In sum, the causation standards for determining actual loss under the Sentencing Guidelines and for restitution purposes are similar. When calculating actual loss for either purpose, the district court should take into account intervening events contributing to the loss unless those events also were reasonably foreseeable to the defendant. See id. at 1334.
At sentencing, Mr. Stein urged the district court in arriving at its loss and restitution calculations to consider that Signalife stock value declined in part because of the short selling of over 22 million shares of Signalife stock and the across-the-board stock market decline of 2008.19 *1156The district court failed to consider these factors, and Mr. Stein argues that this was error. We agree.
Once Mr. Stein pointed to intervening events that may have affected the stock price, the district court was obliged to make findings regarding the effects of these intervening events, if any, and whether these events were reasonably foreseeable to Mr. Stein. Because the court failed to do so, we vacate the sentencing order. On remand, the district court should determine whether these intervening events affected Signalife’s stock price during the fraudulent period and, if so, whether they nonetheless were reasonably foreseeable to Mr. Stein. If the district court finds that these or any other intervening event reduced the value of Signalife stock during the fraudulent period and that the events were not reasonably foreseeable, the district court, to the extent possible, should approximate the effect of such intervening events and subtract this amount from its actual loss calculation.20
III. CONCLUSION
We affirm Mr. Stein’s judgment of conviction because we find no Brady or Giglio violations, but we vacate his sentence and remand to the district court with instructions to calculate anew the amount of loss for purposes of U.S.S.G. § 2B1.1(b)(1) and restitution under the MVRA, consistent with this opinion. To reiterate, this calculation may be an estimate so long as it is based “on reliable and specific evidence” rather than mere speculation. Ford, 784 F,3d at 1396. In particular, on remand, if the government seeks to prove an actual loss figure based on losses suffered by Signalife investors, the government must establish by a preponderance of the evidence that the investors relied on fraudulent information Mr. Stein disseminated. As regards intervening events, if Mr. Stein again offers evidence that a particular event aside from his fraud depressed the stock price during the fraudulent period, the district court must find, based on a preponderance of the evidence, that such intervening event was also reasonably foreseeable or, instead, subtract from the actual loss amount the monetary effect of such intervening event.
AFFIRMED in part, VACATED and REMANDED in part WITH INSTRUCTIONS.
. Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963).
. Giglio v. United States, 405 U.S. 150, 92 S.Ct. 763, 31 L.Ed.2d 104 (1972).
. Signalife was formerly known as Recom Managed Systems, Inc., and later known as Heart Tronics, Inc. Mr. Stein’s wife at the time of the false purchase orders, Tracey Hampton-Stein, was the founder of Signalife and the largest single Signalife shareholder. Thus, Mr. Stein stood to gain directly from the stock’s inflated price.
.“Doc." refers to the numbered entry onto the district court’s docket in this case. The trial transcript is found at Doc. 239 through Doc. 248.
. Faretta v. California, 422 U.S. 806, 95 S.Ct. 2525, 45 L.Ed.2d 562 (1975).
. The government proposed other methods for calculating actual loss, but the district court declined to adopt them.
. Under the applicable 2012 Sentencing Guidelines, a loss of more than $7 million but less than $20 million resulted in a 20-level enhancement. U.S.S.G. § 2B1.1(b)(1)(K).
. The only other document Mr. Stein identifies as supporting a Brady claim is a CHM change of address letter that Mr. Carter purportedly created. Oddly, this letter showed Mr. Carter's wife's uncle as the sender on behalf of CHM. It is unclear how this document could be considered exculpatory, but in any event it cannot support a Brady violation because the government produced the letter to Mr. Stein before trial.
. In Milke, the defendant's postconviction team of "approximately ten researchers ... spent nearly 7000 hours sifting through court records.” Milke, 711 F.3d at 1018. "The team worked eight hours a day for three and a half months, turning up 100 [relevant] cases.... Another researcher then spent a month reading motions and transcripts from those cases to find [the Brady material].” Id. The court held that no reasonably diligent lawyer could have found this material in time to use at *1147trial. Id.; see also United States v. Payne, 63 F.3d 1200, 1209 (2d Cir. 1995) (rejecting the argument that “the government’s duty to produce [an exculpatory document in its possession] was eliminated by that document’s availability in a public court file”).
. The government also argued that‘Exhibit X was not in its possession for Brady purposes. Because we reject Mr. Stem's Brady argument on other grounds, we do not reach this issue.
. The government also told the district court that Mr. Tribou likely would testify that he had no connection with CHM and that he agreed to Mr. Stem's request to sign a blank purchase order. Mr. Stein does not challenge these representations on appeal.
. Mr. Stein also argues that the prosecutor misrepresented the evidence when he asked the jury, "[I]f Tom Tribou, Thomas Tribou, is [CHM],' [then] where’s Tom Tribou’s name, Thomas Tribou’s name [on the purchase order]? ... Take a look closely-See if Thomas Tribou's name appears on there.” Doc. 248 at 114. Mr. Stein argues that Mr. Tribou’s name (in the form of his signature) does appear on the purchase order. But that was not the point of the government's argument. In fact, in closing, the government conceded that Mr. Stein may have obtained a signature on the CHM purchase order. The point—which was true—was that the purchase order did not identify Mr. Tribou as an officer of CHM.
. To be sure, the prosecutor mentioned in passing in his closing argument that Ms. Jones referred to the purchase orders as “phantom purchase orders,” but unlike in DeMarco, the prosecutor did not emphasize or capitalize on this statement by repeating it or making it the centerpiece of an argument for guilt. DeMarco, 928 F.2d at 1076-77 (noting that the prosecutor not only adopted the false statement but also emphasized it in her jury argument). Moreover, the prosecutor never mentioned Ms. Jones's statement that she received no backup for the purchase orders, which was the material aspect of her testimony.
.In support of his Brady and Giglio arguments, Mr. Stein filed a motion for the Court to take judicial notice of portions of a transcript from a summary judgment hearing in the SEC enforcement action against him, Heart Tronics, Inc., and various other defen- . dants. We GRANT this motion but find nothing in the transcript that changes our decision here.
. In a footnote in his opening brief, buried within his Brady argument, Mr. Stein makes a passing reference to an alleged violation of Rule 16 of the Federal Rules of Criminal Procedure. Such a passing reference, without any reasoned analysis whatsoever, is insufficient to preserve the argument on appeal. See United States v. Jernigan, 341 F.3d 1273, 1283 n.8 (11th Cir. 2003) (deeming issue abandoned where defendant made only passing references to it in brief). Accordingly, we do not address it. See id.
. Mr. Stein also challenges the district court’s estimate of the number of victims under U.S.S.G. § 2B 1.1(b)(2)(C), which resulted in an additional 6-level enhancement. This argument is intertwined with Mr. Stein’s § 2B1.1(b)(1) argument, and thus we do not address it separately.
. The government raises a harmless error argument, which we reject. According to the government, the district court could have calculated actual loss based on the value of assets Mr. Stein stole from Signalife or, if loss "reasonably cannot be determined,” U.S.S.G. § 2B1.1 cmt. n.3(B), by estimating Mr. Stein’s gain. Had the court used these alternative figures, the government argues, the Sentencing Guidelines range would have been the same. But the district court made no factual findings regarding the value of stolen assets or Mr. Stein's financial gain, and we will not make those findings in the first instance.
. We vacated the restitution award in Martin, however, because the district court failed to take into account the amount the successor lenders paid to acquire the mortgages. Martin, 803 F.3d at 595-96.
. Although Mr. Stein offered expert testimony regarding the stock market decline, it is unclear whether he offered proof that the short selling occurred or how it may have depressed stock prices.
, Mr. Stein also urges us to follow the lead of two of our sister circuits in importing the proximate cause principles from the civil fraud context, see Dura Pharm., Inc. v. Broudo, 544 U.S. 336, 125 S.Ct. 1627, 161 L.Ed.2d 577 (2005), into the sentencing context for purposes of calculating actual loss. See United States v. Rutkoske, 506 F.3d 170, 179 (2d Cir. 2007); United States v. Olis, 429 F.3d 540, 545-49 (5th Cir. 2005). We decline his invitation because we believe our reasonable foreseeability test strikes the right balance for calculating actual loss under the Sentencing Guidelines and for purposes of restitution.