Appeal from the United States District Court for the Southern District of Illinois. No. 3:17-cv-01349-JPG — J. Phil Gilbert , Judge .

ARGUED MAY 22, 2025 — DECIDED SEPTEMBER 24, 2025

Before BRENNAN, SCUDDER, and KIRSCH, Circuit Judges .

BRENNAN, Circuit Judge . Otis Elion pleaded guilty to distributing methamphetamine in 2017. The federal district court imposed a lengthy sentence for being a “career offender.” U.S. SENT'G GUIDELINES MANUAL § 4B1.1(a). Elion's attorney did not object, because after researching whether his prior convictions qualified as predicate offenses, she concluded that challenging that sentencing enhancement would fail.

Page 2In a federal habeas petition, Elion argued his attorney’s failure to object was deficient and prejudicial, amounting to ineffective assistance of counsel under Strickland v. Washington , 466 U.S. 668 (1984). In Elion v. United States ( Elion I ), we ruled that he was prejudiced and remanded for the district court to examine his attorney’s performance under Strickland in the first instance . 76 F.4th 620, 635 (7th Cir. 2023).

This successive appeal concerns whether his attorney performed deficient ly. We conclude she did not. She identified and researched the correct issues and properly applied the categorical approach. Her failure to reach the correct legal conclusion does not alone establish deficient performance. At the time of Elion’s sentencing, both her conclusion and overall performance were reasonable, so w e affirm the district court.

A federal defendant with two or more prior state convictions for a 'controlled substance offense' may receive a longer sentence under U.S.S.G. § 4B1.1(a). In this case, determining whether a state conviction counts as a “controlled substance offense' requires applying the categorical approach. Elion I , 76 F.4th at 625. Under that approach, the elements of the state statute of conviction are compared to the federal definition of 'controlled substance offense .” Id. The state statute triggers the enhancement “only if its statutory elements are defined in such a way that all possible violations of the statute, however committed, would fall wi thin Congress’s chosen federal benchmark.” United States v. Liestman , 97 F.4th 1054, 1056–57 (7th Cir. 2024) (en banc).

Page 3But if the state statute's elements are broader than the Guideline 's definition , the court then decides whether the statute is divisible. Elion I , 76 F.4th at 626. A divisible statute creates multiple crimes with alternative elements; an indivisible statute creates a single crime with “a single (or ‘indivisible') set of elements.' Mathis v. United States , 579 U.S. 500, 504–05 (2016); see also United States v. Edwards , 836 F.3d 831, 833 (7th Cir. 2016) (“The Supreme Court [in Mathis ] recently clarified that a statute is considered divisible only if it creates multiple offenses by setting forth alternative elements .”). Courts employ different tools to ascertain whether a statute is divisible. Caselaw and textual evidence are examined fi rst. Elion I , 76 F.4th at 630. If the answer is still unclear, a court inspects the defendant's record documents, like an indictment or jury instructions, and “the record would then reveal what the prosecutor has to (and does not have to) demonstrate to prevail.” Mathis , 579 U.S. at 518–19. If indeterminacy remains, the statute is treated as indivisible. Elion I , 76 F.4th at 634; Taylor v. United States, 495 U.S. 575, 602 (1990). “[C]lear signals” are required to “convince us to a certainty that the elements are correct and support divisibility before imposing additional federal consequences for those state convictions.” Najera-Rodriguez v. Barr , 926 F.3d 343, 356 (7th Cir. 2019) (citing Taylor , 495 U.S. at 602); see also Elion I , 76 F.4th at 635 (same).

If a statute is divisible, the “ modified categorical approach” requires a court to “scrutinize certain record documents to determine what crime, with what elements, the defendant was convicted of.' Elion I , 76 F.4th at 626 (citation modified) (quoting Mathis , 579 U.S. at 505-06). If those elements meet the Guideline’s definition , then the Guideline applies, and the defendant can receive a longer sentence.

Page 4Against this backdrop, we consider Elion’s case. The facts are relayed in more detail in Elion I , 76 F.4th at 620. Elion pleaded guilty to three counts of distributing methamphetamine. 21 U.S.C. § 841(a)(1), (b)(1)(C). With several prior convictions, he may have been eligible for a longer sentence as a 'career offender' under § 4B1.1(a). His eligibility turned on whether he had twice committed a “controlled substance offense.” Id . A 'controlled substance offense' is defined as a federal or state offense that 'prohibits the manufacture, import, export, distribution, or dispensing of a controlled substance (or a counterfeit substance)” or possession with intent to engage in those actions, punishable by imprisonment for a term exceeding one year. Id. ; § 4B1.2(b).

Three of Elion's convictions 1 ostensibly qualified as 'controlled substance offense [s].” Elion I , 625 F.4th at 624. Most recent was a 2006 federal conviction for distribution of a cocaine base. 21 U.S.C. § 841(a)(1), (b)(1)(C). Before that was a 2000 conviction for unlawful delivery of a look-alike substance. 720 ILCS 570/404(b) ('§ 404(b)'). The oldest conviction was in 1999 for unlawful delivery of a look-alike substance within 1,000 feet of public housing property. 720 ILCS 570/407(b)(3) (“§ 407(b)(3)”). This is a penalty statute, “elevating sanctions for offenders who commit enumerated offenses in certain locations.” Elion I , 76 F.4th at 62627 (citation modified). The Page 5penalty statute, § 407(b)(3), uses § 404(b)—the same statute as his 2000 conviction—as its underlying offense.

The Probation Office concluded that those convictions rendered Elion eligible as a career offender under § 4B1.1. Elion I , 76 F.4th at 623. His attorney, Judith Kuenneke, also researched this issue. She reached the same conclusion, so she did not object to the application of the enhancement at Elion’s sentencing. Id. at 624. Instead, she argued Elion's health and life circumstances justified a mitigated sentence. The district court sentenced Elion as a 'career offender' to 167 months imprisonment.

After Elion’s sentencing, he petitioned for federal habeas relief. 28 U.S.C. § 2255. To him, neither his 1999 nor his 2000 state convictions were 'controlled substance offense [s].” Kuenneke’s failure to make that argument therefore violated his right to effective assistance of counsel under the Sixth Amendment. The district court disagreed.

We reversed. Elion I , 76 F.4th at 635. Using the modified categorical approach, we 'examine[d] the fit between” § 404(b) and Guideline § 4B1.1(a). Id. at 626-27. First, § 404(b) uses the term “look-alike” but the Guideline uses “counterfeit substance.” Id. Though superficially synonymous, the se terms could be different because the definition of 'counterfeit” may require an intent to deceive, while “look-alike” may not. Id. at 628. Kuenneke, by contrast, had concluded that “counterfeit” and “look-alike” were a categorical match. We ultimately passed on the question. Id.

There was another mismatch between the two terms: § 404(b) punished 'advertising' a look-alike substance, yet the Guideline did not. Id. That meant § 404(b) 'punishes a Page 6broader range of conduct' than the Guideline. Id. at 630. Thus, § 404(b) could not serve as a predicate to § 4B1.1(a) unless it was divisible. Id.

We concluded that § 404(b) was indivisible. After consulting state and federal caselaw, that statute’s text, documents from Elion’s conviction, and applicable jury instructions, we could not “conclusively determine” whether the statute was divisible. Id. at 630-34. That uncertainty meant Guideline § 4B1.1 should not apply to Elion. Id. at 634 (“[W]e cannot satisfy ' Taylor 's demand for certainty.''). This conclusion differed from Kuenneke's, as she believed § 404(b) was divisible. Elion thus was prejudiced by Kuenneke's failure to object, meeting one of Strickland 's two prongs. This case was remanded “so the district court may examine Strickland 's deficient performance prong in the first instance.' Id. at 635.

The district court held an evidentiary hearing. Kuenneke testified to her approach, research, and analysis about whether there was a basis to challenge Elion’s careeroffender designation. She also conceded there were mistakes in an affidavit she had filed previously with the court in response to Elion’s habeas petition.

In a meticulous, lengthy opinion, the district court concluded that Kuenneke did not perform deficiently . Elion v. United States , No. 3:17-cv-01349-JPG, 2024 WL 4527803 (S.D. Ill. Oct. 17, 2024). The court made several findings : Kuenneke had identified the correct statute and applied the modified categorical approach correctly. She researched the many issues and did not rely unduly on out-of-circuit caselaw. Finally, Kuenneke also mistakenly wrote down the wrong statute in her affidavit , but she used the appropriate statute in her analysis.

Page 7The district court then analyzed her performance. She had taken “all appropriate precautions, performed extensive research, used all the avenues at her disposal, and exhausted all available tools.” Id. at *30 (citation modified) . And “[w]hile Kuenneke was ultimately incorrect-her conclusion was based on a reasonable, alternative interpretation based on her professional judgment, informed by research and decades of experience.' Id. The district court concluded her performance was better than that of the atto rney in Bridges v. United States , 991 F.3d 793 (7th Cir. 2021), a case we suggested could assist the court. Elion I , 76 F. 4th at 635.

Elion appeals.² We review factual findings for clear error and legal conclusions de novo. Pettis v. United States , 129 F.4th 1057, 1062 (7th Cir. 2025).

We start with Elion's challenge to the district court's findings of fact.

After Elion fi led his habeas petition, the district court allowed Kuenneke to respond to his claim of ineffective assistance of counsel. She did so in an affidavit. Th at document describes Kuenneke's discussions with Elion and her legal analysis about his career offender status . Relevant here, the affidavit stated that § 407(b)(3) “was subject to review pursuant to the modified categorical approach.' But because § 407(b)(3) is an enhancement statute, it should not have been used for the categorial approach. Rather, the underlying offense, § 404(b), was the correct statute to apply. See Elion I , 76Page 8 F.4th at 626–27. The categorical approach requires identifying elements of a crime, and penalty statutes do not list those elements. Kuenneke also mistakenly attach ed a copy of § 407(b)(3) to her affidavit.

At the post-remand evidentiary hearing, Kuenneke testified she reviewed § 404(b), not § 407(b)(3), but mistakenly wrote down § 407(b)(3). Beyond that, she “couldn’t explain” how the error occurred.

The district court found Kuenneke testifi ed truthfully. It reverse-engineered how the mistake could have happened, relying on contextual clues in the affidavit. The court found no reason why she would lie about her performance, particularly when that falsehood could draw scrutiny and possible perjury charges. Kuenneke had even offered to show her notes to the court in camera, which the court concluded buttressed her truthfulness.

Understandably, Elion asks us to discount the district court’s findings of fact. He argues we should rely on Kuenneke's affidavit r ather than credit her testimony at the hearing. But the district court's factual findings were in part grounded in a credibility determination. In other words, to believe Kuenneke mistakenly wrote down the wrong statute hinges on her truthfulness, which is “fundamentally an issue of credibility.” Elion , 2024 WL 4527803, at *21.

A court of review may disregard a credibility determination only if it is 'completely without foundation.' United States v. Clark , 134 F.4th 480, 482 (7th Cir. 2025) (quoting United States v. Sands , 815 F.3d 1057, 1061 (7th Cir. 2015)). This deferential standard of appellate review exists because trial courts have “the best opportunity to observe the verbal and Page 9nonverbal behavior of the witnesses” by “focusing on the subject’s reactions and responses to the interrogatories, their facial expressions, attitudes, tone of voice, eye contact, posture and body movements, as well as confused or nervous speech patterns in contrast with merely looking at the cold pages of an appellate record.” United States v. Eddy , 8 F.3d 577, 582–83 (7th Cir. 1993) ( citation modified ). As a result, credibility determinations can virtually never be clear error. United States v. Harris , 124 F.4th 1088, 1092 (7th Cir. 2025).

The district court made a credibility determination when it found Kuenneke truthful that she made a mistake. It noted her lengthy experience as a defense attorney and that she was highly unlikely to lie. Also, the court explained why, despite Kuenneke's mistakes in her affidavit, she performed the categorical approach by analyzing § 404(b). For example, § 407(b)(3) references § 404(b), thus “even if Kuenneke put the cart before the horse and jumped straight to the divisibility analysis, even if she treated Elion’s Section 407(b)(3) conviction as a fictitious discrete offense, she would still need to evaluate Section 404(b) as an ‘element’ of a Section 407(b)(3) violation.' In the affidavit, a citation to United States v. Hinkle , 832 F.3d 569, 576–77 (5th Cir. 2016) (a Texas controlled-substance statute is indivisible and not a categorical match with the Guideline), along with other textual clues, was 'further evidence that she analyzed the correct statute.”

Elion counters that the district court made no specific factual findings about Kuenneke's demeanor on the stand, so we should disregard its findings, as in Ray v. Clements , 700 F.3d 993, 1013 (7th Cir. 2012). That overreads Ray ’s holding. There, this court stated: “ the district court made no finding Page 10concerning Ray's demeanor or presentation, and instead based its 'credibility' fi nding on nothing more than a string of speculative doubts, none of which were based on any competent contradictory evidence presented by the state.” Id. Ray thus does not grant free rein to appellate courts to disregard credibility determinations when a district court does not opine on a witness’s demeanor. And in contrast to Ray , the district court here gave reasons why it found Kuenneke truthful: her experience and lack of motive to lie along with textual evidence in the affidavit.

To conclude , we affirm the district court's factual findings. Kuenneke mistakenly wrote down and attached the wrong statute. But she identified § 404(b) as the correct statute for the categorical approach and noticed the potential mismatches between that statute and the Guideline. She performed a divisibility analysis, researched the relevant issues, and employed the modified categorical approach.

This court held in Elion I that Kuenneke's performance prejudiced Elion. 76 F.4th at 635. Whether her performance was deficient turns on whether Kuenneke's 'representation fell below an objective standard of reasonableness.' Strickland , 466 U.S. at 688. Counsel’s performance need not be perfect to be objectively reasonable. “Criminal defendants have a right to a competent lawyer, but not to Clarence Darrow.' United States v. Rezin , 322 F.3d 443, 446-47 (7th Cir. 2003), overruled on other grounds by Lockhart v. United States , 577 U.S. 347 (2016). And courts must make 'every effort … to eliminate the distorting effects of hindsight .' Strickland , 466 U.S. at 689.

Page 11First, we discuss whether Kuenneke’s incorrect legal conclusion alone makes her performance deficient. Next, we consider whether our caselaw foreshadowed our conclusion in Elion I . We then evaluate whether Kuenneke should have at least made the categorical approach argument. Last, we analyze her performance as a whole.

The district court concluded that Kuenneke’s choice not to challenge the sentencing enhancement was not strategic. Rather, she opted not to object because she had reached an incorrect legal conclusion that § 404(b) was divisible (even though this court had not yet determined that § 404(b) was indivisible until Elion I ). She then chose not to advance the claim, as she deemed it not viable.

Elion tells us that our caselaw establishes that if an attorney fails to make a winning argument solely because she reached a legally incorrect conclusion, it is deficient performance. For support, Elion points to Cates v. United States , 882 F.3d 731, 736 (7th Cir. 2018), and Bridges , 991 F.3d at 800.

But that is emphatically not what our caselaw holds. We have repeatedly stated that counsel need not forecast changes in the law. 'Defense attorneys, it is true, are generally not obliged to anticipate changes in the law,' and 'it is not enough to show that a challenge to the career offender enhancement should have prevailed.” Bridges , 991 F.3d at 802, 804; Resnick v. United States , 7 F.4th 611, 623 (7th Cir. 2021) (“[O]ur case law provides that failure to object to an issue that is not settled law within the circuit is not unreasonable by defense counsel.”); see also Scott v. Hepp , 62 F.4th 343, 348 (7th Cir. 2023) (AEDPA case but discussing Strickland ’s principle Page 12that “a failure to argue a point of unsettled law, not foreshadowed by existing case law, ‘is not enough by itself to demonstrate deficient performance '' (quoting Minnick v. Winkleski , 15 F.4th 460, 470 (7th Cir. 2023))). As these cases make clear, a defense attorney’s choice not to make a potentially meritorious argument is not automatically deficient performance , even if it stems solely from a legal error.

Nor can Elion’s reading of our caselaw be reconciled with Strickland , which instructs that the 'proper measure of attorney performance remains simply reasonableness under prevailing professional norms.” 466 U.S. at 688. Strickland necessarily permits mistakes that are reasonable. Only when the defense attorney's error is so appalling that he can no longer be considered 'counsel' for his client is his performance deemed deficient . Harrington v. Richter , 562 U.S. 86, 104 (2011). Strickland does not demand clairvoyance. “[L]aw is no exact science … . The giving of legal advice that later is proven to be incorrect, therefore, does not necessarily fall below the objective standard of reasonableness.' Smith v. Singletary , 170 F.3d 1051, 1054 (11th Cir. 1999). We recognized this point when we said, “[c]riminal defendants have a right to a competent lawyer, but not to Clarence Darrow.' Rezin , 322 F.3d at 446–47. Elion’s proffered rule cannot be reconciled with these principles.

To be sure, we stated in Cates , 'a mistake of law is deficient performance.' 882 F.3d at 736. But that statement must be read in context. First, counsel there did not object to a jury instruction that 'plainly misstated' the law by 'flatly contradict[ing]” the statutory text. Id. at 737. So, as the government correctly points out, the error there was egregious. And the full quote from Cates shows that its holding concerns jury Page 13instructions: “There is no conceivable strategic reason for a defense lawyer to forgo a challenge to a prejudicial jury instruction.” Id. at 736. Since Cates , this court has understood that case’s holding to apply to jury instructions. See Powell v. Fuchs , 4 F.4th 541, 547-48 (7th Cir. 2021) ('When a jury instruction so badly misstates the law and that error relaxes the government’s burden of proof, this court held that there can be no conceivable strategic reason for counsel not to object.” (citing Cates , 882 F.3d at 736)). Second, Cates cited Vineyard v. United States , 804 F.3d 1218, 1225 (7th Cir. 2015), which stated that a 'misunderstanding of law or fact, however, can amount to ineffective assistance.' Id. (emphasis added). So, Vineyard does not support reading Cates to hold that a misunderstanding of law always amounts to ineffective assistance.

Elion also claims a “mistake of law” is when counsel misunderstands the law. That position lacks a limiting principle. If an attorney declines to make an argument that no court has accepted and no other attorney has made, yet which later succeeds, it is doubtful the attorney’s omission was unreasonable under “[p]revailing norms of practice.” Strickland , 466 U.S. at 688. Yet under Elion’s definition, the attorney is deficient.

In addition, under Elion’s rule, defense attorneys would have to raise all possible arguments, even wrong ones. Otherwise, many years later, counsel could be judged deficient for possibly misunderstanding the law. But “trial counsel may undermine the credibility of the defense of his client if he simply presents the court with a barrage of attacks.” Lickers v. United States , 98 F.4th 847, 857 (7th Cir. 2024) (quoting Goins v. Lane , 787 F.2d 248, 254 (7th Cir. 1986)). Elion’s rule therefore runs counter to the interests of criminal defendants.

Page 14An attorney’s mistake of law is not always deficient performance under Strickland . For that reason, Kuenneke’s decision not to raise a potentially meritorious argument-even though not done strategically—is not automatically deficient performance.

Although a single mistake of law is not always deficient performance, there is a narrow exception in the categorical approach context. This occurs when an attorney fail s to recognize that our caselaw foreshadows an argument that later succeeds. We outline the law on this exception, but we reject Elion’s arguments that he falls within it.

Strickland tells courts to assess attorney performance based on “prevailing professional norms.” 466 U.S. at 688. A single legal error rarely demonstrates deficient performance. Richter , 562 U.S. at 111. This includes failing to “anticipate changes in the law.” Bridges , 991 F.3d at 804. But a narrow exception exists for solitary errors that are 'sufficiently egregious and prejudicial.' Richter , 562 U.S. at 111 (quoting Murray v. Carrier , 477 U.S. 478, 496 (1986)). “An attorney 's ignorance of a point of law that is fundamental to his case combined with his failure to perform basic research on that point is a quintessential example of unreasonable performance under Strickland .' Hinton v. Alabama , 571 U.S. 263, 274 (2014).

Our caselaw has applied this exception to an attorney's performance of the categorical approach. For example, in Bridges , the defendant was indicted for Hobbs Act robbery and later pleaded guilty. 991 F.3d at 797. As part of his guilty plea, he agreed that his crime of conviction was a “crime of Page 15violence' under the Guideline's careeroffender enhancement. Id. His attorney failed to object to that label at sentencing. Id. at 797-98. He later sought postconviction relief, alleging his attorney was ineffective for failing to object to the career offender designation. Id . at 799. Had counsel done “minimal research,” he would have “uncovered a Tenth Circuit decision squarely holding that Hobbs Act robbery was no longer a crime of violence,” although the Seventh Circuit had yet to reach that holding. Id. at 797-98.

Though 'defense attorneys … are generally not obliged to anticipate changes in the law,' attorneys can still be deficient if the argument was 'sufficiently foreshadowed.' Id. at 804. Thus, even though this court had not reached such a holding, out-of-circuit caselaw provided a clear path for Bridges’s attorney. Id. at 805. Nor was such an argument obscure, as making Guidelines arguments is a “core competency” of federal defense counsel. Id. at 804 (citation modified). The case was remanded for an evidentiary hearing to determine whether 'defense counsel had a reasonable sentencing strategy.' Id. at 807.

Following Bridges was Harris v. United States , 13 F.4th 623, 625 (7th Cir. 2021). There, the defendant had several prior state drug convictions. Id. When he pleaded guilty to a federal drug offense, his attorney failed to argue whether those prior drug convictions counted as 'felony drug offense [s]” under a sentencing enhancement provision. Id. We concluded that counsel 'should have known about a possible categorical challenge.' Id. at 629. This was true even though the argument had not yet been addressed, as at the time of sentencing “at least one court had considered the issue.” Id. at 629-30. What is more, the categorical approach framework was well Page 16established, and the “analysis called for” was not complex— counsel simply “had to compare the plain language of statutes.” Id. at 630. But given counsel’s options at the time, his performance was not deficient. The ' defense essentially had a bird in the handthe plea offer with a set 20 -year sentence—with a possibility of two in the bush—the novel challenge to the predicate offenses with the risk of a mandatory life sentence.' Id. at 631. Thus, it was “objectively reasonable” for counsel not to raise the categorical challenge and instead accept the plea deal. Id.

Harris and Bridges each recognize this 'failure to foreshadow' exception. When 'case law sufficiently foreshadowed [an] argument ,” the attorney “could be deemed to have been deficient' for not advancing it. Bridges , 991 F.3d at 804. But the definition of 'foreshadow ing” has its limits. After all, an attorney is constitutionally ineffective only if his errors are “so serious” that he no longer functions as “counsel.” Strickland , 466 U.S. at 687. To 'eliminate the distorting effects of hindsight,” courts put themselves “in defense counsel’s shoes at the time of the challenged acts or omissions” to assess his performance. Lickers , 98 F.4th at 857 (quoting Strickland , 466 U.S. at 689). Existing decisions therefore must 'sufficiently' foreshadow the argument. Bridges , 991 F.3d at 804; Harris , 13 F.4th at 629. We hold that a 'sufficiently foreshadowed' argument is one that no reasonable attorney could fail to foresee. Kimbrough v. United States , 71 F.4th 468, 472 (6th Cir. 2023).

A reasonable attorney could, for example, be expected to raise a foreshadowed argument when courts in the jurisdiction expressly predicted the change. Chase v. MaCauley , 971 F.3d 582, 594 (6th Cir. 2020) (“[J]udges on the Michigan Court of Appeals panel … had clearly and forcefully detailed Page 17why Alleyne rendered Michigan's sentencing scheme unconstitutional.'). But persistent uncertainties in an area of law might mean it was not possible to foresee a particular change.

This exception for failing to raise a foreshadowed claim is narrow and rarely applies. This is because 'defense attorneys … are generally not obliged to anticipate changes in the law.” Bridges , 991 F.3d at 804; see also Harris , 13 F.4th at 629 (stating that the 'reasonableness of counsel's performance must be assessed in the context of the law at the time” (citation modified)). Infrequent application of this exception ensures that courts do not allow hindsight to color judgment of an attorney's performance but rather evaluate the performance using “prevailing professional norms.” Strickland , 466 U.S. at 689.

These principles in tow, we move to Elion’s case.

In 2017 this circuit had yet to decide the legal issue in Elion’s case. The question therefore is whether Elion I 's conclusion was 'sufficiently' foreshadowed by existing caselaw, such that Kuenneke's failure to advance the argument is deficient performance.

Elion first claims that Kuenneke did not need to recognize a foreshadowed argument. Instead, she should have concluded that “delivery” and “distribute” in § 404(b) are ambiguous, not synonymous. At the time, Taylor established that courts must be certain when imposing an enhanced sentence under the categorical approach. See Elion , 76 F.4th at 634–35 (stating that a “lack of clear signals” compelled a conclusion that the statute was indivisible and citing Mathis , 579 U.S. atPage 18 519). So, to Elion, Kuenneke should have at least raised an argument that Taylor ’s demand for certainty had not been met.

True, before Elion's sentencing Taylor did demand certainty. See Shepard v. United States , 544 U.S. 13, 21 (2005). But Strickland tells us to put “ourselves in defense counsel’s shoes at the time of the challenged acts or omissions.' Lickers , 98 F.4th at 857. Considering Kuenneke’s position at the time, Taylor ’s principle would have been of little use to her. For her to benefit from it, she needed concrete applications of that principle that were similar to Elion’s case. That is particularly true because 'certainty' is a standard, not a rule. Reasonable minds can disagree as to its application; what one finds certain, another finds ambiguous. Thus, Kuenneke is not deficient for failing to offer the categorical approach argument just because Taylor 's certainty principle was established in 2017.

This court in 2017 had yet to answer two key questions for Kuenneke's analysis. First up is whether 'look-alike' was synonymous with “counterfeit.” In United States v. Hudson , this court had held that Indiana’s statute using “look-alike” was a categorical match with “counterfeit” under the Guideline. 618 F.3d 700 (7th Cir. 2010). And in United States v. Robertson , the Eighth Circuit had held that Illinois's definition of “look-alike” was also a categorical match with “counterfeit” under the Guideline. 474 F.3d 538 (8th Cir. 2007). So, existing caselaw suggested Elion’s argument would not succeed.

Nor had this court answered a second question on the advertising mismatch. Kuenneke, however, researched the issue and located United States v. Hinkle , 832 F.3d at 569. But Hinkle was of little help because it was specific to a Texas statute, and the categorical approach can turn on a state's view of its law.

Page 19See, e.g. , United States v. Franklin , 895 F.3d 954, 958 (7th Cir. 2018) (per curiam) (differentiating between Wisconsin's and Iowa’s burglary statutes). In sum, existing caselaw either did not apply or strongly disfavored Elion’s categorical approach argument.

At issue is whether a reasonable attorney could have failed to foresee the conclusion in Elion I . As explained, a n attorney can be deficient for not raising a foreshadowed argument when obvious caselaw is missed. Harris , 13 F.4th at 629; Bridges , 991 F.3d at 805. But if the attorney does locate that obvious caselaw, yet that precedent strongly disfavors the client's argument, no reasonable attorney would foresee this court holding the opposite. That is what happened here. Robertson analyzed 'the same two terms now at issue' but reached a conclusion that cut against Elion. Elion I , 76 F.4th at 628. Hudson also cast doubt on whether Elion's argument would succeed. And no court had answered the advertising mismatch, except Hinkle , which turned on the particulars of Texas law and thus did not aid Elion’s case. For these reasons, when Elion was sentenced, a reasonable attorney would not foresee Elion I 's conclusion. See United States v. Carthorne , 878 F.3d 458, 466 (4th Cir. 2017) (“[C]ounsel may be constitutionally required to object when there is relevant authority strongly suggesting that a sentencing enhancement is not proper.'). So, Kuenneke did not perform deficiently by failing to raise that ultimately successful argument.

Contrast Kuenneke’s performance with that of the lawyers in Bridges and Harris —two cases falling under the “foreshadowing exception.” Start with Bridges . The chief difference is the depth and quality of the legal research. In Bridges , the attorney missed a 'noteworthy published opinion,' which Page 20answered the issue in his case. 991 F.3d at 805. “With modern methods of legal research, it would not have taken long in 2018 for counsel to have found the Tenth Circuit decision holding that Hobbs Act robbery is not a crime of violence under the 2016 guideline amendment.' Id. Nor was that the only case discussing the issue. “Counsel checking citations … would have also found the Fourth Circuit’s opinion in United States v. McCollum , 885 F.3d 300 (4th Cir. 2018), issued just one month before Bridges signed his plea agreement.” Id. at 806. Because caselaw offered a clear avenue of relief, and reasonable counsel would understand and recognize the categorical approach, counsel may have been deficient for not raising the argument.

Kuenneke, by contrast, did the necessary work. She researched the categorical approach arguments, found the applicable caselaw, and analogized that precedent to Elion's case. When the caselaw provided no answer, she used statutory interpretation and relied on her extensive experience. Kuenneke just reached a different conclusion than Elion I -a conclusion on which reasonable minds could disagree.

Kuenneke's performance was also considerably better than that of the attorney in Harris . There, the attorney “should have known about a possible categorical challenge.” 13 F.4th at 629. A circuit court had applied the categorical approach to the 'federal 'felony drug offense' definition ,' and the issue was an easy one to spot. Id. at 629–30. But here, Kuenneke recognized the categorical approach issue and thoroughly examined its viability.

Elion argues that focusing solely on the amount of legal research treats Strickland as a 'mere billable-hours requirement,” leading to absurd results: A n attorney who Page 21exhaustively researches an issue but misunderstands a core holding would not be deficient. We disagree. How thoroughly an attorney researches an issue can reflect the quality of that attorney's performance. Yet the evaluation entails many other factors, such as the difficulty of the legal question presented . An attorney's error is less justifiable when the categorical approach analysis is “not complex.” Harris , 13 F.4th at 630. So, the opposite is also true—when that analysis is complex, as here, an attorney's legal error is more reasonable.³ And as Bridges emphasized, Strickland ’s performance test also considers the strength of available alternative arguments. 991 F.3d at 807–08. Thus, our caselaw does not focus only on the attorney’s time spent researching the law.

Only in rare circumstances will precedent foreshadow a change in law. “The change must be so obvious that no reasonable jurist or attorney could fail to foresee it.” Kimbrough , 71 F.4th at 472 (citation modified). Elion's categorical approach argument was not foreshadowed because existing caselaw was silent or cut against it. As the district court here correctly recognized, Kuenneke’s performance is not analogous to that of the attorneys in Bridges and Harris , who should have foreseen a successful categorical approach argument.

Elion also points us to United States v. Coleman , 79 F.4th 822 (7th Cir. 2023), for support. There, defense counsel failed to challenge whether the defendant's prior convictions were convictions for a 'felony drug offense.' Id. at 826. This court agreed that a remand was necessary because it was “objectively unreasonable for Coleman's defense counsel to have not even considered a categorical challenge.” Id. at 832.

Coleman reached this conclusion via the foreshadowing exception. Id. at 831. The argument Coleman’s attorney failed to make would have been novel at the time. Id. No circuit court had addressed the question. Id. at 833 (Easterbrook, J., dissenting). And it would take this court six years after Coleman’s sentencing to do so. Id. (citing United States v. Ruth , 966 F.3d 642 (7th Cir. 2020)). Yet, this court in Coleman held that caselaw foreshadowed the argument the attorney failed to make because “the groundwork for such an argument” had been laid in other cases in different contexts. Id. at 831.

Coleman extended the holdings of Bridges and Harris . Indeed, Coleman 's holding can be read so broadly as to be “misaligned with our Strickland jurisprudence . ” Neal v. United States , No. 23-1722, 2025 WL 2553445, at *5 (7th Cir. 2025). To illustrate how, consider an example. For a time, the Confrontation Clause’s promise that the accused has a right to confront the witnesses against him could be overcome if the out-of-court statement had an “adequate indicia of reliability.” Ohio v. Roberts , 448 U.S. 56, 65–66 (1980) (citation modified) . The Court, relying largely on historical evidence, then abandoned that test in Crawford v. Washington to align closer with the Clause’s original meaning. 541 U.S. 36, 43–50 (2004). Taken to its logical limit, this change would mean that every Page 23federal defender with a Confrontation Clause argument before Crawford performed deficiently if by failing to argue that Ohio ’s “adequate indicia” test was wrong as an original matter. One could contend that the “groundwork for such an argument” was already laid because the historical sources were accessible, and the focus on text, history, and tradition had been emphasized in other “contexts.” Cf. Coleman , 79 F.4th at 832 ('[T]he groundwork for such an argument was, at the very least, foreshadowed by numerous decisions … in other contexts.”). That is difficult to reconcile with Strickland 's instructions that the “proper measure of attorney performance” is 'reasonableness under prevailing professional norms.' Strickland , 466 U.S. at 688 (emphasis added).

Rather, the application of Strickland in Coleman must be read in light of cases such as Hinton and Richter . Those cases establish that only egregious errors amount to deficient performance. Hinton , 571 U.S. at 274 (' An attorney ’s ignorance of a point of law that is fundamental to his case combined with his failure to perform basic research on that point is a quintessential example of unreasonable performance under Strickland .”); Richter , 562 U.S. at 111 (' And while in some instances even an isolated error can support an ineffective -assistance claim if it is sufficiently egregious and prejudicial , it is difficult to establish ineffective assistance when counsel ’s overall performance indicates active and capable advocacy.” (citation modified)). Strickland emphasizes that a “fair assessment of attorney performance requires that every effort be made to eliminate the distorting effects of hindsight … and to evaluate the conduct from counsel’s perspective at the time.” 466 U.S. at 689. Coleman suggests c ounsel is deficient any time some amount of “groundwork” has been previously laid and that attorney fails to make an unobvious but ultimately successful Page 24argument. Instead, the test is whether the 'groundwork' sufficiently foreshadows the argument so that no reasonable attorney would fail to make it. Kimbrough , 71 F.4th at 472. Coleman sits at the outer boundary of that test.

This case is not analogous to Coleman . There, t he attorney failed even to consider the categorical approach argument. 79 F.4th at 832. By contrast, Kuenneke considered, researched, and analyzed that argument. Coleman emphasized that the argument the attorney failed to make was not complex. Id. Not so here. Kuenneke’s analysis turned on nuanced difference s between the definitions of “counterfeit” and “look alike,” or whether “distribute” meant “delivery,” questions we called “complex” and “ambiguous” and that we could not “conclusively determine.' Elion I , 76 F.4th at 628, 634. So, we conclude that Elion’s case does not fall into the foreshadowing exception.

But wait, Elion argues, Kuenneke should have at least tried to make a categorical approach argument. After all, she testified that when she began researching Elion’s case, there was potentially an argument pursuant to Mathis about the categorical approach.

For support, Elion relies on Polk County v. Dodson , 454 U.S. 312 (1981). There, the Court stated, “a defense attorney has a duty to advance all colorable claims and defenses.” Id. at 323. But Polk County considered whether public defenders act “under color of state law.' Id. at 314. Polk 's statement is not a rule changing the effectiveness standard under the Sixth Am endment.

Page 25Kuenneke did agree “there may have been an argument pursuant to Mathis .” But that is true for all sorts of challenges; that does not mean she must raise every one. Counsel has a duty to raise only sufficiently foreshadowed arguments, not viable ones. See Bridges , 991 F.3d at 802 (“[I]t is not enough to show that a challenge to the career offender enhancement should have prevailed.”); Kimbrough , 71 F.4th at 473.

There is a good reason for this rule. Defense attorneys often have several potential arguments to advance but only limited resources. Time and effort spent on one argument takes away from another. So, it is natural for defense attorneys to shy away from novel, nuanced, or counterintuitive contentions and stick to previously productive arguments. Kuenneke chose not to offer difficult, counterintuitive reasoning—whether “counterfeit” is synonymous with “look-alike” and whether “distribute” is synonymous with “delivery.” Instead, she made mitigating arguments that she had investigated, which she felt had a higher chance of success. In the end, the court gave Elion a shorter sentence than what the government requested.

As well, “there is a tactical reason not to make weak arguments” because “they may distract the court from the strong arguments and as a result make it less likely to rule in the defendant’s favor.” Rezin , 322 F.3d at 446. Here, after research, Kuenneke judged Elion’s categorical approach argument to not be 'a viable objection.' Nor is it an answer to say Kuenneke should have made the categorical approach argument in addition to the mitigating arguments. She “cannot be faulted for eschewing the proverbial kitchen sink and instead focusing on arguments with better odds. ” Lickers , 98 F.4th at 857. '[T]rial counsel may undermine the credibility of the Page 26defense of his client if he simply presents the court with a barrage of attacks.” Id. (quoting Goins v. Lane , 787 F.2d 248, 254 (7th Cir. 1986)).

In short, it is not whether counsel could have made an argument, but whether counsel unprofessionally erred by not making an argument. Kuenneke 's performance was not deficient solely because she did not offer a potentially valid categorical approach argument. Cf. Williams v. Lemmon , 557 F.3d 534, 538 (7th Cir. 2009) (per curiam) (An egregious error is “an omission of something obviously better ( in light of what was known at the time) than the line of defense that counsel pursued.”).

Last, the government correctly points out that Kuenneke’s performance must be “assessed as a whole; it is the overall deficient performance, rather than a specific failing, that constitutes the ground for relief.' Peoples v. United States , 403 F.3d 844, 848 (7th Cir. 2005). “While in some instances even an isolated error can support an ineffective -assistance claim, it is difficult to establish ineffective assistance when counsel ’s overall performance indicates active and capable advocacy.” Richter , 562 U.S. at 111 (citation modified).

Elion claims Kuenneke’s performance was worse than that of the attorney in United States v. Winstead , 890 F.3d 1082 (D.C. Cir. 2018). There, the defendant's attorney failed to make a novel (for that circuit) Guideline argument that fi ve circuits had already rejected. Id. at 1089. The court concluded that counsel was deficient for not making the argument because “there was no conceivable tactical reason … for not making it.” Id. at 1090. That was not the only error made by the Page 27attorney , however. The court was troubled that “Winstead’s counsel did not put up much opposition at the sentencing hearing' when he said, '[y]our honor … [m]y client is resigned to his fate … . I can only be candid with this court.' Id. at 1087-88. By conceding, rather than making an 'obvious [Guideline] legal argument”—“the only serious argument the defendant had in the entire case”— counsel was deficient. Id. at 1090. So, counsel, judged as a whole, made several mistakes including an egregious concession.

Kuenneke, on the other hand, investigated and offered viable mitigation arguments to advocate for a shorter sentence. She did so after identifying and researching the relevant categorical approach questions. Kuenneke mistakenly reached a different conclusion than Elion I , an error mitigated by the nuance and complexity of the analysis. Kuenneke’s performance as a whole therefore was not deficient.

Kuenneke spotted the issues in Elion's case, researched them, and correctly applied the categorical approach. She decided not to offer a categorical approach argument because she believed the state statute of conviction was divisible. We later held that statute was indivisible. Kuenneke’s conclusion, though incorrect and prejudicial to Elion, was not deficient performance because caselaw did not foreshadow our holding. So, Kuenneke’s representation of Elion was not ineffective assistance of counsel.

AFFIRMED.