United States v. Michael Coscia

4 F.4th 454

• Court: Court of Appeals for the Seventh Circuit
• Decided: July 12, 2021
• Docket: 19-2010
• Status: Published

Opinion

In the

United States Court of Appeals For the Seventh Circuit ____________________ No. 19-2010 UNITED STATES OF AMERICA, Plaintiff-Appellee, v.

MICHAEL COSCIA, Defendant-Appellant. ____________________

No. 20-1032 MICHAEL COSCIA, Petitioner-Appellant,

v.

UNITED STATES OF AMERICA, Respondent-Appellee. ____________________

Appeals from the United States District Court for the Northern District of Illinois, Eastern Division. Nos. 14-cr-00551-1 & 19-cv-05003 — Harry D. Leinenweber, Judge. ____________________

ARGUED DECEMBER 2, 2020 — DECIDED JULY 12, 2021 ____________________ 2 Nos. 19-2010 & 20-1032

Before EASTERBROOK, RIPPLE, and ROVNER, Circuit Judges. RIPPLE, Circuit Judge. A jury convicted Michael Coscia of six counts of commodities fraud, in violation of 18 U.S.C. 1 § 1348, and six counts of spoofing, in violation of 7 U.S.C. §§ 6c(a)(5)(C) and 13(a)(2). On direct appeal, we affirmed his 2 conviction. We now have before us the appeals of two pro- ceedings that Mr. Coscia initiated after we resolved his di- rect appeal. The first is a motion for a new trial on the basis of new evidence in which he alleges (1) that data discovered after trial establishes that there were errors in the data pre- sented to the jury and (2) that subsequent indictments against other traders for similar spoofing activities undercut the Government’s characterization of Mr. Coscia as “unique” or a trading “outlier.” The second proceeding is a motion to vacate his conviction pursuant to 28 U.S.C. § 2255, in which Mr. Coscia claims that his trial counsel, Sullivan & Cromwell LLP, provided ineffective assistance of counsel. Specifically, he alleges that Sullivan & Cromwell had an undisclosed con- flict of interest with several of the Government’s witnesses and that this conflict adversely affected counsel’s perfor- mance. He also alleges that, even if there was no conflict of interest, his trial counsel nevertheless provided constitution- ally deficient representation. The district court denied both motions, and Mr. Coscia

1 Spoofing is a disruptive trading practice in which a person submits bids or offers with the intent to cancel the bid or offer before it is execut- ed. See 7 U.S.C. § 6c(a)(5). 2 United States v. Coscia, 866 F.3d 782 (7th Cir. 2017). Nos. 19-2010 & 20-1032 3

3 now appeals. He submits that the district court abused its discretion when it denied his new trial motion. In his view, the newly discovered evidence demonstrated that key evi- dence relied on by the Government to establish his intent to spoof was false and inaccurate. As for his habeas motion, he contends that the district court correctly found that counsel had a conflict of interest, but incorrectly concluded that there was no adverse effect on counsel’s performance. He further submits that the district court erred in rejecting his argument that, even in the absence of a conflict of interest, his defense counsel’s performance was constitutionally deficient. In the alternative, Mr. Coscia requests further discovery and an ev- identiary hearing on his ineffective assistance of counsel claims. We now affirm the district court’s judgments on both the new trial and § 2255 motions. We conclude that the district court did not abuse its discretion in denying Mr. Coscia’s motion for a new trial on newly discovered evidence grounds. We further conclude that the district court correctly determined that Mr. Coscia failed to demonstrate an adverse effect or prejudice in either of his ineffective assistance of counsel claims.

3 Mr. Coscia filed two separate notices of appeal for each of the two mo- tions denied by the district court. For the sake of judicial economy, we consolidated his appeals. We employ the standard “R.” and “Appellant’s Br.” when referring to Mr. Coscia’s appeal of his new trial motion, and “2255 R.” and “Appellant’s 2255 Br.” when referring to Mr. Coscia’s ap- peal of his § 2255 motion. 4 Nos. 19-2010 & 20-1032

I BACKGROUND A. Mr. Coscia’s Trading Activity Michael Coscia was the principal of a futures trading firm, Panther Trading LLC. He traded commodity futures contracts on electronic exchanges operated by CME Group, Inc. (“CME”) and the Intercontinental Exchange, Inc. (“ICE”). Trading firms such as Mr. Coscia’s use computer programs to execute trades that are carried out in fractions of a second. In our opinion affirming Mr. Coscia’s convic- tion, we described the basic process of high-frequency trad- ing: The simplest approaches take advantage of the minor discrepancies in the price of a security or commodity that often emerge across national exchanges. These price discrepancies allow traders to arbitrage between exchanges by buy- ing low on one and selling high on another. Because any such price fluctuations are often very small, significant profit can be made only on a high volume of transactions. Moreover, the discrepancies often last a very short period of time (i.e., fractions of a second); speed in ex- ecution is therefore an essential attribute for firms engaged in this business. United States v. Coscia, 866 F.3d 782, 786 (7th Cir. 2017). High-frequency trading can also be used “to artificially move the market price of a stock or commodity up and Nos. 19-2010 & 20-1032 5

down, instead of taking advantage of natural market events.” Id. at 787. This artificial movement can be accom- plished “by placing large and small orders on opposite sides of the market.” Id. For example, if an unscrupulous trader wanted to buy, he would place a small order below the cur- rent market price. He would simultaneously place large or- ders to sell on the opposite side of the market. He would place these large sell orders at progressively lower prices un- til the purchase price matched the price at which the small buy order had been placed. The small order then would be executed, and the large orders would be cancelled. “Im- portantly, the large, market-shifting orders that he places to create this illusion are ones that he never intends to execute; if they were executed, our unscrupulous trader would risk extremely large amounts of money by selling at suboptimal prices.” Id. Congress criminalized this practice, called “spoofing,” in 2010 as part of the Dodd-Frank Wall Street Reform and Con- sumer Protection Act, Pub. L. No. 111-203, 124 Stat. 1376 (2010). It became unlawful “to engage in any trading, prac- tice, or conduct on or subject to the rules of a registered enti- ty that … is, is of the character of, or is commonly known to the trade as, ‘spoofing’ (bidding or offering with the intent to 4 cancel the bid or offer before execution).” 7 U.S.C. § 6c(a)(5).

4 “[A] bid is an order to buy and an offer is an order to sell.” Coscia, 866 F.3d at 787. 6 Nos. 19-2010 & 20-1032

B. Mr. Coscia’s Trial In August 2011, Mr. Coscia implemented two high-frequency trading programs that followed a specific pattern: When he wanted to purchase, Mr. Coscia would begin by placing a small order request- ing to trade at a price below the current market price. He then would place large-volume or- ders, known as “quote orders,” on the other side of the market. A small order could be as small as five futures contracts, whereas a large order would represent as many as fifty or more futures contracts. At times, his large orders risked up to $50 million. The large orders were generally placed in increments that quickly approached the price of the small orders. Coscia, 866 F.3d at 788 (footnotes omitted). A grand jury indicted Mr. Coscia for six counts of spoof- ing and six counts of commodities fraud based on his 2011 trading activity. Trial began on October 26, 2015. The Gov- ernment set forth Mr. Coscia’s pattern of trading: placing small orders and large orders on opposite sides of the mar- ket, with the small orders filling once the desired price was met and the large orders immediately cancelled. The same pattern would then repeat in the opposite direction. Each of these patterns took place within one second or less. To estab- lish Mr. Coscia’s fraudulent intent, the Government present- ed (1) the testimony of Jeremiah Park, Mr. Coscia’s computer programmer; (2) testimony of representatives of ICE and Nos. 19-2010 & 20-1032 7

CME, who described Mr. Coscia’s trading activities and pre- sented charts summarizing relevant trading data; (3) testi- mony of other traders on the effect of Mr. Coscia’s trading on their businesses; (4) Mr. Coscia’s deposition testimony taken by the Commodity Futures Trading Commission; and (5) the rebuttal testimony of financial markets expert Hank Bessembinder. Park testified that he created, at Mr. Coscia’s direction, two programs: Flash Trader and Quote Trader. He con- firmed that Mr. Coscia had specified that the programs were 5 to act “[l]ike a decoy” to “pump [the] market.” Specifically, the large-volume orders were designed to avoid being filled and would be cancelled based on (1) the passage of time, (2) the partial filling of large orders, or (3) the complete filling of a small order. These cancellation settings were intended to reduce the risk that the large orders would be filled. After the large orders were cancelled, the program would reenact the trades in reverse. The Government also presented representatives of ICE and CME who testified about trading data summarized in data charts. The court admitted, without objection, six ICE summary charts and six CME summary charts. John Red- man, the director of compliance for ICE, testified about the ICE data and summary charts. Ryan Cobb, a data scientist for CME, testified about the CME data and summary charts. Both Redman’s and Cobb’s testimony supported the Gov- ernment’s view that Mr. Coscia had engaged in a specific trading pattern that was outside trading norms. We briefly

5 R.86 at 231, 235 (Tr. 498, 502). 8 Nos. 19-2010 & 20-1032

review the charts relevant to this appeal. ICE Summary Chart 2 compared the rate at which Mr. Coscia filled his large orders to the rate at which he filled his small orders on the ICE market. Redman testified that between August and October 2011, Mr. Coscia had placed 24,814 large orders and had traded on 0.5% of those orders. In contrast, Mr. Coscia had placed 6,782 small orders 6 and had traded on approximately 52% of those orders. ICE Summary Chart 3 displayed Mr. Coscia’s “or- der-to-trade ratio,” or “the average size of the order he showed to the market divided by the average size of the or- 7 ders filled.” This chart compared the activity of Mr. Coscia’s firm on the ICE market to the activity of others trading at approximately the same volume. ICE Summary Chart 3 showed that Mr. Coscia’s average order size was 39.8 lots, but his average trade size was 2.5 lots. Thus, Mr. Coscia’s order-to-trade ratio was 1,592%. According to the chart, oth- er trading entities had ratios between 91% to 264%. ICE Summary Chart 6 displayed Mr. Coscia’s share of cancellations of large orders that followed the trade of a small order in the opposite direction. This chart showed that, between September and October 2011, Mr. Coscia cancelled large orders 14,563 times following the execution of small-order trades; other market participants followed this pattern only 671 times combined. Thus, according to the chart, Mr. Coscia accounted for 96% of all of the cancelled

6 See R.177-31.

7 Coscia, 866 F.3d at 789; see also R.177-32. Nos. 19-2010 & 20-1032 9

large orders that followed the execution of a small order in the opposite direction on the ICE exchange. CME Summary Charts 2 and 3 compared the rates at which Mr. Coscia filled large and small orders, referred to as a “fill-rate differential.” These charts showed that, on the CME markets, Mr. Coscia filled 35.61% of his small orders, but only 0.08% of his large orders, resulting in a 35.53% fill-rate differential. CME Summary Chart 5 showed how Mr. Coscia ranked among all trading entities in the same markets in terms of large orders placed (“large order entry rank”) and large or- 8 ders actually traded (“volume rank”). Mr. Coscia ranked first, entering the most large orders in eleven of the seven- teen CME commodities. Mr. Coscia’s volume rank for large orders actually traded, however, was significantly lower 9 across all commodities. Finally, the Government introduced testimony of other traders, some of whom lost hundreds of thousands of dollars as a result of Mr. Coscia’s trading activity. Anand Twells, a trading supervisor at Citadel, LLC, testified that in a transac- tion with Panther Trading, it lost “about $480” in “roughly 10 400 milliseconds.” Hovannes Dermenchyan, the global

8 See R.177-5.

9 To illustrate, Ryan Cobb testified that in the Australian dollar market, Mr. Coscia ranked first, entering more large orders than any other partic- ipant, but was only the thirty-third highest participant “in terms of actu- al trade volume.” R.86 at 135 (Tr. 402). 10 R.88 at 30 (Tr. 635). 10 Nos. 19-2010 & 20-1032

head of trading and markets at Teza Technologies, testified that his firm “lost $10,000 over the course of an hour” be- cause its programs were “induced into trading” by this sin- 11 gle participant’s behavior. “[E]very time that participant placed a very large order,” he went on, “it would induce this 12 specific strategy to trade on the opposite side.” Alexander Gerko, who was previously a portfolio manager at GSA Capital, testified that he “noticed a pattern of activity” where “very, very large orders appear[ed] on the market …, and then these orders would disappear from the bid and ap- 13 pear on the offer.” Gerko stated that his firm noticed this activity “because [they] started to lose a substantial amount 14 of money” from “trading with the small trade.” Finally, Jonathan Eddy, senior vice president at D.E. Shaw & Com- pany, testified that his firm’s computer program considered “order imbalance” in the market as a factor in whether it 15 trades. Eddy explained that its program was “more likely to want to sell” when there were more orders to sell in the market; after multiple large orders to sell were placed at de- 16 creasing prices, its program also placed an order to sell.

11 Id. at 51 (Tr. 656).

12 Id.

13 Id. at 90 (Tr. 695).

14 Id. at 91, 105 (Tr. 696, 710).

15 R.89 at 5 (Tr. 762).

16 Id. Nos. 19-2010 & 20-1032 11

In his defense, Mr. Coscia maintained that his trading was legitimate because each order placed was an order ca- pable of being filled prior to its cancellation. He testified that the strategy of his programs was to “make a lopsided market 17 and hope to get traded on the better of the offer.” If the small order was executed, the large order would be can- celled; if the large order was executed, the small order would be cancelled. Mr. Coscia also testified that he had no preference, and sometimes did not even know, whether his small or large orders were filled. In rebuttal, the Government presented testimony from financial markets expert Hank Bessembinder. He testified that Mr. Coscia’s trading was materially different from that of other high-frequency traders. In contrast to Mr. Coscia’s claims of indifference as to which of his orders were filled, Bessembinder testified that “[t]he outcomes don’t seem to reflect that same sort of even balance” of an indifferent trad- 18 er. Rather, “the outcomes are far, far from being 50/50 or 19 equal outcomes on both sides of the market.” Bessembin- der added that “[t]here were more than 10 times as many contracts traded on the small orders as compared to the 20 large orders.” Mr. Coscia “was entering over 60 percent of his orders as large orders, whereas, the other high-frequency

17 Id. at 119–20 (Tr. 876–77).

18 R.91 at 117 (Tr. 1363).

19 Id.

20 Id. at 117–18 (Tr. 1363–64). 12 Nos. 19-2010 & 20-1032

traders were entering only about a quarter of one percent of 21 their orders as large orders.” But Mr. Coscia cancelled “[a] little over 97 percent” of his large orders within one second, while other high-frequency trading firms canceled their large orders within one second “[j]ust under 35 percent” of 22 the time. Further, Bessembinder testified that Mr. Coscia’s large-order fill rate did not account for his successive at- tempts to cancel orders that failed because the order had al- 23 ready fully executed milliseconds earlier. The jury convicted Mr. Coscia on all twelve counts. The court sentenced Mr. Coscia to a term of thirty-six months’ imprisonment, followed by two years’ supervised release. Mr. Coscia filed a motion for judgment of acquittal and for a 24 new trial, both of which the district court denied.

21 Id. at 123 (Tr. 1369).

22 Id. at 125 (Tr. 1371).

23 Id. at 106–09 (Tr. 1352–55) (describing an exhibit that showed an entry that an order had filled, followed by two cancellation entries that ap- peared milliseconds after that generated “order not found error code[s]”). 24 Mr. Coscia’s first new trial motion, which is not before us, was made on the basis that Mr. Coscia’s convictions were “against the weight of the evidence, the jury was not properly instructed, and the Government in- troduced inadmissible, false, and prejudicial testimony.” R.96 at 1. The motion before us is Mr. Coscia’s second new trial motion on the basis of newly discovered evidence. See R.219; R.220. Nos. 19-2010 & 20-1032 13

C. Direct Appeal Mr. Coscia appealed his conviction and sentence, and we affirmed. Coscia, 866 F.3d at 782. In his appeal, Mr. Coscia challenged the anti-spoofing statute as unconstitutionally vague. He also contended that the Government produced insufficient evidence to support his spoofing conviction. We first held that the anti-spoofing statute provided suf- ficient notice and that “Mr. Coscia’s actions [fell] well within the core of the anti-spoofing provision’s prohibited conduct, precluding any claim that he was subject to arbitrary en- forcement.” Id. at 795. With respect to the sufficiency of the evidence, we observed: As we have noted earlier, a conviction for spoofing requires that the prosecution prove beyond a reasonable doubt that Mr. Coscia knowingly entered bids or offers with the pre- sent intent to cancel the bid or offer prior to ex- ecution. Mr. Coscia’s trading history clearly indicates that he cancelled the vast majority of his large orders. Accordingly, the only issue is whether a rational trier of fact could have found that Mr. Coscia possessed an intent to cancel the large orders at the time he placed them. A review of the trial evidence reveals the fol- lowing. First, Mr. Coscia’s cancellations repre- sented 96% of all Brent futures cancellations on the Intercontinental Exchange during the two-month period in which he employed his 14 Nos. 19-2010 & 20-1032

software. Second, on the Chicago Mercantile Exchange, 35.61% of his small orders were filled, whereas only 0.08% of his large orders were filled. Similarly, only 0.5% of his large orders were filled on the Intercontinental Ex- change. Third, the designer of the programs, Jeremiah Park, testified that the programs were designed to avoid large orders being filled. Fourth, Park further testified that the “quote orders” were “[u]sed to pump [the] market,” suggesting that they were designed to inflate prices through illusory orders. Fifth, according to one study, only 0.57% of Coscia’s large or- ders were on the market for more than one second, whereas 65% of large orders entered by other high-frequency traders were open for more than a second. Finally, Mathew Evans, the senior vice president of NERA Economic Consulting, testified that Coscia’s or- der-to-trade ratio was 1,592%, whereas the or- der-to-trade ratio for other market participants ranged from 91% to 264%. Id. at 795–96 (alterations in original) (footnotes omitted). We therefore concluded that, “when evaluated in its totality, the cumulative evidence certainly allowed a rational trier of fact to determine that Mr. Coscia entered his orders with the in- tent to cancel them before their execution.” Id. at 796. We decided Mr. Coscia’s direct appeal on August 7, 2017. On January 10, 2019, Mr. Coscia filed in the district court a second motion for a new trial. The district court denied this motion on May 15, 2019. Two months later, Mr. Coscia filed Nos. 19-2010 & 20-1032 15

a motion to vacate his conviction under 28 U.S.C. § 2255. The district court denied this motion on December 12, 2019. The district court’s decisions on these two motions are before us today. For ease of reading, we discuss each separately in this opinion. We first will address the district court’s denial of the second motion for a new trial; we then turn to the motion under § 2255. In each discussion, we will set forth additional particular facts pertinent to our analysis. II The Motion for A New Trial A. Governing Standards We review a district court’s denial of a motion for new trial based on newly discovered evidence for an abuse of discretion. United States v. Reyes, 542 F.3d 588, 595 (7th Cir. 2008). District courts may “grant a new trial if the interest of justice so requires.” Fed. R. Crim. P. 33(a). Granting a new trial in the “interest of justice” is “‘reserved for only the most extreme cases,’ and we ‘approach such motions with great caution and are wary of second-guessing the determinations of both judge and jury.’” United States v. Hagler, 700 F.3d 1091, 1101 (7th Cir. 2012) (first quoting United States v. Lin- wood, 142 F.3d 418, 422 (7th Cir. 1998); and then quoting 25 United States v. McGee, 408 F.3d 966, 979 (7th Cir. 2005)).

25 See also United States v. Kamel, 965 F.2d 484, 490 (7th Cir. 1992) (“Be- cause of the importance accorded to considerations of repose, regularity of decision-making and conservation of scarce judicial resources, courts exercise ‘great caution’ in setting aside a verdict reached after ful- (continued … ) 16 Nos. 19-2010 & 20-1032

In seeking a new trial, based on newly discovered evi- dence, the defendant must demonstrate that the new evi- dence “(1) was discovered after trial, (2) could not have been discovered sooner through the exercise of due diligence, (3) is material and not merely impeaching or cumulative, and (4) probably would have led to acquittal.” United States v. O’Malley, 833 F.3d 810, 813 (7th Cir. 2016). In an effort to meet these criteria, Mr. Coscia presents two categories of “newly discovered” evidence: (1) ICE and CME data dis- closed post-trial, and (2) subsequent indictments against other traders for similar activities. We address each of these categories in turn. B. Newly Discovered Data The first category of newly discovered evidence prof- fered by Mr. Coscia is data disclosed post-trial by ICE and CME. In Mr. Coscia’s view, this newly discovered infor- mation raises a significant question regarding the accuracy of the charts that the Government employed at trial to estab- lish that he intentionally had engaged in spoofing. 1. Background During pretrial discovery, ICE and CME produced cer- tain data that the Government intended to use at trial through the summary charts that we described earlier. One month before trial, Mr. Coscia’s counsel requested produc-

( … continued) ly-conducted proceedings; this is particularly appropriate when, as here, the action has been tried before a jury.” (footnotes omitted)). Nos. 19-2010 & 20-1032 17

tion of these charts. The Government responded that it had not yet prepared the summary charts but that “the infor- mation that they summarize ha[d] already been produced” 26 to the defense. Mr. Coscia’s counsel independently issued four subpoe- nas: one subpoena to ICE for the audit trail of ten market participants on the ICE Brent Crude Futures market over a 27 two-month period; and three subpoenas to CME for, among other things, the full audit trail for Mr. Coscia’s or- 28 ders, audit trail information from August to October 2011 29 covering eight contracts, and more specific information about the activity of eight high-frequency trading compa- nies. Following trial but before sentencing, CME produced a complete record of all market participants’ order and trading histories covering approximately ten weeks in all seventeen 30 markets in which Mr. Coscia had traded. Through Sullivan

26 R.227-8 at 2.

27 See R.227-4.

28 See R.227-2.

29 See R.227-3.

30 It is not clear from the record how Mr. Coscia obtained the CME data post-trial. The Government notes that Mr. Coscia received data from CME that neither he nor the Government possessed at the time of trial, and that “[d]efense counsel refused to explain how defendant obtained this new CME data.” Appellee’s Br. 24 n.7; see also R.224 at 2 n.2 (“On January 18, 2019, the [Government] requested that defense counsel dis- close when and how defendant sought this data from CME. Defense (continued … ) 18 Nos. 19-2010 & 20-1032

& Cromwell, and later Kobre & Kim LLP, Mr. Coscia at- tempted to obtain the full audit trail data of all market par- ticipants on the Brent Crude Futures market for the period between September 6, 2011 and October 18, 2011, as well as personal and confidential information relating to govern- ment witness Hovannes Dermenchyan. During Mr. Coscia’s attempt to obtain additional data from ICE, that exchange disclosed that it had inadvertently failed to produce all of the data underlying ICE Summary Chart 6 to either the Gov- ernment or the defense before trial. The court denied Mr. Coscia’s request to subpoena ICE for its full audit trail data but ordered ICE to produce the information that had been used to create ICE Summary Chart 6. ICE complied with the court’s order to produce the data used to create ICE Summary Chart 6. In its letter producing the data, ICE explained that the order cancellations dis- played in ICE Summary Chart 6 are “based on an ICE sys- tem tool that, for regulatory purposes, generates an alert 31 when the tool detects suspicious trading activity.” ICE also disclosed that, “due to an inadvertent error in calculating the total alerts from the Backup Data to create Summary Chart 6, the totals for the alerts for Mr. Coscia and ‘All other partici-

( … continued) counsel responded, on January 22, that it would be a ‘substantial burden’ to obtain this information, but confirmed that defendant did not have before trial the CME data on which his [second motion for a new trial] relies.”). 31 R.222-3 at 2. Nos. 19-2010 & 20-1032 19

32 pants’ … were misstated.” Specifically, Mr. Coscia ac- counted for 14,141 of the alerts (not 14,563), and “All other participants” accounted for 1,328 of the alerts (not 671). 2. The Present Motion On January 10, 2019, Mr. Coscia filed a second motion for a new trial based on newly discovered evidence, the motion now before us. In this motion, Mr. Coscia first submitted that the newly disclosed ICE and CME data established that there were errors in the data presented to the jury. Mr. Coscia also submitted that subsequent indictments against other traders for similar spoofing activities undercut the Government’s characterization of Mr. Coscia as “unique” or that he was a trading “outlier.” In response, the Government submitted that Mr. Coscia had failed to establish that this data would have been una- vailable to him prior to trial if he had exercised due dili- gence. The Government pointed out that Mr. Coscia did not subpoena ICE for its trading data until five months after the conclusion of trial. Thus, the Government submitted that Mr. Coscia’s lack of due diligence alone was fatal to his re- quest for a new trial. Mr. Coscia replied that the data that he received after tri- al proved that there were material errors in the evidence presented to the jury. In particular, he invited the court’s at- tention to five charts: ICE Summary Charts 3 and 6, and CME Charts 2, 3, and 5. We therefore briefly review each of the alleged errors that Mr. Coscia identified and the Gov-

32 Id. at 3. 20 Nos. 19-2010 & 20-1032

ernment’s response. ICE Summary Chart 3. ICE Summary Chart 3 showed that, on the ICE New Brent crude market, Mr. Coscia’s or- der-to-trade ratio of 1,592% was significantly greater than the ratio of any other market participant. The closest firm had a ratio of 264%. Mr. Coscia maintained that the new CME data conclusively establishes that there were actually dozens and even hundreds of traders with order-to-trade ratios greater than 1,592% for each of the seventeen com- modities traded on the CME. The Government responded that Mr. Coscia “glosses over his misleading juxtaposition of data from CME markets against data from ICE markets[] without ever establishing: (1) that CME and ICE markets are comparable (i.e., ‘ap- ples-to-apples’); or (2) what defendant’s order-to-trade-size 33 ratios were for each of the seventeen CME commodities.” ICE Summary Chart 6. ICE Summary Chart 6 reflected that Mr. Coscia accounted for 96% of all large orders can- celled after a small order was filled in the opposite direction. Mr. Coscia contends, however, that the new data presents two issues. First, he stated that ICE Summary Chart 6 was presented to the jury as a summary of all order cancellations, when actually it was based on a limited set of data based on ICE’s regulatory tool that generated alerts when the tool de- tected suspicious activity. The full data, he submitted, actu- ally shows that his transactions “represented a fraction of

33 R.224 at 9. Nos. 19-2010 & 20-1032 21

34 one percent of all order cancellations.” Second, he noted that, as ICE disclosed post-trial, the numbers generated in the chart were inaccurate. Mr. Coscia represented 91%, as opposed to 96%, of the alerts. In reply, the Government first submitted that Mr. Coscia misunderstood and misrepresented the testimony at trial. Redman specifically testified that ICE Summary Chart 6 showed a specific type of cancellations: cancellations of large orders that followed the trading of a small order in the other direction. In any event, suggested the Government, demon- strating Mr. Coscia’s overall cancellation rate of less than one percent and ICE’s inadvertent, and minor, counting error only served to impeach the data presented at trial. CME Summary Charts 2 and 3. CME Summary Charts 2 and 3 demonstrated that Mr. Coscia filled 35.61% of his small orders, but only 0.08% of his large orders, resulting in a 35.53% fill-rate differential. Mr. Coscia submitted that the post-trial CME data shows that Mr. Coscia’s 35.53% fill rate was not abnormal or uncommon, as there were “dozens and even hundreds of traders” in each of the seventeen CME 35 markets who had fill-rate differentials greater than 35.53%. In his view, it was simply untrue that his trading patterns were unique. In reply, the Government submitted that Mr. Coscia’s updated data analysis misleadingly “pitt[ed] defendant’s ag- gregate differential against the commodity-specific differentials

34 R.220 at 7.

35 Id. at 19. 22 Nos. 19-2010 & 20-1032

36 of other traders.” The Government further observed that “defendant’s newly-proffered evidence does not demon- strate that any other trader across the CME markets had an 37 aggregate fill-rate differential higher than 35.53%.” In the Government’s view, then, tallying the number of individual traders with higher fill-rate differentials in a single commod- ity market—without knowing what each trader’s aggregate fill-rate differential was across all markets or if each trader traded in all seventeen markets as Mr. Coscia—was of little value when compared against Mr. Coscia’s aggregate fill-rate differential for all seventeen markets. CME Summary Chart 5. Finally, Mr. Coscia contended that the newly-produced data demonstrated that CME Summary Chart 5, which reflected that Mr. Coscia was a market leader in placing large orders but ranked lower in filling large orders, was inaccurate in three ways. First, the chart failed to include modifications of orders. Second, the data compared Mr. Coscia’s individual trading activity to trading activity of firms, which were comprised of dozens of individual traders. Third, the chart showed how Mr. Coscia ranked in large orders placed compared to both large and small orders filled. Mr. Coscia’s updated analysis included order modifications, which reduced Mr. Coscia’s overall cancellation rate. Moreover, considering only large orders filled, as opposed to large and small orders, Mr. Coscia had among the highest fill rates of anyone in the industry, and much more comparable order and fill rankings relative to

36 R.224 at 10.

37 Id. (emphasis added). Nos. 19-2010 & 20-1032 23

other traders. The Government countered that including modifications is misleading, as “defendant’s spoofing algorithm was not 38 programmed to modify—but to cancel—large orders.” And in any event, the exclusion of modifications was discussed on cross-examination. Further, the Government submitted that its rebuttal witness, Bessembinder, “testified that de- fendant’s large-order fill rate offers only a partial picture, overlooking his successive attempts to cancel orders filled 39 milliseconds before his cancellation instructions arrived.” 3. District Court’s Ruling The district court found Mr. Coscia’s presentation of this new evidence problematic because it was not relevant to the actual defense that he had presented at trial. Mr. Coscia’s defense “admitted the substance of his trading activity,” “claimed that this was a legitimate trading strategy,” and “argued that many traders pursued trading strategies simi- lar to his.” 40 Observing that “the most likely use of the so-called newly discovered evidence would be to impeach the government’s witnesses,” which could not serve as the basis for a new trial, the district court concluded that the new statistical evidence would probably not lead to an ac- quittal. 41 The district court also rejected the subsequent in-

38 Id. at 8.

39 Id. at 9.

40 R.233 at 4.

41 Id. at 5. 24 Nos. 19-2010 & 20-1032

dictments, concluding that “[a]ny such evidence would 42 hardly be relevant or material.” As the district court saw it, “[t]hat others may have employed illegal trading strategies does not constitute a defense to a criminal indictment based 43 on the employment of illegal trading strategies.” Finding that none of the new evidence satisfied the requirements for a new trial and that the jury was completely justified in con- cluding that Mr. Coscia was guilty, the district court denied Mr. Coscia’s motion for a new trial. 4. Our Assessment We now evaluate Mr. Coscia’s arguments. We ask first whether he established that the ICE and CME data disclosed after trial constitutes new evidence. In short, Mr. Coscia must demonstrate that he could not have discovered the data sooner through the exercise of due diligence. See United States v. Westmoreland, 712 F.3d 1066, 1073 (7th Cir. 2013). A claim of diligence, however, is seriously undermined when the defendant fails to have a subpoena issued or fails to re- quest a continuance because critical evidence was not avail- able. See United States v. Oliver, 683 F.2d 224, 228 (7th Cir. 1982) (holding that failure to exercise diligence in locating witnesses before trial precluded new trial based on newly discovered evidence). Mr. Coscia must show that the failure to obtain production of this new information was not due to his lack of diligence.

42 Id. at 7.

43 Id. Nos. 19-2010 & 20-1032 25

Recall that, after trial concluded, Mr. Coscia obtained from CME a complete audit trail covering approximately ten weeks of all market participants for every market in which he traded. In addition, while the court was considering the Government’s motion to quash Mr. Coscia’s post-trial sub- poena, ICE disclosed that it inadvertently had failed to turn over the data underlying ICE Summary Chart 6, which set out Mr. Coscia’s share of cancellations of large orders fol- lowing the trade of a small order in the opposite direction. In accordance with the court’s order, ICE then produced the data underlying ICE Summary Chart 6 and disclosed that the original numbers on the chart were misstated: Mr. Coscia accounted for 14,141, not 14,563, of such cancellations, and all other participants accounted for 1,328, not 671, of such cancellations. Mr. Coscia submits that this information constitutes new evidence because he was entitled to rely at trial on the Gov- ernment’s representations that he had all of the data. He points out that, in response to his request for the summary charts, the Government had responded: “We have not yet prepared the summary charts, but the information that they 44 summarize has already been produced to you.” He also notes that ICE witness John Redman confirmed at trial that he had reviewed the data on the ICE disk and that the sum- mary charts were true and accurate summaries of the data on the ICE disk. First of all, to the extent Mr. Coscia intimates that he was not provided any of the data underlying any of the summary

44 R.227-8 at 2. 26 Nos. 19-2010 & 20-1032

45 charts, that contention is, to put it mildly, overblown. Prior to trial, Mr. Coscia obtained data through discovery and his own independent subpoenas. Indeed, Mr. Coscia’s own ex- pert witness used that data to create his own summary charts. We therefore cannot accept Mr. Coscia’s suggestion that he was deprived of complete data underlying all of the ICE and CME summary charts. Indeed, the representations made to him were in large part true. It was only the underly- ing data for ICE Summary Chart 6 that was lacking as re- vealed by ICE’s post-trial disclosure that it inadvertently had failed to produce that material to either party. The newly discovered material with respect to Summary Chart 6 dis- closed errors, but those errors can be characterized accurate- ly as de minimis. Of all the large order cancellations follow- ing small order trades in the opposite direction, Mr. Coscia accounted for 91%, not 96% of such trades. Importantly, the data continues to show that Mr. Coscia, more than any other market participant, engaged in a pattern of cancelling large orders after trading a small order in the opposite direction. Mr. Coscia simply has not demonstrated how the limited non-disclosure of ICE Summary Chart 6 casts doubt on all of the summary charts from both ICE and CME. Nor has he connected the missing data from ICE Summary Chart 6 to any explanation of why he failed to obtain earlier the data he sought after trial. In an effort to meet the latter burden, he points to the four subpoenas that he issued prior to trial as proof of his diligence. 46 These subpoenas were crafted, how-

45 See Appellant’s Br. 39–40, 45–46.

46 See R.227-2; R.227-3; R.227-4; R.227-5. Nos. 19-2010 & 20-1032 27

ever, to produce specifically described information. For ex- ample, Mr. Coscia’s single ICE subpoena requested the audit trail data for only ten trading firms between September 6, 2011 and October 18, 2011, as well as the data underlying the 47 statistics in its Suspicious Trading Report. Mr. Coscia’s subpoenas to CME were cabined to his own orders and trades, orders and trade data of specific entities, and audit trail data for eight markets on four dates. It was not until well after trial that Mr. Coscia sought the full audit trail of all CME transactions and all ICE transactions. Mr. Coscia has not demonstrated why he was unable to obtain, through compulsory process, the full audit trail data he has since ob- tained or why he did not request a continuance prior to trial to obtain those records. Even if we were to assume that this new evidence could not have been discovered sooner through the exercise of due diligence, Mr. Coscia fails to explain convincingly how this new information is material. In the context of a motion for a new trial, evidence is considered material “if there is a rea- sonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different.” United States v. Bagley, 473 U.S. 667, 682 (1985). Mr. Coscia has not carried his burden of demonstrating that the new information here seriously called into question the jury verdict. Instead, the new information would serve only as impeachment evidence against some of the Government’s witnesses. Given the amount and strength of the other evi-

47 See R.227-4. 28 Nos. 19-2010 & 20-1032

dence against Mr. Coscia, this simply does not warrant a new trial. 48 Moreover, his materiality arguments with respect to the post-trial information fail because his proposed modifica- tions to the data analysis presented at trial are either inaccu- rate or misleading. Additionally, many of the purported is- sues with the data could have been elicited on cross-examination at trial. We first examine Mr. Coscia’s at- tempts to recharacterize the analyses presented at trial. We then turn to those matters that could have been examined through cross-examination. Mr. Coscia relies on the post-trial data to recast the sum- mary charts presented at trial. The new data analysis Mr. Coscia urges us to adopt, however, misrepresents the data or requires us to make unjustified inferences. For in- stance, Mr. Coscia requests that we make comparisons be- tween different sets of data that can be compared only by accepting false equivalencies. ICE Summary Chart 3 showed Mr. Coscia’s order-to-trade ratio to be 1,592% on the ICE Brent Futures market, whereas the next firm down had a ra-

48 See Kamel, 965 F.2d at 493 (“A new trial will not be granted if the evi- dence offered is merely impeaching or cumulative; it must be material.”). Although “[i]t is true that, typically, newly discovered impeachment ev- idence does not warrant relief under Rule 33,” United States v. Reyes, 542 F.3d 588, 596 (7th Cir. 2008), we have cautioned that this is not a categor- ical rule, see United States v. Taglia, 922 F.2d 413, 415–16 (7th Cir. 1991). For example, “[i]f the government’s case rested entirely on the uncorrob- orated testimony of a single witness who was discovered after trial to be utterly unworthy of being believed …, the district judge would have the power to grant a new trial in order to prevent an innocent person from being convicted.” Id. at 415. This simply is not the case here. Nos. 19-2010 & 20-1032 29

tio of only 264%. To undermine the ICE data summarized in ICE Summary Chart 3, Mr. Coscia invites our attention to the CME data that he obtained post-trial and asks us to conclude that “literally dozens, and sometimes hundreds” of traders 49 had order-to-trade ratios greater than 1,592%. Mr. Coscia reasons, without any support, that “[g]iven the robustness of the CME data,” we may assume that the additional ICE data would show similar results to the CME data. 50 Focusing on this evidence, he asserts that he was not the outlier the Gov- ernment made him out to be. We decline to rely on CME da- ta to make unsupported assumptions about the validity of the ICE data. Notably, Mr. Coscia has hampered our ability to compare the CME order-to-trade ratios by not sharing what his own order-to-trade ratio was for each of the CME commodities. Mr. Coscia attempts to discredit CME Summary Charts 2 and 3 with another apples-to-oranges comparison. CME Summary Charts 2 and 3 showed that Mr. Coscia had an ag- gregate 35.53% “fill-rate differential,” the difference between Mr. Coscia’s small-order fill rate and large-order fill rate, across all seventeen CME markets. Mr. Coscia, with the new CME data in hand, observes that “dozens and even hun- dreds of traders” had fill-rate differentials greater than his 35.53%, counting 1,189 “unique traders” with larger fill-rate 51 differentials. This summation, however, is achieved by

49 Appellant’s Br. 33.

50 Id. at 33 n.8.

51 Id. at 34–35. 30 Nos. 19-2010 & 20-1032

counting each trader with a commodity-specific differential greater than Mr. Coscia’s aggregate fill-rate differential across all markets. For us to see it Mr. Coscia’s way, we must compare the fill-rate differentials of specific traders in single commodity markets against his aggregate fill-rate differen- tial across seventeen markets. Without establishing whether the other traders traded in each of the same seventeen mar- kets as Mr. Coscia or what the aggregate fill-rate differentials were for each of the “unique traders” Mr. Coscia identifies, Mr. Coscia has not met his burden demonstrating that this is an apt comparison. Mr. Coscia also attempts to use the data to support prop- ositions that we do not think can be fairly maintained. Mr. Coscia challenges several aspects of ICE Summary Chart 6, which showed Mr. Coscia’s share of cancellations of large orders following a small order filled in the opposite direc- tion. Mr. Coscia first contends that ICE Summary Chart 6 was presented to the jury as a chart showing all order cancel- lations among all other participants. Relying on the post-trial data, Mr. Coscia observes that, “of the 71,785,276 cancella- tions in the Brent contracts market traded on ICE, Coscia on- 52 ly accounted for 47,649 or .066% of those canceled orders.” But it is clear that ICE Summary Chart 6 did not display market-wide order data of all cancellations but only revealed a specific subset of cancellations: cancellations of large or- ders following the fill of a small order in the opposite direc- tion. The very first page of the exhibit reflects this subset of cancellations: “Instances where Mr. Coscia cancelled large

52 Id. at 25. Nos. 19-2010 & 20-1032 31

53 orders following an opposite trade.” In addition, Redman clearly confirmed this characterization at trial: So what we did to get to this chart was we looked at how frequently a large order was canceled following the trading of a small order in the other direction. … It looks at everybody else who’s—who had the same instance of large order—small order 54 trades, large orders canceled. Thus, Mr. Coscia’s submission that he actually represent- ed less than 1% of all market-wide cancellations is not at all supported by the evidence. The record is clear that ICE Summary Chart 6 referred to a subset of cancellations only: Mr. Coscia accounted for over 90% of cancellations of large orders that followed the fill of a small order in the opposite direction. The post-trial data correction simply reflects that Mr. Coscia represented 91%, as opposed to 96%, of large or- der cancellations following the trade of a small order in the other direction. We find it difficult to see how this de mini- mis error probably could have led to an acquittal. Mr. Coscia

53 R.177-35 at 1. Mr. Coscia contends that “the presumption that the jury understood the 96% figure” to represent only this subset of cancellations, as opposed to all cancellations, “is belied both by the chart’s more sweep- ing title of ‘Order cancellation comparison.’” Appellant’s Reply Br. 12 n.6. We are unpersuaded by Mr. Coscia’s concerns, as the first page of the exhibit contains the very heading Mr. Coscia claims is lacking; Mr. Coscia has chosen to excerpt the first page of the exhibit from his lead brief. See Appellant’s Br. 23–24. 54 R.86 at 37–38 (Tr. 304–05). 32 Nos. 19-2010 & 20-1032

offers no other justification why this minor error should cast doubt on all of the data evidence. Finally, Mr. Coscia attacks CME Summary Chart 5 by suggesting three ways the data could have been calculated differently. He contends: (1) that order modifications should have been included in the cancellation count; (2) that small and large firms should have been treated differently; (3) and that large orders placed only should have been compared to large orders filled. But where the data or the calculations may have fallen short are matters that should have been dealt with on cross-examination. Indeed, Mr. Coscia’s trial counsel did cross-examine CME witness Ryan Cobb on the 55 exclusion of modifications from the chart. Mr. Coscia also fails to carry his burden of demonstrating that he likely would have been acquitted if the jury had been presented this data or his updated charts. First, as we have discussed, Mr. Coscia’s proposed presentation of the evi- dence does not present the new evidence fairly or accurately. It is a safe assumption that Government counsel would have exposed these shortcomings. More fundamentally, there was a significant amount of other evidence against Mr. Coscia that established his intent to spoof. The jury considered Mr. Coscia’s own testimony, the testimony of programmer Jeremiah Park, the testimony of other traders, and the rebut- tal testimony of Hank Bessembinder.

55 Id. at 150–52 (Tr. 417–19). Nos. 19-2010 & 20-1032 33

For these reasons, we conclude that the district court did not abuse its discretion in denying Mr. Coscia’s motion for a new trial based on the post-trial data. C. Subsequent Indictments of Other Traders Mr. Coscia also submits that subsequent indictments of other traders for spoofing materially undercuts the Govern- ment’s key theory at trial: that Mr. Coscia was an “outlier.” In his view, these subsequent indictments establish that the Government had asserted falsely that Mr. Coscia’s trading activities were unique and that this uniqueness demonstrat- ed his criminal intent. Mr. Coscia also submits a new trial is warranted in his case because the Government treated Park’s testimony differently in the prosecution of another trader, see United States v. Jitesh Thakkar, No. 18-cr-00036 (N.D. Ill.), where it argued that Park’s testimony did not support a finding of intent to spoof. As to whether the subsequent indictments contradict the Government’s characterization of Mr. Coscia as “unique” or as an “outlier,” the district court was entitled to conclude, in the context of a motion for a new trial, that the fact “[t]hat others may have employed illegal trading strategies does not constitute a defense to a criminal indictment based on the 56 employment of illegal trading strategies.” From our review of the record, the case against Mr. Coscia was not built ex- clusively around the Government’s characterization of Mr. Coscia’s trading strategy as “unique.” Significantly, it

56 R.233 at 7. 34 Nos. 19-2010 & 20-1032

included Mr. Coscia’s own admissions about his trading pat- terns, Park’s testimony, and the testimony of other traders. With respect to the Thakkar prosecution, Mr. Coscia con- tends that the Government took a position contrary to the one it took against him. Specifically, Thakkar alleged that the Government engaged in selective prosecution because it de- clined to prosecute Jeremiah Park, the computer program- mer who built Mr. Coscia’s trading program. In reply, the Government stated that “Park’s awareness of Coscia’s intent to spoof is not supported by Park’s own testimony in Coscia. ... Park testified that Coscia never suggested to Park that Coscia was doing something wrong or fraudulent when us- 57 ing Park’s trading programs.” Whether Park was subjec- tively aware of Mr. Coscia’s intent to spoof or whether Park had a subjective intent to spoof, however, is irrelevant to Mr. Coscia’s intent to spoof. We cannot accept Mr. Coscia’s attempt to conflate his own intent with that of Park. Mr. Coscia has failed to establish that the interests of jus- tice warrant a new trial, for either the post-trial data or the subsequent indictments. The district court therefore did not abuse its discretion in denying Mr. Coscia’s new trial mo- tion.

57 Government’s Resp. to Def. Jitesh Thakkar’s Mot. to Dismiss Indict- ment with Prejudice at 9 n.4, United States v. Thakkar, No. 18-cr-00036 (N.D. Ill.). Nos. 19-2010 & 20-1032 35

III The Section 2255 Motion After the district court denied the motion for a new trial, Mr. Coscia filed a motion under 28 U.S.C. § 2255 to vacate his conviction. In this motion, he alleged that his trial coun- sel, Sullivan & Cromwell, had provided ineffective assis- tance of counsel in two ways. He first submitted that Sulli- van & Cromwell had undisclosed conflicts of interest that adversely affected his trial counsel’s performance. Secondly, he alleged that Sullivan & Cromwell had provided ineffec- tive assistance by failing to object to, investigate, or chal- lenge the Government’s statistical evidence. The district court denied his § 2255 motion. We will review each of those allegations in turn. We re- view de novo a district court’s denial of a defendant’s mo- tion to vacate or set aside his convictions pursuant to 28 U.S.C. § 2255. Hall v. United States, 371 F.3d 969, 972 (7th Cir. 2004). We review the district court’s factual findings for clear error. Id. A. The Conflict-of-Interest Allegation 1. Background Mr. Coscia submits that, at the time of his trial, Sullivan & Cromwell represented, either simultaneously or in the past, several government witnesses, including ICE, D.E. Shaw, and Citadel. He further alleges that Sullivan & Cromwell never disclosed such conflict to him and that its representation of an adverse witness constituted an actual conflict of interest. In his view, this conflict of interest incen- 36 Nos. 19-2010 & 20-1032

tivized Sullivan & Cromwell to neglect critical discovery be- cause the information derived from such a discovery process necessarily would impact adversely these clients. Mr. Coscia supported these allegations by noting that, with respect to ICE, Sullivan & Cromwell’s website revealed that it had represented ICE in various transactions over a fourteen-year period. Notably, it represented ICE in a $5.2 billion acquisition that had been finalized on the first day of Mr. Coscia’s trial. Sullivan & Cromwell never disclosed that representation. Furthermore, Sullivan & Cromwell’s lead counsel, Attorney Kenneth Raisler, personally had repre- sented ICE in prior matters. According to Mr. Coscia, Attor- ney Raisler therefore knew that Sullivan & Cromwell had a long-term and valuable attorney-client relationship with ICE. Mr. Coscia contended that, because of its concurrent representation of ICE, Sullivan & Cromwell chose trial strat- egies that would not create difficulties for its long-standing client by failing to ascertain the completeness or accuracy of the summary charts and by failing to cross-examine effec- tively Redman, the ICE representative. Mr. Coscia also alleged that Sullivan & Cromwell previ- ously had represented D.E. Shaw and Citadel, entities whose representatives testified for the Government at his trial. He supported this allegation by submitting attorney profile pages from the Sullivan & Cromwell website showing that various attorneys from the firm had represented D.E. Shaw 58 and Citadel in various transactions. This situation, in Mr. Coscia’s view, amounted to an actual conflict of interest

58 See 2255 R.8-2; R.8-3. Nos. 19-2010 & 20-1032 37

because Sullivan & Cromwell attorneys faced the possibility of having to cross-examine their former clients. Mr. Coscia further speculates that Sullivan & Cromwell “represented other persons or entities who testified at trial, and therefore 59 had yet further conflicts of interest.” Mr. Coscia contended that Sullivan & Cromwell, wary of creating difficulty for these clients, had failed to obtain data from them or cross-examine their representatives effectively. The district court ruled that Mr. Coscia had demonstrat- ed successfully that the firm actively provided legal services to ICE at the time of the trial. It nevertheless concluded that Mr. Coscia failed to show that this simultaneous representa- tion had affected adversely Attorney Raisler’s performance during trial. Taking the same view as it had in disposing of the second motion for a new trial, the district court deter- mined that Sullivan & Cromwell’s strategy was to acknowledge Mr. Coscia’s trading conduct and to justify that trading conduct as legitimate. The court therefore concluded that Mr. Coscia failed to demonstrate that any alleged con- flict adversely affected Sullivan & Cromwell’s representation of him.

59 Appellant’s 2255 Br. 19 n.3 (“In response to a question from one of Coscia’s post-trial attorneys asking whether [Sullivan & Cromwell] had represented any of 28 specific entities who were involved in Coscia’s trial and/or the transactions at issue in Coscia’s case, [Sullivan & Cromwell] provided the following vague but suggestive response: ‘We can confirm that Sullivan & Cromwell LLP represented certain entities (or their affili- ates) listed in your April 3, 2019 letter prior to or during Sullivan & Cromwell’s representation of Mr. Coscia.’” (emphasis omitted)). 38 Nos. 19-2010 & 20-1032

2. Governing Principles The Sixth Amendment guarantees criminal defendants effective assistance of counsel. Included within this right is the right to representation “free from conflicts of interest.” Wood v. Georgia, 450 U.S. 261, 271 (1981). An allegation of the sort presented here is governed by the rule established by the Supreme Court in Cuyler v. Sulli- van, 446 U.S. 335 (1980). Under this rule, the defendant must first establish the existence of a conflict of interest. Once the defendant has established such a conflict, he must further establish that the conflict “adversely affected his lawyer’s performance.” Id. at 348. “[U]ntil a defendant shows that his counsel actively represented conflicting interests, he has not es- tablished the constitutional predicate for his claim of ineffec- tive assistance.” Id. at 350 (emphasis added). This showing, although not as difficult to meet as the Strickland v. Washing- ton, 466 U.S. 668 (1984), prejudice standard, is nevertheless a significant burden. See Spreitzer v. Peters, 114 F.3d 1435, 1450 (7th Cir. 1997); see also Hall, 371 F.3d at 973 (observing that the Sullivan adverse-effect standard is significantly easier to meet than the Strickland prejudice standard). In sum, “[a]n ‘actual conflict,’ for Sixth Amendment pur- poses, is a conflict of interest that adversely affects counsel’s performance.” Mickens v. Taylor, 535 U.S. 162, 172 n.5 (2002). An “adverse effect” can be demonstrated “by showing that ‘but for the attorney’s actual conflict of interest, there is a reasonable likelihood that counsel’s performance somehow would have been different.’” Gonzales v. Mize, 565 F.3d 373, 381 (7th Cir. 2009) (quoting Stoia v. United States, 22 F.3d 766, 771 (7th Cir. 1994)). Nos. 19-2010 & 20-1032 39

3. Our Assessment We first consider Mr. Coscia’s conflict of interest claim as to ICE, and we agree with the district court that there was a conflict of interest with respect to ICE. We conclude, howev- er, that Attorney Raisler’s conflict of interest did not ad- versely affect his performance. With respect to the conflict of interest, Mr. Coscia pre- sented evidence that his trial attorney, Kenneth Raisler, was involved in providing legal and lobbying services to ICE in the years prior to Mr. Coscia’s criminal proceeding. In addi- tion, Sullivan & Cromwell was directly providing legal ser- vices to ICE while Attorney Raisler was representing 60 Mr. Coscia. We have noted the importance of “the pre- sumption that the lawyer will subordinate his pecuniary in- terests and honor his primary professional responsibility to his clients in the matter at hand.” United States v. Jeffers, 520 F.2d 1256, 1265 (7th Cir. 1975). Attorney Raisler’s prior direct involvement and his firm’s simultaneous involvement in the representation of ICE in other matters at the time of Mr. Coscia’s trial, and the failure to disclose such conflict, is cause for concern that loyalties may have been divided. See Rosenwald v. United States, 898 F.2d 585, 587 (7th Cir. 1990) (“The pragmatic pressure on counsel in cases such as these is purely financial—the lawyer does not want to lose a client whether that client is seeking advice on civil or on criminal matters. The ethical dilemma is also the same—the attorney must still guard secrets and confidences and must seek to promote the client’s interests ….”). Here, Redmond’s testi-

60 See 2255 R.8-1. 40 Nos. 19-2010 & 20-1032

mony about Mr. Coscia’s trading conduct on ICE’s exchange, the role that Attorney Raisler had played in advising ICE on regulatory and lobbying matters, and the financial stake Sul- livan & Cromwell had in the simultaneous matters concern- ing ICE, support the district court’s determination. As we noted earlier, the presence of a conflict of interest, standing alone, does not carry the day for Mr. Coscia. Hav- ing established that Sullivan & Cromwell had a conflict with respect to ICE, Mr. Coscia still must demonstrate that there is a reasonable possibility that, absent this conflict of interest, his counsel’s representation was adversely affected. In an effort to carry this burden, Mr. Coscia submits that, because his counsel’s firm had a conflict of interest, the attorney did not obtain or verify the data underlying the ICE summary charts. If this underlying statistical evidence had been avail- able, Mr. Coscia continues, it would have “expose[d] the ob- jective inaccuracies in the prosecution’s charts” and demon- 61 strated that his trading strategy was not unique. Mr. Coscia also maintains that an unconflicted lawyer would have, through more effective cross-examination, rebutted the Gov- ernment’s assertion that Mr. Coscia was an outlier in his 62 trading activity.

61 Appellant’s 2255 Br. 29.

62 For example, Mr. Coscia points to his trial counsel’s failure to cross-examine ICE representative Redman about a $122,180 loss he in- curred. At trial, Redman testified that this loss resulted from a large or- der being filled; Mr. Coscia now alleges that this loss actually resulted from multiple small orders. Redman’s testimony, however, supported Mr. Coscia’s defense theory: he was indifferent to whether large or small orders were filled, and that he “wanted to trade” each large order he (continued … ) Nos. 19-2010 & 20-1032 41

Mr. Coscia’s argument encounters some very strong headwinds. At the outset, the newly discovered post-trial data does not uncover the large inaccuracies he claims. 63 It demonstrates, at most, a mild variation in the cancellation of large trades following small orders in the opposite direction. Moreover, Mr. Coscia’s defense at trial was to acknowledge that his trading activity was unique, but wholly above 64 board. His approach, quite understandably, was to present

( … continued) placed on the market. R.89 at 183 (Tr. 940). In closing, Mr. Coscia’s de- fense counsel told the jury: “[J]ust because the small side actually trades more often than the large side that you really didn’t have anything at stake …. But the truth is, and we’re going to see this, the large side was filled in full or in part more than 8,000 times.” R.92 at 94 (Tr. 1507). De- fense counsel went on: “Sometimes the large side was profitable. Some- times it was not. Sometimes you make money in trading. Sometimes you don’t.” Id. at 103 (Tr. 1516). 63 Mr. Coscia again insists that ICE Summary Chart 6 falsely “claimed that Coscia was responsible for 96% of all order cancellations on the Brent Crude futures market,” and that it was not until post-trial that ICE disclosed that ICE Summary Chart 6 represented a subset of cancella- tions flagged by an ICE system alert tool. Appellant’s 2255 Br. 26–27. Mr. Coscia urges that not only was the methodology of the alert tool “never fully explained either during trial or after,” but that trial counsel failed to contest the “devastating statistic … that Coscia accounted for 96% of all cancellations on the Brent Crude market.” Id. at 27. As we dis- cussed with respect to Mr. Coscia’s new trial motion, the record indicates that ICE Summary Chart 6 was clearly presented to the jury for what it was: large orders cancelled after small orders filled in the opposite direc- tion. The chart’s title and Redman’s testimony clearly describe that ICE Summary Chart 6 reflected a subset of, not all, cancellations. 64 See, e.g., R.82 at 170 (Tr. 170) (“Every order to buy or sell a futures con- tract that he placed into the market was a real, legitimate order that was (continued … ) 42 Nos. 19-2010 & 20-1032

his trading activity as legitimate and to argue that there was no evidence that his trading behavior manifested an intent to spoof. He submitted that he was indifferent as to whether his large or small orders were filled and that every order he placed, regardless of size, was capable of being filled and therefore legitimate. The focus of Mr. Coscia’s defense was that he was not at- tempting to rig the market through spoofing. The post-trial statistical evidence recovered is only mildly relevant and probative to this defense. By diluting, somewhat, the Gov- ernment’s assertion that he was an outlier in his trading methodology, the new evidence provides some circumstan- tial evidence relevant to whether he was attempting to rig the market. In light of the other evidence, however, this mild variation of ICE Summary Chart 6 could not have made a significant difference in the jury’s determination. Consider- ing the design of the program and the testimony of Park, Mr. Coscia’s programmer, that Mr. Coscia wanted the pro- gram to act “[l]ike a decoy” to “pump [the] market,” it was entirely reasonable for trial counsel to pursue the chosen strategy. 65 In the end, Mr. Coscia set up a system designed to spoof the market.

( … continued) available for others in the market to trade. … Sometimes he lost money. Sometimes he made money.”); R.92 at 52 (Tr. 1465) (“We don’t dispute, in short, that he had a different strategy, but there’s nothing wrong or unlawful about having a different strategy. There’s nothing wrong or unlawful about having an unusual strategy.”). 65 R.86 at 231, 235 (Tr. 498, 502). Nos. 19-2010 & 20-1032 43

The situation is somewhat different with respect to D.E. Shaw or Citadel. Here, Mr. Coscia has not demonstrated that a conflict of interest existed. Relying on Hall and Enoch v. Gramley, 70 F.3d 1490 (7th Cir. 1995), Mr. Coscia contends that “even the possibility of cross-examining a former client constitutes an actual conflict.” 66 This position misses the mark: In Enoch, 70 F.3d at 1498, we declined to adopt a rule “that any lawyer has a conflict of interest when he cross-examines a former client.” Although the possibility of having to cross-examine a former client may lead to an actual conflict of interest, we have held that the defendant must show one of two things: “(1) that the attorney’s representa- tion of the first client was ‘substantially and particularly re- lated to his later representation of defendant’ or (2) that the attorney actually ‘learned particular confidential information during the prior representation of the witness that was rele- vant to defendant’s later case.’” Hall, 371 F.3d at 973 (quoting Enoch, 70 F.3d at 1496–97). Mr. Coscia has not established that Sullivan & Crom- well’s representation of D.E. Shaw in its unrelated structured private transactions or of Citadel in its unrelated investment transactions was substantially and particularly related to the representation of Mr. Coscia. The cases to which Mr. Coscia invites our attention involve the same matter or concern co-defendants, and therefore do not govern the situation be- fore us. For instance, in Hall, we held that there was an actu- al conflict of interest when an attorney’s representation of a witness “enabled him to learn confidential information per-

66 Appellant’s 2255 Br. 18. 44 Nos. 19-2010 & 20-1032

taining directly to Hall’s case.” Id. at 973. In Ross v. Heyne, 638 F.2d 979, 982 (7th Cir. 1980), “one attorney represented the defendant while his law partner represented co-defendants who testified for the prosecution.” This type of multiple representation presented an actual conflict be- cause the “two co-defendants testified against Ross in ex- change for favorable treatment by the state,” and the de- fendant’s attorney “was unable to cross-examine them effec- tively.” Id. at 983. 67 And in United States v. Moscony, 927 F.2d 742, 747–51 (3d Cir. 1991), the Third Circuit held that an ac- tual conflict of interest existed when defense counsel previ- ously had represented employees of the defendant who would be testifying for the government. Defense counsel could not effectively impeach the witnesses “without reveal- ing information ‘relating to’ his representation of them.” Id. at 750. Mr. Coscia has not established that Sullivan & Cromwell learned of relevant and confidential information from its representations of D.E. Shaw and Citadel in unrelated trans- actions. He has presented no facts to suggest that Mr. Coscia’s trial lawyers were torn between the duty of

67 See also McElrath v. Simpson, 595 F.3d 624 (6th Cir. 2010) (finding an actual conflict of interest in joint representation of multiple defendants in same case); Boykin v. Webb, 541 F.3d 638 (6th Cir. 2008) (same); McFarland v. Yukins, 356 F.3d 688 (6th Cir. 2004) (same); United States v. Basham, 918 F. Supp. 2d 787 (C.D. Ill. 2013) (finding an actual conflict of interest when prior client and defendant were charged with the same offense, and prior client admitted defendant’s involvement); United States v. Ring, 878 F. Supp. 134 (C.D. Ill. 1995) (finding an actual conflict of interest when counsel previously represented in related matter client who would be testifying as material government witness). Nos. 19-2010 & 20-1032 45

confidentiality to their firm’s former client and their duty of loyalty to Mr. Coscia. Inviting our attention to United States v. Alex, 788 F. Supp. 359 (N.D. Ill. 1992), Mr. Coscia further contends that “a law- yer cannot represent a defendant if he previously represent- ed a victim of the crime.” 68 Alex is not binding precedent, but, in any event, Mr. Coscia overstates the district court’s conclusion. In Alex, an attorney undertook representation of a criminal defendant while simultaneously representing sev- eral of the alleged victims of the defendant’s extortionate conduct in a grand jury investigation. Id. at 362. The court concluded that, because of his representation of the victims, counsel was “aware of certain matters which could be used to attack the credibility of the witnesses at trial.” Id. at 364. The court’s concern was grounded in the fact that the attor- ney sought “to represent one of the alleged perpetrators of the criminal activity when he and his firm previously repre- sented individuals who were allegedly victims of the very same criminal activity.” Id. Even if we were to assume that D.E. Shaw and Citadel can be characterized as victims of Mr. Coscia’s spoofing conduct, Mr. Coscia does not allege, and there is nothing in the record to suggest, that Sullivan & Cromwell represented D.E. Shaw or Citadel as victims in the context of Mr. Coscia’s criminal activity. See Enoch, 70 F.3d at 1497 (concluding no actual conflict when attorney represented adverse witness for “entirely separate” matters “four years apart, and none of the relevant people involved had cross-cutting relation-

68 Appellant’s 2255 Br. 21–22. 46 Nos. 19-2010 & 20-1032

ships”). Here, Mr. Coscia has offered only speculation that Sullivan & Cromwell’s prior representation of D.E. Shaw and Citadel involved a possible conflict of interest. “[T]he possibility of conflict,” however, “is insufficient to impugn a criminal conviction.” Sullivan, 446 U.S. at 350. 69 B. Ineffective Assistance of Counsel 1. Background As a second ground for relief under § 2255, Mr. Coscia also brings an ineffective assistance of counsel claim on the ground that, even in the absence of a conflict of interest, his trial counsel provided ineffective assistance. His petition al- leged that trial counsel’s failure to object to, investigate, and

69 Mr. Coscia insists that his trial counsel “took a disturbingly light touch” by failing to obtain the algorithm program logic of the trading firms. Appellant’s 2255 Br. 29. He submits that “[t]he program logic from D.E. Shaw would likely have shown that it, too, was designed to cancel orders … which would have been important to show Coscia was not an ‘outlier’ or ‘unique’ in his trading.” Id. at 30. But again, Mr. Coscia’s de- fense theory was to not dispute his algorithm or that he was an outlier. See R.82 at 172 (Tr. 172) (“Let me get one thing right out of the way. We don’t disagree that Michael Coscia came up with the idea for a comput- er-driven trading program. He’s admitted that. We don’t dispute that it worked generally as the prosecution has described.”); R.92 at 53 (Tr. 1465) (“We don’t dispute how Michael’s strategy worked. We don’t dis- pute that it worked differently from other high-frequency traders. We don’t dispute that Michael placed more large orders than other high-frequency traders, and we don’t dispute that Michael traded … more large orders than other high-frequency traders.”); Id. at 129 (Tr. 1542) (“Now, as Michael told you and Professor Bessembinder also con- firmed, there’s really no dispute about how the algorithm worked.”). Nos. 19-2010 & 20-1032 47

challenge certain evidence rendered his trial counsel’s per- formance constitutionally deficient. The Government coun- tered that Mr. Coscia failed to satisfy either the defi- cient-performance or prejudice prong of Strickland v. Wash- ington, 466 U.S. at 668. It emphasized that trial counsel’s de- cisions were strategic: He characterized Mr. Coscia’s conduct as a legitimate trading strategy. And in any event, the jury saw and heard a significant amount of evidence, including Mr. Coscia’s own testimony, along with that of his pro- grammer, Park. The district court agreed with the Government that Mr. Coscia could not establish either Strickland prong. In its view, Mr. Coscia’s case presented “the common situation” where an attorney concludes that “the client stands a better chance of success by admitting the underlying actions al- leged to have been taken by the client which appear to be easily provable, and instead argue to the jury that the actions 70 do not amount to a crime.” The court concluded that trial counsel’s strategic decision was objectively reasonable, and there was no reasonable probability that a different strategy would have led to a different outcome. Accordingly, the dis- trict court denied Mr. Coscia’s § 2255 motion. 2. Governing Principles It is well established that the right to counsel is the right to effective assistance of counsel. Id. at 686. Under the two-prong test set forth in Strickland, 466 U.S. at 687–88, inef- fective assistance of counsel is established by showing that

70 2255 R.15 at 6–7. 48 Nos. 19-2010 & 20-1032

trial counsel’s performance fell below an objective standard of reasonableness and that the deficient performance was prejudicial. To satisfy the first prong, the petitioner must first “show that counsel provided constitutionally deficient performance, meaning counsel made errors so serious he was not functioning as the counsel guaranteed the defendant by the Sixth Amendment.” Winfield v. Dorothy, 956 F.3d 442, 451 (7th Cir. 2020) (internal quotations omitted). In evaluat- ing such claims, we must “indulge a strong presumption that counsel’s conduct falls within the wide range of reason- able professional assistance.” Strickland, 466 U.S. at 689. To satisfy the second prong, the petitioner also must “show that this deficient performance prejudiced his defense—meaning there is a ‘reasonable probability that, but for counsel’s un- professional errors, the result of the proceeding would have been different.’” Winfield, 956 F.3d at 451 (quoting Strickland, 466 U.S. at 694). 3. Mr. Coscia’s Contentions Mr. Coscia submits that his trial counsel failed to investi- gate and offer probative evidence to undermine the Gov- ernment’s case by (1) failing to investigate or challenge inac- curate summary charts; (2) failing to obtain and introduce evidence supporting Mr. Coscia’s good faith; and (3) failing to impeach Government witnesses. The Government submits that Mr. Coscia’s “claims of de- ficient performance relate to strategic choices made by de- 71 fendant’s trial counsel.” The data and summary charts in-

71 Appellee’s 2255 Br. 49. Nos. 19-2010 & 20-1032 49

dicated clearly that Mr. Coscia’s trading practices were dif- ferent and unusual from that of other high-frequency trad- ing firms. We already have rejected Mr. Coscia’s argument that the data was as inaccurate as he claims. More importantly, the strategic choice to accept the trading data as presented but argue that nothing was wrong with his trading practices was a reasonable decision in light of other evidence against Mr. Coscia. Park’s testimony that Mr. Coscia directed him to design a program to avoid large orders being filled made it difficult for Mr. Coscia to deny his trading choices. Mr. Coscia’s prior deposition testimony for the Commodity Futures Trading Commission confirms that the choice to admit Mr. Coscia’s conduct but to argue that it was legal was entirely reasonable. Mr. Coscia insists that his trial counsel failed to pursue a good-faith defense. But, as the Government submits and the district court recognized, Mr. Coscia’s defense strategy at 72 trial was a good-faith defense. From the outset, Mr. Coscia’s defense acknowledged the differences between his trading activities and that of other traders. Trial counsel submitted

72 2255 R.15 at 5 (“At trial, Petitioner’s defense was to acknowledge his trading activity, which was testified to by employees of ICE and the oth- er entities he claims created the conflict. Petitioner attempted to justify his trading activities as being wholly legal and proper. … His defense was that each and every order he placed, both large and small, was a legitimate order that was capable of being filled prior to cancelation.” (citations omitted)). 50 Nos. 19-2010 & 20-1032

that Mr. Coscia’s choice of conduct was “just good trad- 73 ing.” Mr. Coscia next asserts that trial counsel failed to im- peach various government witnesses. For example, he points to counsel’s failure to use prior inconsistent statements to impeach Dermenchyan, an employee at Teza Technologies. In an earlier interview with the Financial Services Authority, Dermenchyan stated that he did not know whether the same participant who placed the large orders was the same partic- 74 ipant placing small orders on the other side. At trial, Der- menchyan stated: [L]arge size[] [orders] were placed in the mar- ket in order to induce participants to trade on the opposite side, and … it was clear it was an individual participant doing this. And if you follow the logic, it was clear that it was a par- ticipant playing with supply and demand in order to push prices in one direction and then 75 push them back in the other direction. These statements, however, are not inconsistent with one another. Dermenchyan’s testimony at trial concerned large orders only, not whether the same participant was placing large orders and small orders in the opposite direction. Mr. Coscia further complains that his trial counsel did not

73 R.82 at 186 (Tr. 186).

74 See 2255 R.5-22 at 15.

75 R.88 at 80 (Tr. 685). Nos. 19-2010 & 20-1032 51

adequately dispute the trading practice; yet his defense be- fore the jury was to acknowledge and admit to his trading practices. Mr. Coscia also submits that his trial counsel was defi- cient in failing to cross-examine Alex Gerko of GSA Capital about GSA Capital’s settlement with CME regarding matched orders. Gerko testified to his observation of a par- ticular pattern of activity, where he “would see very, very large orders appearing on the market indicating some kind of very sharp imbalance between buyers and sellers, and then these orders would disappear from the bid and appear 76 on the offer and … repeat tens of times in a row.” Cross-examining Gerko on GSA Capital’s settlement with CME, however, would have done little for Mr. Coscia’s de- fense because his defense strategy was to admit to his trad- ing practices. The failure to cross-examine here is insufficient to overcome the “strong presumption” that counsel’s deci- sions fell “within the wide range of reasonable professional assistance.” Strickland, 466 U.S. at 689. We agree with the district court’s assessment that Mr. Coscia’s trial counsel was presented with “the common situation” where “the client stands a better chance of success by admitting the underlying actions alleged to have been taken by the client which appear to be easily provable, and instead argue to the jury that the actions do not amount to a crime.” 77 That the jury did not accept his defense does not

76 Id. at 90 (Tr. 695).

77 2255 R.15 at 6–7. 52 Nos. 19-2010 & 20-1032

render it constitutionally deficient. We cannot conclude that his trial counsel’s performance was so deficient as to fall be- low an objective standard of reasonableness. Even if we were to assume that trial counsel’s perfor- mance was deficient, Mr. Coscia has not demonstrated prej- udice. Given the strong evidence of Mr. Coscia’s intent, it is highly improbable that the introduction of a weak statistical characterization of Mr. Coscia’s trading patterns would have led to a different outcome. Not only did Park testify as to his instructions, but Mr. Coscia, himself, testified to how he de- signed his trading programs to work. His ineffective assis- tance of counsel claim therefore must fail. C. Evidentiary Hearing Finally, we turn to the district court’s denial of Mr. Coscia’s request for an evidentiary hearing. We review this denial for abuse of discretion. See Kafo v. United States, 467 F.3d 1063, 1067 (7th Cir. 2006). When a petitioner “alleg- es facts that, if proven, would entitle him to relief,” the dis- trict court must grant an evidentiary hearing. Bruce v. United States, 256 F.3d 592, 597 (7th Cir. 2001) (quoting Stoia, 22 F.3d at 768). The court, however, is not required to grant an evi- dentiary hearing when “the motion and the files and records of the case conclusively show that the prisoner is entitled to no relief.” 28 U.S.C. § 2255(b). “In addition, a hearing is not necessary if the petitioner makes conclusory or speculative allegations rather than specific factual allegations.” Daniels v. United States, 54 F.3d 290, 293 (7th Cir. 1995); see also Aleman v. United States, 878 F.2d 1009 (7th Cir. 1989) (rejecting hear- Nos. 19-2010 & 20-1032 53

ing request when petitioner “offer[ed] conjecture, not facts” that certain witnesses were informants). Mr. Coscia submitted that he became aware only after trial that trial counsel simultaneously or previously repre- sented ICE, D.E. Shaw, and Citadel. He sought an eviden- tiary hearing and requested discovery in the form of a doc- ument subpoena to Sullivan & Cromwell to determine, among other things, whether Sullivan & Cromwell repre- sented ICE, D.E. Shaw, Citadel, or any other government witnesses or trading firms identified in the indictment; the subject matter and time period of those representations; which attorneys were involved or knew about those repre- sentations; the conflict check procedures at Sullivan & Cromwell; and the communications between, and the fees obtained from, any parties Sullivan & Cromwell represent- ed. Mr. Coscia also requested to depose his trial counsel. As we already have concluded that trial counsel present- ed an actual conflict with ICE, we evaluate whether the dis- trict court erred in denying his request for an evidentiary hearing concerning his conflict allegations related to D.E. Shaw and Citadel. He contends that Sullivan & Cromwell’s prior representation of these entities explains what he char- acterizes as the “disturbingly light touch” given by trial counsel. 78 Mr. Coscia submits that, without such a conflict of interest, his trial counsel would have obtained data or elicit- ed testimony from these entities to show that Mr. Coscia’s trading patterns were not abnormal or an outlier.

78 Appellant’s 2255 Br. 29. 54 Nos. 19-2010 & 20-1032

This defense, however, was the opposite of Mr. Coscia’s actual defense strategy to admit Mr. Coscia’s trading prac- tices but submit that they were above board. And as we have discussed, in light of Park’s testimony and Mr. Coscia’s own testimony, Mr. Coscia’s good-faith defense was a more than reasonable strategy. Mr. Coscia has submitted only that trial counsel had pre- viously represented adverse witnesses in unrelated transac- tional matters and that they therefore must have obtained confidential information. He has not provided any specific allegations as to what relevant information might have been obtained from those transactions or how such information would have affected this case. We therefore conclude that the district court did not abuse its discretion in denying Mr. Coscia’s request for an evidentiary hearing. Conclusion The district court did not abuse its discretion in denying Mr. Coscia’s new trial motion on the basis of new evidence. We also affirm the district court’s denial of Mr. Coscia’s § 2255 motion. Even though we agree with the district court that Mr. Coscia’s trial counsel had a conflict, the district court properly concluded that counsel’s performance was not adversely affected. Finally, Mr. Coscia cannot establish either prong of his ineffective assistance claim. For these rea- sons, we affirm the judgments of the district court. AFFIRMED