United States v. Keith Berman

• Court: Court of Appeals for the D.C. Circuit
• Decided: July 25, 2025
• Docket: 24-3044
• Status: Published

Opinion

United States Court of Appeals FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued April 1, 2025 Decided July 25, 2025

No. 24-3044

UNITED STATES OF AMERICA, APPELLEE

v.

KEITH BERMAN, ALSO KNOWN AS MATTHEW STEINMANN, APPELLANT

Appeal from the United States District Court for the District of Columbia (No. 1:20-cr-00278-1)

José F. Girón argued the cause for appellant. With him on the briefs were Kevin B. Collins and Jonah Panikar. A. J. Kramer, Federal Public Defender, entered an appearance.

Ethan A. Sachs, Attorney, U.S. Department of Justice, argued the cause for appellee. With him on the brief was Jeremy R. Sanders, Assistant Chief and Appellate Counsel.

Before: KATSAS and GARCIA, Circuit Judges, and GINSBURG, Senior Circuit Judge.

Opinion for the Court filed by Circuit Judge GARCIA. 2 GARCIA, Circuit Judge: Keith Berman pleaded guilty to securities fraud, wire fraud, and obstruction of proceedings in connection with a scheme to fraudulently increase the share price of his company, Decision Diagnostics Corp. The district court sentenced Berman to 84 months’ imprisonment. On appeal, he challenges the district court’s calculation of the amount of loss caused by his fraud. We affirm. I A At the onset of the coronavirus pandemic in early 2020, Keith Berman saw a business opportunity for his struggling publicly traded medical device company (which we, like the parties, refer to using its stock ticker, DECN). In late February, he asked DECN’s South Korean supplier whether the company’s existing blood glucose monitor sticks might be used to detect the presence of coronavirus. The supplier responded that although that capability was “theoretically possible,” the supplier was “not sure” and “further study” would be required. App. 250. On March 3, without further investigation or confirmation that such a blood test could detect coronavirus, Berman began issuing press releases claiming that DECN had the “technology perfected” and that its finger-stick blood test would “be commercial ready in the summer of 2020.” App. 239. Through the following months, further press releases claimed that the blood test could detect the presence of coronavirus in less than a minute, that DECN had received positive signals of forthcoming approval from the FDA, and that the company was projected to sell as many as 525 million test kits in its first year of production. Following the announcements, the price of DECN’s stock (which sold over the counter as a penny stock) rose sharply. 3 The Securities and Exchange Commission took notice of DECN’s announcements and the spikes in its stock price. On April 23, 2020, the SEC suspended trading of the stock for ten days pending an investigation. An accompanying one-page announcement noted “questions regarding the accuracy and adequacy of information in the marketplace since at least March 3, 2020,” and highlighted several of the claims in Berman’s press releases. App. 260. Berman denied all wrongdoing in a sworn statement, and trading of DECN resumed. On May 20, the SEC publicly released a document laying out the basis for the April trading suspension. That document contained the SEC’s conclusion that while issuing his earlier press releases, Berman knew the company had no working prototypes and no FDA approval, and did not know how many kits DECN could produce. On July 10, Berman nevertheless released another statement touting that the finger-stick tests were “currently producing results,” and stating that the company was also testing a saliva test for coronavirus. App. 244. In light of questions over the veracity of DECN’s claims, by at least July 20, the penny-stock trading platform otcmarkets.com added a caveat emptor (buyer-beware) warning on DECN’s purchase page. Throughout that summer, Berman also used an alias to post more than a thousand messages in an online investor forum, continuing to promote the blood test and downplaying the SEC investigation as improper and misinformed. He also secretly orchestrated three purportedly independent shareholder letters to the SEC that again called the investigation into question; an illustrative passage accused the agency of trying to “destroy Mr. Berman, DECN, and by extension its shareholders.” Suppl. App. 211. 4 B In December 2020, a federal grand jury in the District of Columbia indicted Berman. Three years later, he pleaded guilty to three counts in a superseding indictment: securities fraud under 15 U.S.C. §§ 78j & 78ff, wire fraud under 18 U.S.C. §§ 1343 & 1342, and obstruction of proceedings under 18 U.S.C. §§ 1505 & 1502. The parties failed to reach a plea agreement because they did not agree on the value of the loss caused by Berman’s fraud. The district court held an adversarial sentencing proceeding on that issue. In cases concerning fraudulent inflation of securities, the Sentencing Guidelines instruct courts to determine the “loss” that “resulted from the offense,” and to use that value to determine the corresponding Guidelines range. U.S.S.G. § 2B1.1(b)(1)(C). Although the Guidelines permit a court to use “any method” that is “appropriate and practicable” to calculate loss, they recommend “calculating the difference between the average price of the security . . . during the period that the fraud occurred” and the average price “during the 90- day period after the fraud was disclosed to the market,” and then “multiplying the difference in average share price by the number of shares outstanding”—a method commonly called the modified rescissory method. Id. § 2B1.1 cmt. n.3(E)(ix). To assure that amount “is a reasonable estimate of the actual loss attributable” to the fraud, the Guidelines note that a court “may” subtract from that amount any “significant changes in value not resulting from the offense,” such as “changes caused by external market forces.” Id. The parties here agreed to calculate loss for sentencing using the modified rescissory method. After hearing from the parties’ experts, the district court defined the “period that the fraud occurred” as beginning with Berman’s first press release about the blood test on March 3, 5 2020, and ending on December 17, 2020, the day before the indictment was unsealed. The average stock price during that period was 19.54 cents per share. The 90-day period “after the fraud was disclosed to the market” began the next day (December 18, when his indictment was unsealed) and ended on March 17, 2021. The average price during that period was 2.13 cents per share. The difference in those average prices, 17.41 cents per share, multiplied by the 160 million outstanding shares, yielded a loss amount of $27.8 million. The district court credited the government expert’s testimony, based on econometric analysis, that the change in price was not attributable to any external market conditions. Based on that loss value, the district court applied a 22-level sentence enhancement. See id. § 2B1.1(b)(1). The district court also added a two-level enhancement for offenses that “involved sophisticated means,” and a four-level enhancement for offenses that “resulted in substantial financial hardship to five or more individuals.” Id. § 2B1.1(b)(10), (b)(2)(B). The resulting Guidelines range was 168 to 210 months’ imprisonment. After considering the mitigating and aggravating factors outlined in 18 U.S.C. § 3553, the district court granted a downward variance and imposed a sentence of 84 months. II Berman raises several challenges to the district court’s calculation of the applicable Sentencing Guidelines range and his sentence. We have jurisdiction under 18 U.S.C. § 3742(a), which provides that defendants may seek review of a sentence if they claim an “incorrect application of the sentencing guidelines.” 6 A Berman first challenges the district court’s determination that his fraudulent scheme was “disclosed to the market” when his indictment was unsealed on December 18, 2020. The district court concluded that Berman’s fraud was disclosed by the indictment because it was not until that point that “the full scope of the Defendant’s misconduct became known to the public.” App. 510. In particular, the court found that “a core component” of the charged criminal scheme was Berman’s ongoing (and not-yet-public) effort throughout the summer of 2020 “to persuade prospective investors to ignore the SEC, doubt its credibility and keep investing in DECN.” App. 511. We review those factual findings for “clear error” and will “affirm unless we are ‘left with the definite and firm conviction that a mistake has been committed.’” United States v. Brockenborrugh, 575 F.3d 726, 737–38 (D.C. Cir. 2009) (quoting United States v. U.S. Gypsum Co., 333 U.S. 364, 395 (1948)). The record amply supports the district court’s finding. As detailed above, through at least August 2020 Berman used aliases to post more than a thousand comments on investor message boards seeking to discredit the SEC’s investigation and encourage further investment. He also used an alias to orchestrate shareholder letters with similar content to the SEC in June, July, and August. None of those facts were public until the indictment was unsealed in December 2020. It was plainly appropriate for the district court to find that the fraud had not been “disclosed” until that key part of the ongoing fraud was made public. The district court further explained that the “objective evidence” also supported its determination that the fraud was disclosed only by the indictment. App. 514. DECN’s stock price in 2020 roughly followed a bell-curve shape, increasing 7 shortly after Berman’s first fraudulent statements and returning to its original level only after the indictment was unsealed. See Suppl. App. 362 (government exhibit depicting DECN’s stock price over time). Before the fraud began, the “stock was around two cents a share.” App. 514. During the period in which Berman conducted the charged fraud, the stock traded “significantly above that true fair market value.” Id. And after the indictment disclosed the fraud, “the price returned to that true fair market value of two cents per share.” Id. That bell-curve, the district court reasoned, provides “further support” for the December 18 fraud-disclosure date. Id. And the court reinforced that finding by crediting the government expert’s econometric analysis, which compared DECN’s price movements to those of a biotechnology stock index to show that the price freefall in December was not attributable to general market conditions. App. 512; see App. 303–07. Berman responds that the relevant disclosure date was one of three earlier days: April 23 (when the SEC temporarily suspended trading), May 20 (when it publicly described the key information underlying that suspension), or July 20 (when otcmarkets.com added the buyer-beware warning). Using any of those dates, Berman argues, would make the loss amount zero. Appellant’s Brief 19. Berman fails to show that the district court committed clear error in identifying December 18 as the date the fraud was disclosed. Most importantly, Berman barely acknowledges the district court’s focus on his ongoing, post-May fraudulent statements aimed at discrediting the SEC’s notices. He offers no explanation of why the district court erred as a factual matter in deeming his post-May fraudulent statements a “core component” of the fraud. Nor can he explain how his proposed disclosure dates—none of which prompted the DECN stock price to return to its pre-fraud baseline the way disclosure of the indictment did—correlate with the evidence 8 of stock-price fluctuation. Indeed, his expert’s only attempt to explain the failure of DECN’s stock price to drop after the SEC announcements was to guess that investors are “irrational,” App. 355, which the district court “d[id] not find . . . to be a credible answer to the Government’s evidence,” App. 513. For support, Berman relies on cases finding securities fraud adequately disclosed by SEC announcements similar to the May 20 announcement here. See Appellant’s Brief 29 n.9 (citing In re Merrill Lynch Auction Rate Sec. Litig., 851 F. Supp. 2d 512, 527 (S.D.N.Y. 2012); United States v. Peppel, 2011 WL 3608139, at *9 (S.D. Ohio Aug. 16, 2011); United States v. Ferguson, 584 F. Supp. 2d 447, 450, 454 (D. Conn. 2008)). But given the record in this case, the attempted comparison falls flat. None of those cases involved ongoing fraud that continued to discredit the SEC and inflate the stock price even after the SEC announcements. Indeed, the district court here explicitly acknowledged that “in the normal case,” an SEC notice like the May 20 notice here likely would have sufficed. App. 511. But, the court explained, this was the “unusual” case because “Berman’s own conduct undercut the value that [the SEC] announcements might otherwise have had.” App. 510–11. Berman also argues that the district court committed two legal errors in choosing the December date. First, he claims that the district court erred by starting the disclosure period only after the fraud was “fully” disclosed, App. 510, even though the Guidelines do not use the word “fully,” see U.S.S.G. § 2B1.1 cmt. n.3(E)(ix) (starting the disclosure period “after the fraud was disclosed to the market”). The suggestion is that the crux of Berman’s fraud was “disclosed” well before December and that the indictment added only minor details. As the foregoing demonstrates, that is not a fair characterization of the district court’s detailed ruling. Rather, the court determined with ample record support that in this case 9 a “core component” of the fraud was not disclosed until December. App. 511. Second, Berman argues that the district court proceeded as if the Guidelines require the 90-day disclosure period to begin immediately upon the conclusion of the fraud. Had it not made that legal error, Berman contends, it might have determined that the disclosure period began before the fraud ended. But the court did no such thing. As discussed above, the court’s ruling rested on its factual finding that a key portion of the fraud was not disclosed until December. We therefore do not agree that the court’s analysis turned on the legal issue Berman seeks to tee up for our review. B Berman’s second challenge is that the district court did not have sufficient evidence to conclude that his fraud caused the calculated loss. We review the district court’s application of the Guidelines’ loss calculation method to the facts with “due deference,” 18 U.S.C. § 3742(e), which “presumably falls somewhere between de novo and clearly erroneous” review, United States v. McCants, 554 F.3d 155, 160 (D.C. Cir. 2009) (citation modified)). 1 We find no reversible error in the district court’s loss-causation analysis. The parties agreed to use the modified rescissory method suggested by the Guidelines to calculate the “reasonably foreseeable pecuniary harm that resulted from the offense.”

1 As this court recognized in McCants, although the Supreme Court has held that the statutory origin of the due deference standard, 18 U.S.C. § 3742(e), is unconstitutional “insofar as it required courts to reverse sentences falling outside the applicable Sentencing Guidelines range, we have since held that this section continues to provide the standard by which we review a district court’s application of the Sentencing Guidelines.” 554 F.3d at 160 n.3. 10 U.S.S.G. § 2B1.1(b)(1)(C)(i). As described above, the modified rescissory method asks the court to calculate the difference in the average stock price between the fraud and disclosure periods, multiply that difference by the number of outstanding shares, and ensure that the loss attributable to the change in value was not caused by “external market forces, such as changed economic circumstances, changed investor expectations, and new industry-specific or firm-specific facts, conditions, or events.” Id. § 2B1.1 cmt. n.3(E)(ix). The modified rescissory method does not contemplate that any additional proof of causation is necessary to show the fraud caused the loss; by calculating the change in price, assessing the number of shares that price change affected, and ensuring that the change was not caused by external factors, this method is meant to demonstrate that the fraud satisfies the Guidelines’ causation requirement. The district court’s application of the modified rescissory method tracked what the Guidelines suggest. First, the district court determined as a matter of fact that the fraud period lasted from March 3, 2020, to December 17, 2020, and that the 90- day disclosure period began on December 18, 2020, and ended on March 17, 2021. Next, the district court accepted the government expert’s calculation of the loss amount, which took the difference in the average price during those respective periods (17.41 cents) and multiplied that by 160 million outstanding shares, yielding a loss value of $27.8 million. And finally, the district court determined that the changes in stock price did not result from “external market forces.” The court credited the government expert’s statistical determination that “the changes in DECN’s stock price were not the result of industry-specific fluctuations in the market or other financial trends.” App. 512. The expert ran a regression comparing DECN’s price fluctuations to stocks in the NASDAQ biotechnology index and concluded that the 11 “absence of a consistent statistically significant relationship between” the two indicated that DECN’s stock price “was not responding to industry-specific news concerning the biotech industry.” App. 306. The court also noted the “temporal proximity” between Berman’s fraudulent statements and spikes in the stock’s price. App. 515. And the court emphasized that the defense presented no other viable theory for the stock price’s behavior apart from Berman’s statements. App. 514. Those factors, the court concluded, further supported its finding that Berman’s fraudulent actions, not some other factor, “drove investor demand and thus caused investor losses.” App. 516. Berman insists that the record does not support the district court’s loss calculation. He argues that the government was required to offer additional evidence that each investor who was considered in the loss calculation relied specifically on his fraudulent statements. And the government, he continues, did not make that showing here. To support his assertion, Berman looks primarily to the Eleventh Circuit’s decision in United States v. Stein, 846 F.3d 1135 (11th Cir. 2017). We are not persuaded. Most directly, Stein did not involve an application of the modified rescissory method. Instead, the government in Stein relied on a “buyer’s only” method, which looked at specific customers who purchased shares while the fraud was ongoing and compared the price at which they did so to the value of the shares at the end of the fraud period. Id. at 1144. Here, by contrast, Berman specifically agreed to the application of the Guidelines’ suggested method. And as just explained, the district court applied that method exactly as the Guidelines describe it. In any event, the government’s evidence met the evidentiary bar set even in the different context addressed in Stein. There, the Eleventh Circuit concluded the government 12 could show the relevant purchasers relied on the defendant’s fraudulent statements through circumstantial evidence. The court observed that “requiring individualized proof of reliance for each investor is often infeasible or impossible.” Id. at 1153. But it still found the government’s indirect evidence “far too limited,” because it consisted of only one investor’s testimony and a handful of general victim impact statements. Id. at 1154. The government also did not account for “extrinsic market factors.” Id. at 1145 (citation omitted). As detailed above, however, the record here is more extensive. Indeed, following remand, the Eleventh Circuit upheld a renewed loss calculation that relied on evidence quite like the evidence the government offered here. That evidence included expert testimony showing that the stock-price fluctuations coincided with the fraudulent statements and the disclosure of the fraud, and a “statistical analysis” showing no correlation between general market movements and the stock at issue. See United States v. Stein, 964 F.3d 1313, 1320–21 (11th Cir. 2020). Berman raises one more attack on the district court’s causation analysis. He argues that the district court should have altered the loss amount to account for the possibility that some investors purchased DECN stock solely because of his claim that it was developing a distinct saliva-based test in addition to the blood-based test. Berman’s statements about the saliva-based test were not charged as fraudulent. So, the argument goes, any purchases based on claims about the saliva test should have been excluded from the loss calculation. As support, Berman points to the July 10, 2020, press release claiming DECN would provide the new “saliva testing kit option,” App. 244, and to two victim impact statements that reference the saliva test as motivating, at least in part, the victims’ investment decisions. 13 The district court rejected that argument on the basis that any impact on the stock price attributable to the saliva test could not be disentangled from the charged fraud—it was part and parcel of the “overall fraud scheme.” App. 519. Berman fails to show reversible error in that determination. Recall that, under the Guidelines, “the ‘loss need not be determined with precision,’ and ‘the court need only make a reasonable estimate of the loss, given the available information.’” United States v. Bras, 483 F.3d 103, 112 (D.C. Cir. 2007) (quoting U.S.S.G. § 2B1.1 cmt. n.3). With that standard in mind, the district court’s finding that there was no need to adjust the loss amount to account for the saliva test was reasonable. The only saliva-test reference that Berman identifies in the record—the July 10 announcement of that test—illustrates the district court’s point, because it was embedded in a press release again proclaiming, falsely, that DECN’s “finger stick kits are currently producing results.” App. 244. The district court’s approach also accords with the data recounted above, which showed the stock price plummeting when the blood-test fraud was disclosed. Moreover, Berman did not provide the district court with any reasonable method for deducting investor losses based on his single saliva-test reference from the losses caused by Berman’s voluminous blood-test statements. Nor has he reasonably explained his failure to do so on appeal.2 Against that backdrop, the possibility that some small portion of investors relied on the saliva test in making their decision to invest in DECN does not warrant remand.

2 Berman argues that he cannot be faulted for his failure to introduce such evidence because the district court excluded evidence related to the saliva test on relevance grounds. But Berman did not argue that the saliva test was relevant to any loss calculation or investor reliance; he argued only that his efforts to develop the saliva test might speak to whether he had engaged in a “good faith” or “genuine” effort to develop a functional product. App. 106. 14 C Berman next challenges the district court’s application of an enhancement for causing “substantial financial hardship to five or more victims.” U.S.S.G. § 2B1.1(b)(2); see also id. cmt. n.4(F). The district court found that the victim impact statements “leave[] no room for serious argument that the standard [for substantial financial hardship] is not met.” App. 517. Berman’s argument on this front mirrors his last challenge to the loss finding: Because some investors may have relied on the saliva test, the government was required to elicit more evidence that five or more investors relied specifically on the charged fraudulent statements. That argument fails for a more straightforward reason in this context. True, two victims mentioned the saliva test. But more than five explained that they had invested in DECN based on Berman’s fraudulent statements without ever mentioning the saliva test. App. 517– 18. Those facts suffice to support the district court’s application of the substantial-hardship enhancement. D Finally, Berman argues that his sentence was substantively unreasonable. This claim, however, expressly turns on his argument that the district court miscalculated the loss caused by his fraud. Because we find no reversible error in the district court’s loss-causation analysis, Berman’s substantive reasonableness challenge fails. III For the foregoing reasons, we affirm the district court’s judgment. So ordered.