John C. Coffee Jr. ()
Old frauds never die. Nor do they fade away. Rather, they mutate and morph into new configurations in response to new opportunities (which new technologies usually create). Thus, the traditional boiler room “pump and dump” scheme was a product of the widespread adoption of the telephone, which allowed high pressure salesman to reach hundreds of gullible customers in a day. Today, an analogous new technological development is inviting new forms of fraud. The new development is algorithmic trading (which by some estimates now accounts for 30 percent of stock trading1). Computers are programmed to trade in a micro-second once they detect certain triggering quantitative data. Obviously, this is how high frequency traders have come to dominate the market.
But can the computer be duped? The answer is: definitely and sometimes easily. A pending SEC litigation shows how the contemporary financial world in its hunt for quantitative “big data” exposes itself to fraudsters. In SEC. v. Lidingo Holdings,2 a pending action in the Southern District of New York, the defendant described itself as a “social media consultant” but the SEC characterized it instead as a “stock promotion firm” that received high fees for commissioning and posting articles (and even tweets) on their client firms, written by a variety of ghost writers that they commissioned and paid. These articles were usually posted under pseudonyms (such as “Trader Maven” or “Swiss Trader”) and appeared on reputable websites, such as Motley Fool, Benzinga, and Seeking Alpha. No attempt seems to have been made to directly contact investors (as in the traditional “pump and dump” scheme), but probably far more investors can be reached today through these web sites. According to the SEC’s complaint, this process generated “hundreds of bullish articles about public companies,” without ever disclosing that these articles were “paid promotions.” In one case, the SEC alleges that Lidingo drove up the stock price of Galena Biopharma, a small pharmaceutical company, by some 925 percent.3
This case is not unique, and the SEC has similarly sued more than a dozen companies, recently settling with a Florida-based company, called “The Dream Team Group LLC,” “that charged its clients $25,000 for 90 days of ‘social media relations.’”4 Even before these cases, predatory traders focused on algorithm trading through the practice of “spoofing”—that is, submitting thousands of buy or sell orders and then cancelling them before execution. The goal was to induce the algorithms of high frequency traders to submit similar or higher orders with which they could transact. The strategy worked (and it only cheated algorithm traders who could respond in a microsecond). Still, “spoofing” has now been successfully criminally prosecuted,5 which tends to chill its attraction.
The new practice of disseminating numerous ghost-written, anonymous articles has caught the attention of the press, but the press sees this as just another example of “fake news” dominating real news. That may miss the most interesting point here. Focus now on who is the intended target of this fraud. Although retail customer could also be deceived, the real target is the high frequency trader. In recent years, legitimate research firms have appeared that seek to cull quantitative data about investor sentiment and then disclose the results to their high frequency trading clients. One example is Social Media Analytics, a firm that reviews “millions of tweets a day to give ‘quantitative’ traders a sense of what direction a stock may be moving.”6 Of course, this is lawful, but now suppose a high frequency trader uses a hypothetical metric: 10 positive references to a stock in tweets or postings on certain web sites within a defined period will lead it to buy 10,000 shares of the stock. At this point, the potential impact of postings or tweets by Lidingo (or a similar firm) comes into full view. Their postings need not contain fraudulent misstatements; they need only express a positive opinion. If enough of these stories are posted within the relevant time frame, they will trigger the algorithm.
This brings us to the legal issue: Is this unlawful? If we look only at the statements themselves in these postings, close questions can arise as to their materiality. Use of an obvious pseudonym, such as “Trader Maven,” does not seem to be a material falsehood. But many of these postings were signed with an assumed name—say, “Arthur Rhoades, MBA.” Is it materially misleading to claim falsely that the writer has an MBA? Perhaps, the reasonable investor should know that such a credential does not mean terribly much. More disturbing is the fact that none of these postings disclosed that the author had been paid, directly or indirectly, by the issuer. Indeed, some postings denied this. That sounds more material.
But here we come to the SEC’s most potent weapon. Section 17(b) of the Securities Act of 1933 sets forth a strict and prophylactic anti-manipulation rule that makes it unlawful “to publish, give publicity to, or circulate any notice, circular, advertisement, newspaper, article, letter, investment service or communication, which … describes such security for a consideration received or to be received, directly or indirectly, from an issuer, underwriter or dealer, without fully disclosing the receipt, whether past or prospective, of such consideration and the amount thereof.”7 In short, this provision does not bar the receipt of consideration for publicity or published reports, but does require full disclosure of the amount so paid (at least when it comes from the issuer). Thus, the various firms that (like Lidingo) were receiving payments from issuers for several years were acting in flagrant defiance of this law. One wonders if they knew this or if they ever consulted a competent counsel. Certainly, the term “communication” in §17(b) is broad enough to cover a tweet, an email or any other posting. Frankly, these firms are fortunate that the U.S. Attorney did not prosecute, as §17(b) can be enforced civilly or criminally.
By now, a number of blogs and websites are aware of §17(b) and require that those posting on it must certify that they are not receiving compensation for their article or posting. This would protect the blog or website from any claim that it was aiding and abetting the author’s violation. Other websites need to clean up their act and impose closer controls, both to avoid SEC scrutiny and, even more, to protect themselves from reputational damage.
Although the law’s prohibition is clear when the issuer pays for the report or publicity, let’s next consider the consequences when other traders try to exploit those relying on quantitative research in order to move the stock price. Suppose these predatory traders also commission research that reaches strong conclusions, and they post dozens of articles or tweets. Now, §17(b) does not apply, because that provision covers only payments from an “issuer, underwriter, or dealer.” These traders could pay for ghost writers to post stories under pseudonyms and not violate §17(b). As a result, the question becomes whether such conduct amounts to a “pump and dump” scheme that violates either Rule 10b-5 or the rules under §17(a). Assume that our hypothetical trader takes precautions so that its ghost writers do not make objectively false statements, but only express broad statements of opinion. For example, “This stock is going to take off precisely as oil prices sink because …” Or, “Widget Corp’s price will collapse the moment the health care bill in Congress dies …” Or, “The consensus view among informed traders is that Widget’s price will soon explode.” This trader’s goal might be less to convince gullible retail investors than to dupe the algorithm. If this trader knows (or suspects) that a certain volume of reports stating opinions about a stock will cause high frequency traders to buy or sell automatically, he has an easy target—a “sitting duck” algorithm. Over time, high frequency traders may learn to revise their algorithms and employ subtler variations (but so may the manipulation-inclined trader improve its techniques).
In these cases, plaintiffs and the SEC can assert either or both of two theories: First, they will predictably claim that the payments to these ghost writers are material facts, which were not disclosed. Possibly, this may lead a smarter trader to use full-time employees who receive no special payment and are purportedly posting their own personal opinions. Second, the claim will be raised that the sheer volume of these reports, all emanating from the same source at about the same time, is “manipulative” within the meaning of §10(b) and Rule 10b-5 (or, in the case of the SEC, §17(a) and its rules). Here, a high volume of postings does seem suspicious. Fifty reports or tweets appearing in one week would be consistent with a manipulative intent (while three or four posts may not be). Similarly, there may be a difference between an obvious pseudonym (such as “Trader Maven”) where everyone realizes that the speaker’s identity is being hidden, and a false pseudonym with non-existent credentials (“Arthur Rhoades MBA”). The latter is arguably material because it implies that someone is staking his reputation on his statement of opinion.
Predictably, if these practices continue to prove profitable, more and more effort will be expended trying to learn the quantitative triggers that high frequency traders use. In some cases, this could involve fact patterns that might be actionable under traditional misappropriation theory. For example, if trader X induces a high frequency firm to disclose the quantitative tests they employ to run their algorithms, such conduct might resemble the kind of “feigning fidelity” to the source of the information that gave rise to liability in United States v. O’Hagan.8 But, the remaining question here might be whether there was any fiduciary or similar relationship of trust and confidence, which is usually a necessary precondition to liability under misappropriation theory.
Another plausible tactic would be for the would-be manipulator to tip off friends and allies about its campaign for Company XYZ, telling them that it thinks its efforts will give the stock price a distinct upward bump. Such a “gift” of information could violate the Supreme Court’s recent decision in Salman v. U.S.,9 but the plaintiffs would need to identify a fiduciary (or similar) duty that was breached.10
A last possibility might involve hacking into the high frequency trader’s emails or system to learn its quantitative criteria. This can be analogized by the conduct in SEC v. Dorozhko,11 where the defendant hacked into a public relation firm’s website to obtain the still unreleased financial results of one of its clients. The defendant in Dorozhko raised the defense that he was not a fiduciary to anyone (and so in his view could not be held liable for insider trading), but the Second Circuit found that his behavior could be considered “deceptive” within the meaning of §10(b).12
Another strategy for the would-be manipulator would be to seek to penetrate the firms that collect “Big Data.” If such a person could gain access to the data being compiled by a firm such as Social Media Analytics, and learned what they were about to tell their clients, it might not need to know the clients’ algorithms. It could assume that the data was valuable, and if it showed a major change, someone would trade in reliance on it. Again, there is no fiduciary relationship here, but there could be a misappropriation (on some facts) or a “deceptive” hacking (on other facts).
What steps should sensible professionals take? First, websites and blogs need to insist on representations that persons posting with it have not been compensated. Possibly, they should insist on a further representation that any claimed credentials (such as an advanced degree or a professorship) are valid. Names of authors could be checked on Google (or elsewhere) to see if real persons are associated with them. The SEC is currently approaching this problem on a case-by-case basis. But at some point in the not-distant future, it should issue a release announcing the factors that it will consider in determining whether concentrated trading after multiple publications amounts to a manipulation. Finally, the U.S. Attorney in a strong case should prosecute criminally. That should generate deterrence.
‘Fake News’ and ‘Fake Views’
Skeptics may respond that there is no need to protect the high frequency trader. If such traders use a “sitting duck” algorithm, this argument runs, they have no one to blame but themselves. Yet, even if we have no tears to shed for these traders, two points must be remembered: First, there are always innocent bystanders (such as retail investors) who are also injured. Second, “fake news” (and its younger sibling, “fake views”) distort and corrupt the market. Transparency is the true goal and holy grail of the federal securities laws, and it must be protected against all who would cloud the market.
1. This estimate comes from the Tabb Group.
2. Case No. 17-2540 (S.D.N.Y. April 10, 2017).
3. See Renae Merle, “Allegations of ‘Fake News’ Stretch Beyond Political World.” Washington Post, July 5, 2017, at p. A10. See note 4 infra.
4. Id. For more about the “Dream Team,” Lidingo, and Galena Biopharma, see In re Galena Biopharma Sec. Litig., 117 F. Supp. 3d 1145 (D. Ore. 2015). Eventually, the plaintiff’s attorneys received a fee award in this case of $9 million plus costs for their work in what the court described as a “‘pump and dump’ insider trading scheme.” See In re Galena Biopharma Sec. Litig., 2016 U.S. Dist. LEXIS 82693 (D. Ore. April 24, 2016). This column will not cover the broader issues involved in an issuer’s retention of “social media” firms, and violations of both Rule 10b-5 and Regulation FD are easy to imagine. See Maggie D. Vito, “Regulation Fair Disclosure and Social Media: Why Companies Must Be Careful Not to Tweet Their Way to SEC Investigations,” 18 Duq. Bus. L. J. 161 (2016).
5. See U.S. v. Coscia, 177 F. Supp 3d 1087 (N.D. Ill. 2016) (upholding conviction of “spoofing” commodities trader on six counts of spoofing and six counts of commodities fraud). See also “What Coscia Conviction Means for High Frequency Traders,” Law360, Jan. 7, 2016.
6. See Merle, supra note 3 (quoting Joe Gits, the CEO of this firm).
7. 15 U.S.C. §77q(b).
8. 521 U.S. 642 (1997).
9. 137 S. Ct. 420 (2016).
10. For example, if a social media firm was hired by the issuer, its tipping of its “friends” in the industry might breach its fiduciary duty (as a constructive insider) to the company. Even though the tippee paid nothing in return to the tipper for this information (and therefore did not breach the “personal benefit” test normally required), it might be a close enough ally to be a “close friend” under Salman. Much will turn on the facts here.
11. 574 F. 3d 42 (3d Cir. 2009). The public relations firm was Thomson Financial, which handles the release of financial results for many clients.
12. The Second Circuit drew a difficult distinction between deceptive misbehavior and “mere theft.” 574 F. 3d at 51. It ruled that “depending on how the hacker gained access, it seems to us entirely possible that computer hacking could be, by definition, a ‘deceptive device or contrivance’ that is prohibited by Section 10(b) and Rule 10b-5.” Id.