Before WILLETT, HO, and DUNCAN, Circuit Judges. PER CURIAM: A jury convicted Garland Bernell Harper of murder and sentenced him to death. After his direct appeal and habeas petitions were both denied in state court, Harper raised 31 claims in a federal habeas petition. The district court denied all his claims and also denied a certificate of appealability (COA). Harper asks us to issue a COA on eight of those claims which he presents as posing five distinct legal issues. We DENY Harper a COA on all of his claims for the reasons explained below. I Harper was convicted of murdering his girlfriend, Triska Rose, and her two daughters: Mya, aged seven, and Briana, aged sixteen. The jury sentenced Harper to death in a separate punishment phase. Harper filed a direct appeal raising eight claims. The Texas Court of Criminal Appeals (TCCA) found no error and affirmed Harper’s conviction. Harper later filed an application for a writ of habeas corpus. The trial court drafted proposed findings of fact and conclusions of law, recommending that Harper’s application be denied. The TCCA adopted these findings of fact and conclusions of law with a few minor adjustments and denied Harper’s application. Harper then filed a petition for a writ of habeas corpus in federal court, which was 291 pages long and contained thirty-one claims. The district court denied each of Harper’s claims in a sixteen-page opinion and did not certify any issue for review on appeal. Harper asks us to issue a certificate of appealability on eight of the thirty-one claims. Some claims overlap and, as a result, Harper presents them as five issues: (1) a Confrontation Clause claim (claim 2 of Harper’s habeas petition); (2) a Strickland claim premised on ineffective assistance of counsel during voir dire (claim 11); (3) a Batson claim (claims 12, 13, and 28); (4) a second Strickland claim premised on counsel’s failure to argue that Harper’s mental illness rendered his confession involuntary (claims 15 and 16); and (5) a third Strickland claim premised on counsel’s failure to object on reliability grounds to the government’s expert on future dangerousness, Dr. Moeller (claim 7). II We may issue a certificate of appealability “only if the applicant has made a substantial showing of the denial of a constitutional right.” 28 U.S.C. § 2253(c)(2). The Supreme Court has clarified that this “substantial showing” requires demonstrating that “reasonable jurists could debate whether . . . the petition should have been resolved in a different manner.” Miller-El v. Cockrell, 537 U.S. 322, 336 (2003) (quoting Slack v. McDaniel, 529 U.S. 473, 484 (2000)). In a capital case, “any doubts as to whether a COA should issue must be resolved in the petitioner’s favor.” Nelson v. Davis, 952 F.3d 651, 658 (5th Cir. 2020) (quoting Clark v. Thaler, 673 F.3d 410, 425 (5th Cir. 2012)). Where, as here, “a state court has reviewed a petitioner’s claim on the merits, our review is constrained by the deferential standards of review found in the Antiterrorism and Effective Death Penalty Act (‘AEDPA’).” Id. (citing 28 U.S.C. § 2254). “Under these circumstances, we may not issue a COA unless reasonable jurists could debate that the state court’s decision was either ‘contrary to, or involved an unreasonable application of, clearly established Federal law,’ or ‘was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.’” Id. (internal citations omitted) (quoting 28 U.S.C. § 2254(d)(1)–(2)). A Harper’s first claim is that the district court erred by failing to consider his Confrontation Clause claim and, in the alternative, that his trial and appellate counsel were ineffective for failing to raise it. This issue received only a single footnote in Harper’s state and federal habeas petitions. So it isn’t surprising that the habeas court and the district court below did not discuss it. After the district court dismissed his habeas petition, Harper filed a Rule 59(e) motion to alter or amend the judgment, arguing that the district court had improperly overlooked his claim. The district court denied this motion, finding that the argument had not been fairly placed before the court. We agree with the district court that Harper did not sufficiently plead his Confrontation Clause claim. A conclusory footnote in a 291-page federal habeas petition is not enough to put a district court on notice of a claim. Habeas petitions must “specify all the grounds for relief available to the petitioner” and “state the facts supporting each ground.” Mayle v. Felix, 545 U.S. 644, 655 (2005) (quoting RULES GOVERNING § 2254 CASES 2(c)). “[C]onclusory allegations of ineffective assistance of counsel do not raise a constitutional issue in a federal habeas proceeding.” Miller v. Johnson, 200 F.3d 274, 282 (5th Cir. 2000) (citing Ross v. Estelle, 694 F.2d 1008, 1012 (5th Cir. 1983)). And even if a footnote were enough to raise the issue under the federal rules, it was not enough to comply with the exhaustion requirement. “The exhaustion requirement is satisfied when the substance of the federal habeas claim has been fairly presented to the highest state court.” Adekeye v. Davis, 938 F.3d 678, 682 (5th Cir. 2019) (quoting Soffar v. Dretke, 368 F.3d 441, 465 (5th Cir. 2004)). A fair opportunity requires that “all the facts necessary to support the federal claim were before the state courts” and “the habeas petitioner must have ‘fairly presented’ to the state courts the ‘substance’ of his federal habeas corpus claim.” Anderson v. Harless, 459 U.S. 4, 6 (1982) (quoting Picard v. Connor, 404 U.S. 270, 275, 277–78 (1971)). “Arguments raised in a perfunctory manner, such as in a footnote, are waived” because they do not give the state court a fair opportunity to consider the claim. Bridas S.A.P.I.C. v. Gov’t of Turkmenistan, 345 F.3d 347, 356 (5th Cir. 2003) (quoting United States v. Hardman, 297 F.3d 1116, 1131 (10th Cir. 2002)). Harper’s purported Confrontation Clause claim was just as buried in his state habeas petition. (Indeed, the two petitions are almost identical.) It was placed in a single footnote deep in the body of a 232-page petition that presented 24 claims. Unsurprisingly, neither the Texas District Court nor the Texas Court of Criminal Appeals treated this as a separate claim. The state and federal courts were not the only ones that overlooked Harper’s Confrontation Clause footnote. Harper overlooked it too. His state and federal habeas petitions failed to list it as a separate claim. And Harper did not ask the TCCA to reconsider its decision in light of his Confrontation Clause claim after the court issued its opinion without discussing footnote 12. The fact that no party—not even Harper’s own lawyer—understood him to have raised a Confrontation Clause claim is evidence enough that this issue was neither fairly presented to the state court nor adequately pleaded before the federal district court. The district court did not abuse its discretion by refusing to reconsider its decision in light of this unexhausted and forfeited claim. Neither of Harper’s alternative arguments have any merit either. He first argues that the district court should have sua sponte construed his reply brief, which did raise a Confrontation Clause argument, as a Rule 15 motion to amend his habeas petition. We need not decide this argument. Even if true, Harper’s claim would still be unexhausted for failing to present the claim to the state habeas court. Second, Harper argues that it doesn’t matter that he put his Confrontation Clause claim in a footnote because that footnote cited a transcript in which his trial counsel made the essence of a Confrontation Clause objection. Harper bases this argument on Dye v. Hofbauer and its progeny. They hold that a claim properly presented in an earlier proceeding can be incorporated by reference into a federal habeas petition. Ramey v. Davis, 942 F.3d 241, 248 (5th Cir. 2019) (citing Dye v. Hofbauer, 546 U.S. 1, 4 (2005) (per curiam)). But Dye and Ramey involved claims that were properly raised in prior briefs—not exhibits. See id. (claim properly raised in direct appeal brief and state habeas petition, but insufficiently briefed in federal habeas petition, was incorporated by reference); Dye, 546 U.S. at 4 (claim properly raised in a state habeas brief was incorporated by reference in federal habeas petition). Requiring courts to search through every exhibit cited in a habeas petition to discover additional possible arguments would be a massive and unwarranted extension of Dye. Dye didn’t turn habeas petitions into matryoshka dolls. Claims cannot be hidden inside of voluminous exhibits cited in footnotes hidden inside of habeas petitions that are hundreds of pages long. See RULES GOVERNING § 2254 CASES 2 advisory committee note (noting that Rule 2 does not require judges to grope through “two thousand pages of irrational, prolix and redundant pleadings” (quoting Passic v. Michigan, 98 F. Supp. 1015, 1016 (E.D. Mich. 1951)); Adams v. Armontrout, 897 F.2d 332, 333 (8th Cir. 1990) (“[D]espite our firm conviction that the pleading requirements in habeas corpus proceedings should not be overly technical and stringent, it would be unwise to saddle district judges with the burden of reading through voluminous records and transcripts in every case.” (quoting Williams v. Kullman, 722 F.2d 1048, 1051 (2d Cir. 1983)). No COA will issue on this claim. B Harper’s second claim is that his counsel’s performance was ineffective for failing to strike or challenge for cause three jurors—Dowlin, Basey-Higgs, and Williams. To succeed, Harper must show “(1) that his counsel’s performance was deficient, and (2) that the deficient performance prejudiced his defense.” Hughes v. Dretke, 412 F.3d 582, 589 (5th Cir. 2005) (citing Strickland v. Washington, 466 U.S. 668, 689–94 (1984)). “Regarding the first prong, ‘[t]o establish deficient performance, a petitioner must demonstrate that counsel’s representation “fell below an objective standard of reasonableness.”‘” Id. (alteration in original) (quoting Wiggins v. Smith, 539 U.S. 510, 521 (2003)). “Regarding the second prong, ‘to establish prejudice, a “defendant must show that there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different. A reasonable probability is a probability sufficient to undermine confidence in the outcome.”‘” Id. (quoting Wiggins, 539 U.S. at 534). Harper suggests that our decision is governed by two lines of Supreme Court cases that establish (1) that the death penalty cannot be mandatory, see Tuilaepa v. California, 512 U.S. 967, 972 (1994); Sumner v. Shuman, 483 U.S. 66, 77 (1987); Roberts v. Louisiana, 428 U.S. 325, 333 (1976) (plurality); Woodson v. North Carolina, 428 U.S. 280, 301 (1976) (plurality), and (2) that jurors must be willing and able to give effect to mitigating evidence when deciding whether to approve the death sentence, Morgan v. Illinois, 504 U.S. 719, 733–35 (1992); Penry v. Lynaugh, 492 U.S. 302, 327–28 (1989); Eddings v. Oklahoma, 455 U.S. 104, 115 (1982). But the state court found that these three jurors were all willing to consider mitigating evidence. The court reasoned that “none of the cited jurors indicated that they would automatically vote for the death penalty in every case and all expressed the opinion that they could answer the special issues in such a way that either life or death would result based on the evidence and the law.” Based on this evidence, the state court concluded that Harper’s counsel was not ineffective for failing to use a peremptory strike against these jurors. While Harper may disagree with this conclusion, the state court’s decision was supported by the evidence and its legal conclusion that Harper did not satisfy Strickland‘s first prong was therefore not unreasonable. And even if Harper could meet Strickland‘s first prong, he cannot meet the second. He does not even argue that the outcome would have been different if his counsel had objected to these three jurors. No COA will issue on this claim. C Harper’s third claim is that the prosecutor discriminated against prospective black jurors (Harper’s “Batson” claim). The Government argues that Harper’s Batson claim has evolved considerably since it was first presented on direct appeal, and as a result most of it is procedurally defaulted. We first consider whether any of Harper’s Batson arguments are procedurally defaulted, before evaluating the merits of his exhausted arguments. 1 Federal habeas review of a claim is procedurally barred if the highest available state court “dismissed the claim on a state-law procedural ground instead of deciding it on the merits.” Rocha v. Thaler, 626 F.3d 815, 820 (5th Cir. 2010) (citing Harris v. Reed, 489 U.S. 255, 262 (1989)). But to qualify, that state-law procedural ground must be both an “independent and adequate ground for dismissal.” Id. (quoting Nobles v. Johnson, 127 F.3d 409, 420 (5th Cir. 1997)). Independent means “independent of the merits of the federal claim.” Id. at 821 (quoting Finley v. Johnson, 243 F.3d 215, 218 (5th Cir. 2001)). Adequate means that the rule is “strictly or regularly applied evenhandedly to the vast majority of similar claims.” Id. (quoting Amos v. Scott, 61 F.3d 333, 339 (5th Cir. 1995)). The TCCA dismissed Harper’s Batson claim in part under its procedural rule that arguments that could have been raised on direct appeal but were not are procedurally defaulted. See Ex parte Nelson, 137 S.W.3d 666, 667 (Tex. Crim. App. 2004) (en banc) (“It is well-settled ‘that the writ of habeas corpus should not be used to litigate matters which should have been raised on direct appeal.’” (quoting Ex parte Gardner, 959 S.W.2d 189, 199 (Tex. Crim. App. 1996) (en banc))). We have previously recognized this rule as an adequate state ground that bars federal habeas relief, Aguilar v. Dretke, 428 F.3d 526, 533 (5th Cir. 2005) (citing Ex parte Gardner, 959 S.W. 2d at 199), and Harper does not argue that we should reconsider that decision. So the first question we must answer is which of Harper’s Batson arguments— if any—are procedurally defaulted.[1] Harper’s arguments in support of his Batson claim fall into two buckets. The first is that the prosecution’s strikes followed a pattern that gave rise to an inference of discrimination. The second is that each of the prosecution’s five stated reasons for using a peremptory strike against juror Banks were pretextual. We start with Harper’s “pattern of strikes” argument. Harper did make a version of this argument on direct appeal. But he didn’t present any evidence of the racial makeup of the jury pool as a whole. The TCCA held that the juror questionnaires, which would have provided evidence of the racial makeup of the jury pool, were essential to substantiate Harper’s claim that there was a racial disparity between the venire and the seated jury. In other words, by failing to introduce the juror questionnaires, Harper was effectively asking the TCCA to take his word for it that the seated jurors were more likely to be white than the venire was. Because evidence of the racial makeup of the accepted members was not in the record, Harper could not substantiate this claim. As a result the TCCA dismissed his claim on direct appeal. When Harper tried to raise this argument in his state habeas litigation, this time with the evidence he needed, the TCCA held that Texas law barred him from relying on arguments or evidence that he could have but failed to raise on direct appeal. Harper does not argue that the TCCA misapplied Texas law on this point. Therefore, while Harper has properly exhausted a pattern-of-strikes claim, he cannot rely on evidence that he did not present on direct appeal.[2] Next we consider Harper’s argument that the state’s five proffered reasons for striking juror Banks were pretextual. Again, we find that some of these arguments are fully preserved, some have been expanded to rely on additional theories or evidence, and others are entirely procedurally defaulted. Harper did procedurally exhaust his objections to the state’s first and second proffered reasons. But he only partially exhausted his objection to the state’s third proffered reason. The prosecutor’s third proffered reason was that Banks had said that “everybody is capable of rehabilitation.” Harper did argue in the direct appeal that this reason was pretextual because jurors Cotton and Basey also expressed that the chance for rehabilitation was the most important factor to them and they were seated anyways. But in his habeas petition, Harper relied on a comparison to five additional jurors— jurors Price, Moore, Pavlovich, Summer, and Vaughan—who, like Cotton, Basey, and Banks, expressed that rehabilitation is the most important goal of criminal punishment. Because Harper did not rely on a comparison to these additional jurors on direct appeal, he cannot do so now. Harper next argues that the State’s fourth proffered reason—that Banks failed to answer the question about whether life in prison is more effective than the death penalty—was pretextual. But this argument too was almost entirely procedurally defaulted. In his direct appeal, Harper’s only objection to this proffered reason was that there was no evidence that Banks was being “deceptive, or untruthful, in failing to answer that question.” In his two habeas petitions, Harper radically reshaped his argument, relying on evidence not presented in his direct appeal. He now argues that the prosecution failed to strike three other jurors who also failed to answer parts of the questionnaire. Fair point. But again, Harper cannot raise it now because this argument was not presented on direct appeal. Finally, Harper argues that the State’s fifth proffered reason— Banks’s strong belief in the importance of forgiveness based in part on her background in ministry—was pretextual. But Harper did not make any version of this argument on direct appeal. This argument is entirely procedurally defaulted. 2 Having sorted through which arguments in support of Harper’s Batson claim were procedurally defaulted and which were not, we consider whether the district court’s denial of the non-defaulted Batson arguments is debatable. We conclude that it was not. Where, as here, the defendant has made out a prima facie case that race motivated the challenged strikes, and the prosecutor has provided a race-neutral explanation for the strike, the trial court must weigh the evidence and decide whether the prosecutor’s explanation is credible or mere pretext for discrimination. Batson v. Kentucky, 476 U.S. 79, 96–98, 98 n.21 (1986). Because this question “largely will turn on evaluation of credibility,” id. at 98 n.21, and the best evidence of credibility is “the demeanor of the attorney who exercises the challenge,” Hernandez v. New York, 500 U.S. 352, 365 (1991) (plurality opinion), the trial court’s decision “is entitled to ‘great deference’ and ‘must be sustained unless it is clearly erroneous,’” Felkner v. Jackson, 562 U.S. 594, 598 (2011) (citation omitted) (quoting Batson, 476 U.S. at 98 n.21 and Snyder v. Louisiana, 552 U.S. 472, 477 (2008)). “[T]he federal court’s role is to ‘determine whether the trial court’s determination of the prosecutor’s neutrality with respect to race was objectively unreasonable and has been rebutted by clear and convincing evidence to the contrary.’” Hoffman v. Cain, 752 F.3d 430, 448–49 (5th Cir. 2014) (quoting Murphy v. Dretke, 416 F.3d 427, 432 (5th Cir. 2005)). We start with Harper’s pattern of strikes argument. For statistical evidence like this to be relevant, “data concerning the entire jury pool is necessary. The number of strikes used to excuse minority . . . jury pool members is irrelevant on its own.” Medellin v. Dretke, 371 F.3d 270, 278–79 (5th Cir. 2004) (per curiam). As noted above, Harper is barred from using evidence that he did not present on direct appeal.[3] Because he cannot rely on the juror questionnaires, and the questionnaires were necessary to prove Harper’s statistical claim, the district court’s dismissal of this argument is not debatable. Next, we consider Harper’s argument that each of the prosecutor’s five stated reasons for striking Banks were pretextual. The trial court did not clearly err by crediting the prosecutor’s first reason for striking Banks, which was that Banks did not answer questions directly and tended to “ponder,” out loud, and at length. Banks was indeed loquacious and noncommittal. At first, she seemed to say that she was categorically opposed to the death penalty, saying: “I mean, I’m pretty settled—I feel like I’m pretty settled on my thoughts concerning the death penalty. I don’t like to see people die. [Prosecutor:] Right. [Banks:] Period. I mean, who does?” She then indicated that the death penalty might be appropriate for serial killers. When the prosecutor later asked if the death penalty should only be used for serial killers, she responded that “[t]hose are not the only cases” and indicated that the murderer’s remorse or chance for rehabilitation were the most important factors to her. Three pages of the transcript are then taken up by Banks’s extended discussion of the facts of an unrelated DUI homicide case, and why she felt that 30 years imprisonment was justified because the crash took the lives of five children. Even after dozens of pages of the transcript were taken up discussing the death penalty, Banks still said that she had made only an “initial response,” and that if she had another “30 minutes to ponder on the question” she might change her mind on when the death penalty is appropriate. To be sure, the prosecutor’s questions likely played a role in Banks’s meandering and noncommittal responses. But in light of these facts, we cannot conclude that the trial court clearly erred in crediting the prosecutor’s explanation that she was striking Banks because she would not answer the question of her position on the death penalty clearly. Nor did the trial court clearly err in crediting the prosecutor’s second and third reasons: (2) that Banks was opposed to the death penalty, and (3) that she believed strongly in giving people a chance to rehabilitate themselves in prison. Both of these justifications had substantial basis in the record. True, Banks later walked back her statement that she was “pretty settled” in her opposition to the death penalty. But even then, she continually insisted on the importance of rehabilitation and tied her distaste for the death penalty to her religious beliefs. There is nothing wrong with expressing a distaste for the death penalty. Nor is it racially discriminatory for a prosecutor to use a peremptory strike because a juror expresses such distaste. We agree with the district court that the trial court did not clearly err in crediting the prosecutor’s second and third stated reasons. Harper fares no better with his argument that the prosecutor’s fourth proffered reason was pretextual. In his direct appeal, Harper argued that Banks was not being untruthful or deceptive by failing to respond to one of the items on the questionnaire. But this argument does nothing to demonstrate that the prosecutor’s stated reason was pretextual. Both things can be true: Banks could have been fully truthful and forthcoming, and the prosecutor could have been concerned that she failed to respond to one of the most important items on the questionnaire. Therefore, the TCCA did not clearly err in crediting the prosecutor’s fourth stated reason. Finally, we note that even if Harper had evidence tending to disprove some of the prosecutor’s proffered reasons, that is not enough. “[A] Batson claim will not succeed where the defendant fails to rebut each of the prosecutor’s legitimate reasons.” Sheppard, 967 F.3d at 472 (citing Fields v. Thaler, 588 F.3d 270, 277 (5th Cir. 2009) and Stevens v. Epps, 618 F.3d 489, 500 (5th Cir. 2010)). Harper failed to rebut the prosecution’s fifth stated reason for striking Banks at all. Therefore, the district court’s rejection of Harper’s Batson claim is not debatable. 3 Harper next argues that his counsel on direct appeal rendered ineffective assistance by making an incomplete Batson claim. He faults his counsel on direct appeal for conducting only a limited comparative juror analysis, failing to rebut the prosecutor’s fifth stated reason for striking Banks at all, and failing to include the questionnaires of the jurors into the record. An ineffective assistance of appellate counsel claim “requires a showing that (1) counsel’s performance was legally deficient, and (2) the deficiency prejudiced the defense.” United States v. Bernard, 762 F.3d 467, 471 (5th Cir. 2014) (citing Strickland, 466 U.S. at 687 (1984)). “Applying AEDPA deference to Strickland‘s already deferential standard, we must deny relief if ‘there is any reasonable argument that [appellate] counsel satisfied Strickland‘s deferential standard’ despite failing to make the argument [in question]. In other words, we must deny relief ‘if there was a reasonable justification for the state court’s decision.’” Higgins v. Cain, 720 F.3d 255, 265 (5th Cir. 2013) (first alteration in original) (quoting Harrington v. Richter, 562 U.S. 86, 105, 109 (2011)). The habeas court conducted an extensive argument-by-argument review of Harper’s comparative juror analysis argument. It considered each argument that Harper said should have been raised. It found that each of these arguments was meritless, and that as a result, Harper’s appellate counsel was not ineffective for failing to raise them. But Harper does not take issue with what the habeas court concluded. Instead, he takes issue with how the habeas court reached that conclusion. He argues that the habeas court improperly relied on the prosecutor’s subsequent affidavit which explained some of the inconsistencies in her proffered reasons for striking Banks. For example, the prosecutor averred that her inaccurate statement that Banks wanted to do away with the death penalty was an “honest mistake based on the prosecutor’s impression after Banks repeatedly emphasized her belief in forgiveness and rehabilitation.” The habeas court also relied on the prosecutor’s testimony that while other jurors mentioned rehabilitation, the prosecutor only struck Banks because those jurors “did not reach the intensity of Banks’ belief in rehabilitation and forgiveness.” Harper argues that by relying on this “post hoc” reasoning, the habeas court violated Miller-El II‘s “stand or fall” rule. Harper’s argument that Miller-El II prevents the prosecution from ever introducing additional evidence to resist a Batson claim is wrong. Miller- El II‘s “stand or fall” rule means that prosecutors and later reviewing courts cannot accept “either entirely different substituted reasons or post hoc reasons for strikes.” Chamberlin v. Fisher, 885 F.3d 832, 841 (5th Cir. 2018) (en banc) (discussing Miller-El v. Dretke (Miller-El II), 545 U.S. 231, 251–52 (2005)). But Miller-El II “does not extend to preventing the prosecution from later supporting its originally proffered reasons with additional record evidence, especially if a defendant is allowed to raise objections to juror selection years after a conviction and to allege newly discovered comparisons to other prospective jurors.” Id. “Nothing in the ‘stand or fall’ statement means that the prosecutor would forfeit the opportunity to respond to such contentions.” Id. The prosecutor’s affidavit falls squarely within the type of evidence that Chamberlin said later reviewing courts may consider: a prosecutor’s “opportunity to respond” to “newly discovered comparisons to other prospective jurors.” Id. This makes sense. If the prosecution was not able to explain why it did not strike certain jurors after the fact, it would have to foresee future Batson claims and explain why it was not striking each prospective juror during jury selection. See Chamberlin v. Fisher, 855 F.3d 657, 674 (5th Cir. 2017) (Clement, J., dissenting) (“[T]o avoid the result reached by the majority here, during jury selection the prosecution would not only have had to explain why it struck specific black jurors—as it did—but also why it did not strike all white prospective jurors as well. There is nothing in Batson, Miller-El II, or any other case that compels anything of the sort.”), rev’d, 885 F.3d 832 (5th Cir. 2018) (en banc) (Judge Clement’s position was later adopted by the en banc court). Miller-El II requires consistency, not prophecy. No COA will issue on this claim. D Harper next argues that his trial counsel was ineffective for failing to object to the introduction of Dr. Moeller’s testimony on the basis that it was unreliable under the standard set forth in Kelly v. State, 824 S.W.2d 568, 573 (Tex. Crim. App. 1992) (en banc).[4] The Supreme Court has acknowledged that testimony from psychologists on likelihood of future dangerousness is rather shaky in general because studies have shown that such testimony is wrong more often than it is right. See Barefoot v. Estelle, 463 U.S. 880, 901 (1983) (considering a report from the American Psychiatric Association (APA) stating that psychiatric opinions regarding future dangerousness are wrong “most of the time”); see also Coble v. State, 330 S.W.3d 253, 275 n.53 (Tex. Crim. App. 2010) (noting that “[m]ore recent psychiatric and legal articles have reached a similar conclusion [to the APA's amicus brief in Barefoot], although some conclude that the accuracy of clinicians’ predictions may now be slightly better than chance when they also use risk assessment and actuarial tools”). Nevertheless, the Supreme Court in Barefoot, 463 U.S. at 901, and the Texas Court of Criminal Appeals in Coble, 330 S.W.3d at 275– 77, both concluded that expert testimony on future dangerousness may be admissible in some cases if it is reliable. Even if we assumed that Harper’s counsel rendered ineffective assistance, Harper cannot meet Strickland‘s second prong by showing prejudice, for two reasons: (1) the record shows that the court would have denied a Kelly objection, and (2) there was ample evidence of future dangerousness in the record apart from Dr. Moeller’s testimony. First, a Kelly objection would have been futile. While Harper’s counsel did not specifically make a Kelly objection, he did “object to [Dr. Moeller's] testimony as having any value if it’s from a lay perspective.” Harper’s counsel also told the court, as a part of his argument that Dr. Moeller should not be able to testify, that the prosecution was “not calling Dr. Moeller to explain his report . . . . [t]hey’re calling Dr. Moeller to disavow that report.” The court overruled those objections. It is exceedingly unlikely that the court would have changed its mind had Harper’s counsel merely uttered the words “Kelly” or “ Daubert.” See Clark v. Collins, 19 F.3d 959, 966 (5th Cir. 1994) (no ineffective assistance of counsel for failing to make an objection that would have been overruled). Second, there was ample evidence supporting the jury’s finding on future dangerousness even without Dr. Moeller’s testimony. First and foremost, there was the evidence of this crime: Harper brutally murdered four people, including two young girls. The jury also heard evidence that he raped and killed another woman back in 1989, put a knife to a woman’s neck and robbed her, forced a different woman into his car and assaulted her, assaulted a cab driver and refused to pay his fare, stole a purse from a former co-worker and withdrew $800 from her account before being arrested, stole a woman’s purse and pushed her pregnant sister into a shopping cart before stealing a car and fleeing the scene, and took another woman’s purse, knocked her down, and again fled in a car. In light of this substantial (indeed, overwhelming) evidence that Harper was dangerous, Dr. Moeller’s testimony did not cause prejudice. See Coble, 330 S.W.3d at 281 (erroneously admitted testimony from psychologist on likelihood of future dangerousness did not cause prejudice because there was “ample evidence” of such dangerousness “quite apart from [the psychologist's] testimony”); see also Busby v. Davis, 925 F.3d 699, 723 (5th Cir. 2019), cert. denied, 140 S. Ct. 897 (2020) (no ineffective assistance of counsel because it was “highly likely” that the result would have been the same without the error). E Harper’s final argument on appeal is that his trial counsel was ineffective for not arguing that his mental illness rendered his confession involuntary. He argues that had counsel raised this argument, there is a reasonable probability that the trial court would have suppressed his confession, or that at least one juror would have voted to acquit. We are not considering this issue on a clean slate. The habeas court considered Harper’s claim and held against him. It found that trial counsel was not ineffective because the trial court did consider whether Harper’s claim was voluntary. The habeas court noted that the trial court watched the video of Harper’s confession which showed that Harper was given his Miranda warnings, asked questions about the warnings, and did not appear intoxicated or otherwise impaired. Based on these facts, the trial court found that Harper was not coerced in any way, and admitted the confession. In light of the trial court’s findings, the state habeas court concluded that the result would not have been any different had trial counsel made this additional argument. The habeas court also concluded that it was “speculative at best that at least one juror would have found the applicant’s confession in voluntary.” The habeas court’s findings were not an “unreasonable determination of the facts.” 28 U.S.C. § 2254(d)(2). The trial court and the jury were both able to watch Harper’s confession and heard testimony about his mental health. Based on these facts, they were able to consider what probative value Harper’s confession had even without a specific argument from Harper’s trial counsel. Because no reasonable jurist could find that the habeas court’s decision was unreasonable, no COA will issue on this claim. III We DENY Harper’s application for a COA on all claims.
