On November 3rd, high-frequency trader Michael Coscia was found guilty in Chicago in one of the most-watched financial trials in recent years. His conviction under Dodd-Frank’s new anti-spoofing provision is important on a number of levels: what it means for a number of other market manipulation investigations, its deterrence value going forward, and the failure of the void for vagueness challenge mounted by the defense. Coscia gives the imprimatur of the Federal courts to Dodd-Frank Section 747 as a tool to police the markets in the new financial world where “trades are not the basic unit of market information—the underlying orders are.” Perhaps more abstractly, Coscia is also interesting because it deals with the problem of intent concerning actions carried out by algorithms. It therefore has wider significance for the developing body of law assigning responsibility for actions carried out by computer algorithms. Coscia is clearly an important case of first impression.
First, what did Michael Coscia do? The indictment alleges that he engaged in “spoofing,” or more precisely “layering,” in six separate incidents involving various commodities markets. These layering transactions involved placing a series of large “quote” orders on one side of the book, while simultaneously placing a small “trade” order on the opposite side. As the quote orders began to move the market, the small trade order would be filled. Then the quote orders would be cancelled before they could be filled. The entire process would then be repeated on the other side of the market. Count One of the indictment gives an example of how this worked: Coscia began by placing a buy order in the Euro FX market for 14 contracts at 14288. 11 milliseconds later he placed three large sell orders: 91 contracts at 14291, 99 at 14290, and 61 contracts at 14289. These quotes caused market prices to fall, and seven milliseconds later the buy order was filled. Six milliseconds later, he cancelled his sell orders. Then he reversed this process, placing a sell order for 14 contracts at 14289, higher than any other bid. He then entered four buy orders: 88 contracts at 14284, 88 contracts at 14286, 88 contracts at 14287, and 61 contracts at 14288. Nine milliseconds later, his sell order was filled, and five milliseconds after that he canceled his buy orders. “By entering large orders that he intended to cancel at the time he placed them, and caused to be canceled before other traders could fill them, Coscia made a profit by buying 14 contracts at 14288 and selling them at 14289 less than one second later.”
This conduct is a textbook example of manipulative layering: posting a small quote on one side of the market, then large quotes on the other side to get the price to move towards the small quote, then cancelling the large quotes before they can be filled. Michael Coscia would buy at an artificially induced low price, then turn around and artificially induce the price back up, and then finally sell at that level. While the amounts of money made in each episode were small, totaling only $1,070 for the six spoofing counts he was charged with, the indictment states that he made approximately $1.5 million using this strategy from August through October 2011.
While he had already settled civil charges with both the U.S. Commodity Futures Trading Commission (the “CFTC”) and the U.K.’s Financial Conduct Authority, he was charged with six criminal violations of the new anti-spoofing provision in the Commodity Exchange Act (the “CEA”) as well as six counts of commodities fraud. CEA Section 4c(a)(5)(C) prohibits “any trading, practice, or conduct [that] . . . is of the character of, or is commonly known to the trade as ‘spoofing’ (bidding or offering with the intent to cancel the bid or offer before execution).” While this very broad language left the provision open to a constitutional due process challenge, the indictment describes quotes that were left open for as little as 13 milliseconds, and were cancelled in as little as five milliseconds after the order on the other side of the market filled. While there is no minimum time a quote must be left open, the fact that these quotes were open for such a short period of time, combined with their position in an apparently coordinated sequence consisting of large orders on one side of the market and a smaller order on the other side, pointed to the conclusion that the large “quote” orders were never intended to be filled. Evidence at trial supported this, including language in programmer Jeremiah Park’s handwritten notes that the quote orders would be “used to pump market” [sic].
Despite—or perhaps because of—these disadvantageous facts, Coscia’s defense team mounted a wholesale attack on the statute, claiming that it was “void for vagueness” and thus constitutionally impermissible. This strategy had worked before in federal court in Houston in United States v. Radley. There, four defendants working for British Petroleum were charged with various criminal violations of the CEA including price manipulation and conspiracy to corner the propane market. Judge Gray Miller dismissed the case, ruling that “manipulation” in this context was impermissibly vague, and that orders including stacked bids and “show” offers designed to increase prices while BP held long positions did not create an “artificial” price. Speculative market activity for the sake of making a profit is a legitimate business activity, and since the offers themselves were not fraudulent, the government’s indictment failed.
Judge Harry Leinenweber of the Northern District of Illinois rejected the defense’s contention that the new anti-spoofing provision is likewise vague. Unlike other cases where the CEA was found to be impermissibly vague, the parenthetical in CEA Section 4c(a)(5)(C) provides a definition of spoofing: “(bidding or offering with the intent to cancel the bid or offer before execution).” While there may be real uncertainty as to what exactly are the outer limits of spoofing in a world where more than 95% of quotes are cancelled before they are filled, if anything is obviously spoofing it would seem to be what Coscia’s algorithms did: layer a series of quotes on one side of the market resulting in a fill on the other side, and then immediately cancel the outstanding quotes before they can be hit. His lawyers referred to numerous expressions of concern about the definition of spoofing voiced at a 2010 CFTC Roundtable on Disruptive Trading Practices, but outside of the First Amendment context, a party claiming unconstitutional vagueness must demonstrate that the statute is vague as applied to the facts of her case. While there may be questions as to whether “kill or fill” orders, partial-fill orders, or other types of conduct would fall within the boundaries of spoofing, Judge Leinenweber ruled that given the facts alleged, the statute was not void and the case could proceed.
What are Coscia’s implications? For current defendants in spoofing actions it may offer, paradoxically, a strong incentive not to settle a civil action. Coscia himself faces a potentially lengthy prison sentence and fines, on top of the already substantial fines he has paid in the earlier civil actions. There are legitimate questions as to whether his punishment fits the crime, but in an environment of widespread public questioning as to why so few financial executives have been prosecuted for misdeeds associated with the financial crisis, regulators may have felt a significant temptation to make an example of him. Assuming it is not overturned upon appeal, this conviction should have a significant deterrent effect in the years to come.
Finally, it is interesting to reflect on the implications of Coscia in the world of the “Second Machine Age,” where more and more activities formerly done directly by humans are executed by computer algorithms. Just as the anti-spoofing statute allows for the extension of anti-manipulation doctrine to orders that are not themselves fraudulent—unlike Judge Miller’s construction of the CEA in Radley—Coscia takes the natural step of inferring a second-order intent in the programming of algorithms to accomplish certain tasks. Such an inference may become more and more common as the law grapples with legal responsibility in realms as diverse as self-driving cars, drone warfare, and computer-assisted medical practice. Coscia then may be an early example of attribution of responsibility to those in charge of programming the computers in the world of finance.
 See Bradley Hope, Commodities Trader Found Guilty in ‘Spoofing’ Case, Wall St. J., Nov. 3, 2015.
 Maureen O’Hara, High Frequency Market Microstructure, 116 J. Fin. Econ. 257, 263 (2015).
 Indictment, United States v. Coscia, No. 14-CR-00551 (N.D. Ill. Oct. 1, 2014), available at http://www.justice.gov/sites/default/files/usao-ndil/legacy/2015/06/11/pr1002_01a.pdf.
 Id. at 6-7.
 The EUR/USD foreign exchange contract is priced to four decimal places, so 0.0001 is the “percentage in point” (or “PIP”), the smallest unit a contract can fluctuate.
 Id. at 7.
 See Andrei A. Kirilenko and Andrew W. Lo, Moore’s Law versus Murphy’s Law: Algorithmic Trading and Its Discontents, 27 J. Econ. Persps. 51, 66-67 (2013).
 Indictment at 4, United States v. Coscia, No. 14-CR-00551 (N.D. Ill. Oct. 1, 2014).
 In the Matter of Panther Energy Trading LLC and Michael Coscia, CFTC Docket No. 13-26 (July 22, 2013); Final Notice: Michael Coscia, FCA (July 22, 2013).
 Entitled “Disruptive Practices,” new § 4c(a)(5) of the Commodities Exchange Act was enacted as § 747 of the Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010, Pub. L. No. 111-203, 124 Stat. 1376, 1739. The anti-spoofing provision is codified as amended at 7 U.S.C. § 6c(a)(5)(C). 7 U.S.C. § 13(a)(2) makes spoofing a felony punishable by a fine of not more than $1 million or punishment not more than 10 years, or both.
 18 U.S.C. § 1348.
 7 U.S.C. § 6c(a)(5)(C).
 Indictment at 6-7, United States v. Coscia, No. 14-CR-00551 (N.D. Ill. Oct. 1, 2014).
 See Brian Lewis, Annie Massa & Janan Hanna, From Pits to Algos, an Old-School Trader Makes Leap to Spoofing, BloombergBusiness, Nov. 12, 2015.
 Defendant’s Motion to Dismiss 15-26, United States v. Coscia, No. 14-CR-00551 (N.D. Ill. Dec. 15, 2014)
 United States v. Radley, 659 F. Supp. 2d 803 (S.D. Tex. 2009), aff’d on other grounds, 632 F.3d 177 (5th Cir. 2011).
 Id., 659 F. Supp. 2d, at 815.
 Memorandum Opinion and Order, United States v. Coscia, No. 14-CR-00551 (N.D. Ill. Apr. 16, 2015).
 7 U.S.C. § 6c(a)(5)(C).
 CFTC Staff Roundtable on Disruptive Trading Practices (Dec. 2, 2010), available at http://www.cftc.gov/idc/groups/public/@swaps/documents/dfsubmission/dfsubmission24_120210-transcri.pdf.
 See, e.g., Jed S. Rakoff, The Financial Crisis: Why Have No High-Level Executives Been Prosecuted?, New York Review of Books (Jan. 9, 2014).
 See Erik Brynjolfsson & Andrew McAfee, The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies (W.W. Norton 2014).
 See Ryan Calo, Robotics and the Lessons of Cyberlaw, 103 Calif. L. Rev. 513 (2015).
The preceding post comes to us from Steven McNamara, Assistant Professor of Business Law at the Olayan School of Business, The American University of Beirut.