Kervinton Valentino v. Harold Clarke

PUBLISHED UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT No. 18-7295 KERVINTON VALENTINO, Petitioner – Appellant, v. HAROLD CLARKE, Director, Virginia Department of Corrections, Respondent – Appellee. Appeal from the United States District Court for the Eastern District of Virginia, at Alexandria. Anthony John Trenga, District Judge. (1:17-cv-00373-AJT-TCB) Argued: December 10, 2019 Decided: August 26, 2020 Before KING, AGEE, and RICHARDSON, Circuit Judges. Affirmed by published opinion. Judge Richardson wrote the opinion, in which Judge King and Judge Agee joined. ARGUED: Bradley Rittenhouse Haywood, Arlington, Virginia, for Appellant. Katherine Quinlan Adelfio, OFFICE OF THE ATTORNEY GENERAL OF VIRGINIA, Richmond, Virginia, for Appellee. ON BRIEF: Mark R. Herring, Attorney General, Victoria N. Pearson, Deputy Attorney General, Donald E. Jeffrey, III, Senior Assistant Attorney General, OFFICE OF THE ATTORNEY GENERAL OF VIRGINIA, Richmond, Virginia, for Appellee. RICHARDSON, Circuit Judge: According to Virginia prosecutors, Kervinton Valentino beat, shot, and robbed a prostitute in an Alexandria hotel room. But Valentino claimed an unknown assailant sprung upon him—wounding the woman as well as Valentino himself. The trial was a proverbial swearing match: “If you believe [the victim],” the prosecution summarized, then “the defendant is guilty of all charges. If you don’t believe her, if you believe the Defendant’s version, then [] he’s not guilty.” J.A. 461. Although he concurred with the prosecution’s premise, Valentino argued his story was sound. Yet the jury found him guilty, and the trial judge sentenced Valentino to twenty years in prison. In state habeas proceedings, Valentino moved to overturn his convictions on the theory that his trial attorney was so ineffective as to violate the Sixth Amendment right to counsel. See Strickland v. Washington, 466 U.S. 668, 687 (1984). Valentino claimed that his lawyer should have sought forensic testing of various items in the hotel room to bolster his story. The state post-conviction court agreed that Valentino’s bloody sock deserved DNA testing. Even so, the court found this failure did not harm Valentino’s defense. As for the rest of Valentino’s claims, the state court held trial counsel’s performance neither deficient nor prejudicial. Thus, the state court denied post-conviction relief without ordering new forensic testing. Valentino then brought a federal habeas petition again raising these Sixth Amendment claims. See 28 U.S.C. § 2254. First, the district court found the state post- conviction court previously “adjudicated” Valentino’s “claim[s] . . . on the merits.” J.A. 761; see § 2254(d). This finding triggered our highly deferential standard of review for 2 state-court convictions. Next, applying that standard, the district court held that the state court’s adjudication was neither unreasonable nor inconsistent with Supreme Court precedent. See J.A. 764; § 2254(d)(1). So the court dismissed Valentino’s habeas application and rejected his request for an evidentiary hearing. Because we agree that the state court adjudication was not unreasonable, we affirm. I. Background In the early hours of May 2, 2012, Kervinton Valentino arranged to pay Aaliyah Islam for sex in a hotel room. On that foundational fact, the Parties agree. But from there, they paint starkly different pictures of the subsequent robbery and shooting. According to the Commonwealth, Valentino pulled a gun on Islam, robbed her, beat her, and shot her. The defense, however, asserted that Islam was not alone: she and an accomplice planned to rob Valentino—they pulled a gun on him. We summarize each case in turn. A. The prosecution’s case 1. Islam’s testimony The Commonwealth built its case on Islam’s testimony. On May 1, 2012, Islam flew from Northern California to Virginia to “have sex for money.” J.A. 19, 32. After landing at Dulles Airport, Islam took a shuttle to an extended-stay hotel in Alexandria where she planned to entertain clients. Her room included a kitchen, bathroom, and bedroom with a doorless coat closet. Although Islam attested to occupying her room alone, she kept a pair of toothbrushes and two deodorant sticks in the bathroom. One toothbrush, she explained, was for her teeth, and the second was used to apply hair product. Islam 3 further claimed that she layered on two brands of women’s deodorant since she “sweat[s] a lot.” J.A. 117. After perusing Islam’s online profile on “backpage.com,” Valentino arranged for an encounter. He drove to Islam’s hotel, parked in the lot, and met Islam at the agreed time. Islam let Valentino into the hotel and escorted him to her third-floor room. Seeing that Islam was alone, Valentino claimed to have forgotten his wallet. So he went back to his car, leaving his cell phone in Islam’s room. In reality, the prosecution argued, Valentino went to retrieve his gun after confirming Islam lacked security. Islam again let Valentino into the hotel, they returned to the room, and Valentino retrieved $140 in cash from his pocket. (Islam never saw a wallet.) Valentino then undressed, carefully placed his clothes on the floor, and tried to have sex with Islam. 1 When Valentino was unable to maintain an erection, the encounter quickly devolved into a violent robbery. Islam looked down at her cell phone to text her boyfriend in California as Valentino retrieved his clothes. 2 When Islam looked up, Valentino had “a gun in [her] face.” J.A. 57. Valentino commanded, “Give me everything. Everything you got give it to me. Where’s all the money at?” J.A. 58. He then swiped Islam’s cell phone from her hand, took back the $140, and stole another of Islam’s phones from the dresser 1 Islam testified that this action was odd to her as “usually people place [their clothes] on the table or on the chair . . . but [do] not lay their clothes on the floor.” J.A. 54. The prosecution argued that Valentino did so to conceal his gun. J.A. 471. 2 The defense would insinuate that Islam’s boyfriend was the assailant. But cell- site location data offered by the prosecution showed that the recipient of Islam’s texts remained in Northern California during the encounter. See J.A. 424−25. 4 drawer. Valentino again pointed the gun in Islam’s face, threatening that, if Islam said “anything smart[,] . . . he was going to shoot [her] in the foot.” J.A. 59. After quickly searching the room and Islam’s belongings, Valentino discovered a laptop hidden under the bed. Valentino then “cocked the gun back,” saying, “I’m going to give you ten seconds to give me everything you got . . . . [I]f you don’t give me everything you got and I find it, I’m going to shoot you.” J.A. 63. But Islam had just arrived in Virginia and explained that she had nothing else to give. Valentino continued to threaten Islam, and eventually, Islam “jumped up and pushed the gun out of [her] face.” J.A. 64. But Islam—at 5’3” and 110 pounds—was little match for Valentino—armed, half-a-foot taller, and more than fifty pounds heavier. Grabbing Islam by the hair, Valentino slung her to the floor, ripping the hair extensions from her head and punching her in the face. At one point, Valentino dropped the gun. Islam reached for it, but Valentino got there first. He shot her in the leg. After shooting Islam, Valentino “hesitated to leave,” debating whether to take her laptop. J.A. 71. He left it and fled the hotel through a side door. Islam testified that, as he left, Valentino was walking normally, and he did not “wince or cry in pain.” J.A. 118. Islam got up and went to the hallway window, leaving bloody footprints in her wake. She watched Valentino enter a black SUV. Islam then called 911 from her hotel-room phone. Responding to a report of gunshots and a woman screaming, law enforcement was already headed to the hotel. When they arrived, they observed the bloody footprints leading from a hotel room to a window in the hallway and back to the room. They knocked on the door, and Islam let them in. Her left leg was wrapped in a towel, “soaked with 5 blood.” J.A. 122. Officers helped Islam from the doorway and sat her down in the hall. Inside, blood plastered the bedroom and the bathroom. Clothes were also strewn around the bedroom, and Islam’s torn hair extensions were scattered on the floor. Police searched Islam’s room and found a single bullet casing in the bedroom. 3 Islam was taken to the hospital and treated for her injuries. At trial, she authenticated photos that showed where the bullet entered and exited her leg. Islam also identified x-rays showing where the bullet “crossed either the fibula or the tibia and the smaller bone” and fragmented inside her leg. J.A. 81, 83–84. 4 According to Islam, Valentino shot her; no one else was involved in the altercation. 5 Islam testified that Valentino was not shot during the fight. And, she continued, when Valentino left the hotel room, he was not limping and never indicated that he was in pain. 6 3 A casing is the part of a bullet cartridge ejected from a semi-automatic handgun (like Valentino’s) after a bullet is fired. It is not the bullet (the projectile) itself. Law enforcement did not find a bullet in Islam’s hotel room. As one crime scene investigator testified, bullets are not always found. And here, law enforcement declined to rip up the carpet, remove the A.C. unit, or take other measures to find a bullet or its fragments. 4 Islam proceeded to explain how “the x-ray . . . didn’t show that [her bones] [were] going to shorten” as a result of the gunshot and so she didn’t need surgery. J.A. 81. The prosecution relied on Islam’s testimony about her injuries and exhibits from the hospital, never calling a medical expert. Valentino’s counsel did not object. 5 The prosecution called Islam’s previous client that night to confirm that she was alone. According to that client, no one was in Islam’s room, no one waited outside the room, and no one approached him. 6 Yet another of Islam’s clients from that night testified that, before the police arrived, he heard a gunshot and then saw a man “c[o]me out [of the side entrance of the hotel] running to his car. When he was next to his car for a second he saw me and then he got into his car and left.” J.A. 341, 345. Although the client thought the man was (Continued) 6 2. Valentino’s arrest and interview The next day, the police tracked Valentino to a Chesapeake motel using a “find my Android” feature on one of Islam’s stolen phones. Based on the number in Islam’s cell phone, Valentino had texted her father: “Tell [your] daughter to stop prostituting. This is the police.” J.A. 310. Valentino also tried to speak with Islam’s father on the phone and tried to call another contact that he believed to be Islam’s mother. Outside the motel, officers spotted a black SUV and ran its plates. The car was registered to a Kervinton Valentino. Officers then witnessed a man matching Valentino’s description (5’9”, around 165 pounds with a beard and a tattoo on his right forearm) leaving the motel with a rolling suitcase. One officer approached the man and asked for his name. In response, Valentino began “physically shaking,” his hands “were stuttering,” and there was “some nervousness in his voice.” J.A. 182. Officers patted Valentino down, finding a semi-automatic handgun concealed at his waist. Julien Mason, a firearms expert with the Virginia Department of Forensic Science, would later match this gun to the bullet casing found in Islam’s hotel room. Valentino “screamed” that he “just found the gun in the room.” J.A. 189, 507. But Valentino would not answer which room. J.A. 189. When asked what he had been doing in the room, Valentino said he was “having sex with a girl.” J.A. 189. But then Valentino “denied being in the hotel at all.” J.A. 191. Valentino, he was “not 100 percent sure.” J.A. 343. The client did, however, identify Valentino’s car and explained that the man was not limping as he ran to it. 7 One of Islam’s cell phones was in Valentino’s pocket. The other was in his black SUV. When later asked about the rolling suitcase he was pulling, Valentino said that it belonged to a woman who was a friend of his. He told the officer “[t]hat they had had an argument” and that she possibly “owed him money.” J.A. 392. He then said that the two were “going to negotiations.” J.A. 392. The day after officers arrested Valentino, they conducted an interview with him on tape. In that interview, Valentino claimed that he was sitting on the bed with Islam when a man opened the closet door, pulled a gun on him, and patted him down. (Yet, as a detective explained, the closet had no door, and an individual sitting on the bed could see directly into the closet area.) Valentino also claimed that he had never undressed during the interaction. (Still, at another point in the interview, Valentino claimed to have disrobed and attempted to put on a condom. And at trial, he described being at least partially disrobed.) Valentino at first described the assailant as a black male about his height. But officers “kept having to go back and bring [Valentino] to tell [them] what the guy looked like.” J.A. 390. (And, at still another point, Valentino described the man as smaller than he was, lacking “a masculine build.” J.A. 390.) While fighting off the assailant and Islam, Valentino claimed that he managed to gain control over the gun and both of Islam’s phones. And after the encounter, Valentino claimed the assailant called him on Islam’s cell phone. Officers also asked Valentino about his injured foot. Valentino’s foot experienced some type of injury, and his boot had a small hole. Valentino claimed to have shot himself with “my gun.” J.A. 396–97. Officers asked if they could take Valentino’s boots for testing. He agreed. 8 B. The defense’s case 1. The hotel struggle At trial, Valentino told a very different story. Citing difficulties with his wife and a persistent sex addiction, Valentino acknowledged that he arranged to meet with Islam. At the hotel, Valentino testified that Islam handed him a condom, which he tried to put on. He “didn’t like the texture” and asked Islam “if she [had] another one that is more durable and stronger.” J.A. 266. When Islam said no, Valentino said that he had another in his truck. She agreed to let him retrieve it, and Valentino then claimed to spend “several minutes” retrieving the condom and charging his iPhone in his vehicle. J.A. 267. Islam then again opened the side door of the hotel to let him back in. 7 “Talking and laughing,” the pair returned to the third-floor room. J.A. 268. But “as soon as the door shut, a guy came out of the closet and pointed a gun” at Valentino. J.A. 268. 8 He demanded Valentino’s money, and Valentino threw his wallet on the bed. Islam then checked Valentino’s back pockets. 7 Valentino twice told Islam that he was “driving a white Chrysler with New York plates.” J.A. 44; see also J.A. 52. Yet Islam testified that she did not ask about his car and told him that she didn’t need to know what car he was in. The prosecution argued that Valentino did so to mislead the police if Islam called them after his planned robbery. 8 There were inconsistencies in this testimony. On cross examination, Valentino could not recall whether he was robbed the first time he went into the room or the second time. J.A. 304 (“Q: So nobody robbed you the first time you go into the room. It’s the second time you go into the room? A: Ma’am, I have no idea.”). And in a recorded interview with detectives, Valentino claimed that he had been sitting on the bed with Islam when the man opened the coat closet. 9 When the man went to pat Valentino down, Valentino noticed “the clip in the gun wasn’t all the way down,” so he “started to fight the guy.” J.A. 271. Islam then joined the fray. At trial, Valentino described a physical struggle that lasted “[a]bout three minutes.” J.A. 271–72, 274–76, 279. As he simultaneously fought Islam and her accomplice, Valentino managed to take Islam’s cell phone and put it in his pocket. See J.A. 275 (“I don’t know why I put the phone in my pocket . . . I just did it.”). He would later take Islam’s second cell phone 9 and the gun he wrestled from the man. At one point during the altercation, Valentino was “shaking” Islam when he heard a gunshot. J.A. 382−83; see also J.A. 276. At that moment, Valentino explained, Islam was between him and the man with the gun. When the man fired the gun, Valentino said he immediately entered “survival mode”: I dropped the gun and I dropped the phone and I panicked. [The man] tried to pull me back by my T-shirt. I pulled [the shirt] out of his hand and ran out of the room. I grabbed my keys that were under the TV. J.A. 278–79. Valentino then “deactivated [his] Facebook” out of fear that the assailant would gain information about his family from it and “went straight home.” J.A. 284, 286. Valentino never explained how he had come to hold the gun after the assailant fired it. Nor did he explain how he had retrieved the gun (or the phone) from the room after he dropped it. 9 Valentino claimed that he planned to use her phone to call 911. But he never did, explaining that he “didn’t want to go to the police because [he] [didn’t] want [his] wife to find out [he] was with a prostitute.” J.A. 314. 10 2. Leaving the hotel About 45 minutes after leaving the hotel, Valentino experienced foot pain. When he took his shoe off, Valentino saw that his “whole foot was bloody.” J.A. 287. He then found a bullet in his shoe and “just put it away.” J.A. 287. The defense admitted Valentino’s bloody sock and boots into evidence. And while an image of the boots was projected before the jury, Valentino identified the hole “where the bullet entered [his] foot.” J.A. 290. The defense also displayed a picture of Valentino’s foot, and Valentino pointed to where he claimed to have been injured. Still on the stand for cross examination, Valentino described the circumstances of his arrest. He acknowledged that he tried to call Islam’s parents “a couple times.” J.A. 309. According to Valentino, he wanted to give the gun back and get his green card back, which he claimed to have left in the hotel room. Valentino also explained that he sent a text message to Islam’s father claiming to be the police “because [he] wanted to initiate a conversation” after his phone calls went unanswered. J.A. 310. Valentino then claimed to have been arrested after eating at a Waffle House and going to the motel to use the bathroom. When asked why he had a suitcase with him if he was just using the bathroom, Valentino replied that he had “met a person and [] loaned them [his] suitcase,” and that the person in return “gave [him a] suitcase.” J.A. 319. Valentino first testified that he could not remember the person’s name. When pressed, he claimed that he could not pronounce it. Valentino also discussed his recorded interview with detectives. He claimed that, while he was telling his story, the detectives “were showing signals behind [his] back so 11 [he] lost trust” in them. J.A. 306. Valentino also asserted that, because he has a “broken accent” and “speak[s] fast,” he was misinterpreted and misquoted. J.A. 312. Similarly, Valentino claimed that he never told the officers that he was in a room to have sex with a girl. Nor did he tell the officers that he was taking the suitcase to “negotiations.” J.A. 319. 3. The defense’s physical evidence The defense’s physical evidence focused on Valentino’s boot. Mason, the firearms expert, described testing Valentino’s boot “for any indication that the bullet may have impacted the boot.” J.A. 358. Mason found “lead residue along the margins of the damage . . . indicative of a lead projectile such as a bullet or bullet fragment having impacted the boot.” J.A. 359. “The damage,” Mason explained, was “consistent with a slow-moving projectile or an unstable projectile striking the boot.” J.A. 359−60. He elaborated that whatever the lead object was it did not make a “direct impact” on the boot. J.A. 360, 362. And if the lead object was a bullet, “any impact on another object prior to impacting the boot would slow it down.” J.A. 360. For instance, “a bullet passing through a person’s leg striking bone would slow it down and that could be a possible reason why a bullet would . . . do this type of damage.” J.A. 360–61. But Mason’s testimony also cast doubt on the possibility of a ricochet damaging Valentino’s boot. On cross examination of Islam, defense counsel elicited testimony that Valentino was about eight feet away when he shot her. Mason explained that “bullets tend to travel in a straight line,” so it was unlikely that a bullet could pass through Islam’s leg and then rebound with sufficient energy to damage Valentino’s boot. J.A. 360–61. According to defense counsel, this meant that a ricochet could not have occurred under 12 Islam’s account. See J.A. 485 (“Mason . . . said with a reasonable degree of scientific certainty that cannot happen. Her version, whatever else you think about her, can’t happen.”). 10 The defense also called Shiao-Mei Smith, an analyst at the Virginia Department of Forensic Science, who tested Valentino’s boots. First, Smith performed a blood indication test on the exterior of the boots to determine whether blood was present. It was—the indicator came back positive. Then, she tested a stain on the inside of Valentino’s boot for DNA. The DNA analysis showed a “mixture profile”—that is, evidence of more than one person’s DNA. J.A. 372. Although Islam and her boyfriend were excluded as contributors to the major profile in the mixture, Valentino could not be eliminated as the major profile contributor. Given this testimony’s context and results, the jury could reasonably infer that Valentino was the main contributor of DNA to the stain. See State v. Wright, 253 P.3d 838, 843–44 (Mont. 2011). But available testing methods prevented any definitive analysis of the minor profile. In other words, Smith’s analysis indicated that another person’s DNA was in the stain, but that person could not be identified. Smith also noted the limitations of her analysis. She explained that although she knew there was DNA on the stain, she could not tell when that DNA was deposited. Nor 10 Recall, however, that Islam testified that Valentino was not shot during the encounter. And, in any event, the prosecution disagreed with defense counsel’s argument. See J.A. 506 (“Mason told you that bullets sometimes hit hard objects like human bone and they break up. He said fragments can ricochet. Perhaps when he went to shoot her, a piece of the bullet flew back into his foot.”). 13 could Smith determine the source of the DNA—whether blood, “sweat or semen or any other number of substances that would contain a person’s DNA.” J.A. 376. Along with the bloody sock and Valentino’s boots, the defense introduced security- camera footage from the hotel’s front desk. Valentino explained that Islam was shown on the security footage at around 2:04 A.M. He also testified that he was “60 to 70 percent” sure that another man in the security video was the man who later attacked him in Islam’s room. J.A. 263–64. Valentino claimed to have recognized him based on the man’s slim body shape, white shirt, the shape of his head, and the way he walked. During cross examinations of government witnesses, the defense tried to achieve two objectives. First, it sought to support Valentino’s story. So, for instance, on cross examination of Islam, the defense highlighted the two toothbrushes and deodorant sticks that she kept in her bathroom, implying that someone else was staying in the room. Similarly, on cross examination of one of Islam’s clients, the defense elicited the testimony that two men left the hotel after the gunshot but before the police arrived—one running out the side door to Valentino’s car and another walking out the front door. Second, the defense sought to paint a picture of a rushed investigation in which the state “was more interested in finding evidence to convict than in giving the full and fair investigation.” J.A. 491. So on cross examination of the state’s investigators, the defense called their decision-making into question. For instance, the defense highlighted the state’s choice to test the hole in Valentino’s boot for indicia of a gunshot rather than DNA. On the outside of the boot near a small hole, Smith performed a blood indication test, which came back positive. But since the 14 hole itself was so small, Mason and Smith explained, performing both tests would have been impossible. Conferring with detectives, Smith and Mason agreed to test the hole for gunshot indicia and inside the boot for DNA. In the defense’s view, this choice to test the hole for evidence of a gunshot, not DNA, strayed from the state’s obligation to conduct a full and fair investigation. J.A. 491; see also J.A. 490 (The detective “promised and committed to doing a full and fair investigation. . . . He made the call do we look for DNA on the exterior of the boot or do we look for gun powder residue. He’s trying to convict and not trying to give a full and fair investigation.”). Similarly, the defense emphasized the prosecution’s failure to test Valentino’s bloody sock for DNA. 11 As defense counsel explained, the detective “just threw [the socks] in his locker and left them there . . . he sure didn’t bother to take them down to the lab to be analyzed.” J.A. 490. This, counsel explained, was more evidence of the state’s poor investigation. C. The verdict and sentencing The jury convicted Valentino of all counts: robbery, Va. Code § 18.2-58, malicious wounding, § 18.2-51, and the use of a firearm while committing those offenses, §§ 18.2- 11 Neither the state nor the defense formally sought forensic testing of Valentino’s sock. The socks, which Valentino claims were the ones he was wearing, were allegedly given to Valentino’s attorney by one of Valentino’s friends. J.A. 601–02 (defense motion for testing). The defense attorney then gave the socks to the prosecution at a preliminary hearing. At the time, Valentino’s counsel told the detective “to do as you wish.” J.A. 332. The detective did not interpret this statement as a request by the defense to test the bloody sock. Rather, he interpreted defense counsel’s action as simply turning over evidence. 15 53, -53.1. 12 Valentino moved for a new trial on the grounds that the Commonwealth should have conducted DNA testing on his bloody sock. He asserted that, if Islam’s DNA was on Valentino’s sock as he claimed, it would have definitively proved his version of events (that Islam was between Valentino and an assailant when the gun went off) rather than one of the Commonwealth’s theories (that bullet fragments ricocheted after hitting Islam, that Valentino shot himself later, that Valentino was never really injured, or that the injury was caused by something else entirely). 13 Although defense counsel claimed that he believed the DNA testing would be done, he acknowledged that he never filed a motion to have the evidence tested. Moreover, when the defense realized that testing had not been performed, counsel admitted to reaching a 12 The defense requested a poll of the jury, see Va. Sup. Ct. R. 3A:17(d), and the record shows that all twelve jurors affirmed their verdicts. The trial judge then affirmed the unanimity of the verdict. J.A. 521; see Va. CONST. art. I, § 8 (guaranteeing the accused in a “criminal prosecution[] . . . the right to . . . an impartial jury of his vicinage, without whose unanimous consent he cannot be found guilty”); cf. Ramos v. Louisiana, 140 S. Ct. 1390, 1397 (2020). No one objected. But two days later, one juror phoned defense counsel claiming that she did not, in fact, affirm the verdict in open court. According to defense counsel, “[b]ecause the defendant and his mother were crying loudly while the jury polling process was being conducted, neither the defendant nor counsel realized at the time the jury was polled that a juror did not answer.” J.A. 523. At a motions hearing, the judge rejected Valentino’s motion for a new trial on this basis. 13 As the Commonwealth explained at oral argument, the jury did not need to definitively resolve this question to convict Valentino. And the Commonwealth declined to bind itself to any single theory. See Oral Arg. 16:20–17:35; see also J.A. 506 (“What about the bullet? [Defense counsel] says it must be that the bullet traveled through [Islam’s] leg and then into the Defendant’s shoe. If you guys want to go back there and try to play jury twister and figure out how that happened, by all means try but I don’t think you’re going to figure out that is how it happened.”). 16 “tactical decision” about how to use that fact—to argue that the investigation was unfair and rushed—instead of asking for a continuance. J.A. 571. Citing these reasons, the court rejected Valentino’s motion for a new trial. The trial judge sentenced Valentino to twenty years in prison. J.A. 582–83. His conviction was affirmed in the Virginia Court of Appeals. See Valentino v. Commonwealth of Virginia, Va. Ct. App. Rec. No. 2005-12-4 (Va. Ct. App. June 4, 2013). The Virginia Supreme Court declined to review his case. D. Post-conviction proceedings After his direct appeal was rejected, Valentino filed a habeas petition in state court raising five claims of ineffective assistance of counsel: (1) Failure to seek forensic testing of the bloody sock, the clothing in Islam’s room, the two toothbrushes, and the two deodorant sticks; (2) Failure to challenge Islam’s medical testimony; (3) Failure to challenge the admission of evidence of the circumstances surrounding Valentino’s arrest; (4) Failure to object to the admission of cell-site location information; and (5) Failure to pay attention to the jury poll. And as to his forensic testing claim, Valentino asked the state court to order an analysis. The state circuit court rejected Valentino’s habeas petition, stating (in full): IT APPEARING that the trial record provides a sufficient basis upon which to determine the merits of the Habeas Petition without the necessity of an evidentiary hearing, and UPON FINDING that while trial counsel’s failure to seek forensic testing of Petitioner’s bloody socks was unreasonable, petitioner has failed to establish “a reasonable probability that, but for counsel’s unprofessional error[], the result of the proceedings would have been different,” Strickland v. Washington, 466 U.S. 668, 694 (1984); and UPON FURTHER FINDING that, as to all other claims set forth in the Habeas Petition, petitioner has failed to show that trial counsel’s performance 17 was deficient and that he was prejudiced as a result; it is ORDERED that Respondent’s Motion to Dismiss be, and hereby is, granted. J.A. 726–27 (alterations in original). Valentino timely appealed the circuit court’s judgment to the Supreme Court of Virginia. Finding no reversible error, it refused Valentino’s petition for appeal. Valentino then filed the instant § 2254 petition in federal district court, re-raising all his claims except the failure to object to the cell-site location information. Again, Valentino sought forensic testing of his sock, Islam’s clothing, and the duplicate toiletries. The district court dismissed Valentino’s petition. First, the district court found that the state circuit court’s ruling was an adjudication on the merits. See Valentino v. Clarke, No. 117-cv-373-AJT-TCB, 2018 WL 9814680, at *2 (E.D. Va. Sept. 19, 2018). Thus, it reviewed that ruling for reasonableness under § 2254(d), a “highly deferential standard.” Id. And finding the adjudication reasonable, the district court rejected Valentino’s claims and refused to order discovery. The district court issued a Certificate of Appealability “limited to whether the Court should order further discovery and an evidentiary hearing, including whether to compel . . . forensic testing for the items at issue in [Valentino’s claim that his attorney was ineffective for failing to pursue further forensic testing].” Id. at 4. At Valentino’s request, we expanded that certificate to explicitly include the full substance of Valentino’s claim that his attorney was ineffective for failing to pursue further forensic testing. Valentino v. Clarke, No. 18-7295 (4th Cir. Feb. 19, 2019) (Order). 18 II. Discussion At the outset, we review the vantage point from which we assess constitutional challenges to state court proceedings. The Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”), Pub. L. No. 104-132, 110 Stat. 1214, authorizes federal courts to “entertain an application for a writ of habeas corpus” for a convicted state prisoner “on the ground that [his] custody [] violat[es] the Constitution or laws” of the United States. 28 U.S.C. § 2254(a). But the manner in which the federal courts may entertain such an application depends considerably on how the state court treats a petitioner’s claims. AEDPA contains several procedural requirements structured to ensure the state courts remain “the principal forum for asserting constitutional challenges to state convictions,” rather than “a preliminary step for a later federal habeas proceeding.” Harrington v. Richter, 562 U.S. 86, 103 (2011). First, AEDPA requires all state prisoners to “exhaust[] the remedies available in the courts of the State” before seeking federal habeas relief. § 2254(b)–(c); see also Harrington, 562 U.S. at 103. This exhaustion requirement gives state courts “a meaningful opportunity to consider allegations of legal error without interference from the federal judiciary.” Vasquez v. Hillery, 474 U.S. 254, 257 (1986). State courts are “presumptively competent[] to adjudicate claims arising under the laws of the United States.” Tafflin v. Levitt, 493 U.S. 455, 458 (1990). Thus, “[i]t would be unseemly,” the Supreme Court has explained, “for a federal district court to upset a state court conviction without an opportunity to the state courts to correct a constitutional violation.” Darr v. Burford, 339 U.S. 200, 204 (1950). 19 When state courts embrace their leading role in correcting constitutional errors and righting wrongful convictions, the demands of “comity and federalism” reach their apex. Rhines v. Weber, 544 U.S. 269, 273 (2005). Recognizing these “foundational principle[s] of our federal system,” AEDPA “erects a formidable barrier to federal habeas relief for prisoners whose claims have been adjudicated in state court.” Burt v. Titlow, 571 U.S. 12, 19 (2013). For these already-adjudicated claims, § 2254(d) permits only two paths to federal habeas relief. See Harrington, 562 U.S. at 98. First, if the state decision turned on a factual finding, that finding may be disturbed only if “objectively unreasonable in light of the evidence presented in the state-court proceeding.” Miller-El v. Cockrell, 537 U.S. 322, 340 (2003); see also § 2254(d)(2). Otherwise, the state ruling must have been “contrary to, or involved an unreasonable application of, clearly established federal law” for federal habeas intervention. § 2254(d)(1); see also Virginia v. LeBlanc, 137 S. Ct. 1726, 1728–29 (2017). Moreover, in conducting § 2254(d)’s deferential review, a federal court is limited to reviewing the record as presented to the state post-conviction court. Because “review under § 2254(d)(1) focuses on what a state court knew and did,” we measure the reasonableness of the state court’s decision based on the information in the record before it. Cullen v. Pinholster, 563 U.S. 170, 182 (2011); see also Lockyer v. Andrade, 538 U.S. 63, 71–72 (2003). And since “evidence later introduced in federal court is irrelevant to § 2254(d)(1) review,” a district court need not hold an evidentiary hearing before denying a petitioner’s claim based on § 2254(d)(1). Cullen, 563 U.S. at 184; see also Schriro v. Landrigan, 550 U.S. 465, 474 (2007); see generally Samuel R. Wiseman, Habeas After 20 Pinholster, 53 B.C. L. REV. 953, 968–70 (2012). Likewise, § 2254(d)(2) provides for a limited review of factual determinations “in light of the evidence presented in the State court proceeding.” And “[t]his backward-looking language” similarly “requires an examination of the state-court decision at the time it was made.” Cullen, 563 U.S. at 182, 185 n.7; see also Elmore v. Ozmint, 661 F.3d 783, 850 (4th Cir. 2011). On the other hand, if a state court shuns its primary responsibility for righting wrongful convictions and refuses to consider claims of error, the weighty concerns of federalism and comity are diminished. See Winston v. Kelly, 592 F.3d 535, 555, 557 (4th Cir. 2010) (“Winston I”). AEDPA thus permits a federal court to bypass § 2254(d)’s limitations on relief when a state court has refused to “adjudicate” a procedurally proper claim “on the merits.” § 2254(d). In this rare scenario, the gloves come off: The federal habeas inquiry is more penetrating, and—if consistent with statute and the Rules Governing § 2254 Cases—we may hear evidence that would otherwise be immaterial under § 2254(d)’s limited review. See Gordon v. Braxton, 780 F.3d 196, 202 (4th Cir. 2015); Winston v. Pearson, 683 F.3d 489, 496 (4th Cir. 2012) (“Winston II”); Winston I, 592 F.3d at 556. Accordingly, the scrutiny and scope of our review of Valentino’s petition turns on whether the state post-conviction court previously “adjudicated” his “claim[s]” “on the merits.” § 2254(d). If so, AEDPA requires us to apply § 2254(d)’s limitation on relief. If not, we must remand to the district court for de novo review and a possible evidentiary hearing. See Gordon, 780 F.3d at 202. 21 A. The state post-conviction court denied Valentino’s claims on the merits Whether the state court decided Valentino’s claims on the merits is a legal question that we review de novo. Muhammad v. Kelly, 575 F.3d 359, 367 (4th Cir. 2009). “When a federal claim has been presented to a state court and the state court has denied relief,” we “presume[] that the state court adjudicated this claim on the merits.” Harrington, 562 U.S. at 99. Valentino bears the burden of overcoming this “strong but rebuttable” presumption. Johnson v. Williams, 568 U.S. 289, 301 (2013). Valentino asserts the state court failed to adjudicate his claim on the merits when it rejected his petition without ordering forensic testing of his bloody sock. But this argument does not fit neatly within the plain meaning of a “claim that was adjudicated on the merits.” § 2254(d). One might think that a state court adjudicates a claim on the merits when it “rule[s] on,” (Adjudicate), an “assertion of an existing right,” (Claim), based on “[t]he elements or grounds of [that] claim,” (Merits). Black’s Law Dictionary 52, 311,1185 (11th ed. 2019); see also Sellan v. Kuhlman, 261 F.3d 303, 311 (2d Cir. 2001) (“‘Adjudicated on the merits’ has a well settled meaning: a decision finally resolving the parties’ claims, with res judicata effect, that is based on the substance of the claim advanced, rather than on a procedural, or other, ground.”); cf. Cullen, 563 U.S. at 187 (“[T]hat claim was adjudicated on the merits in state-court proceedings. . . . The California Supreme Court denied each of those petitions ‘on the substantive ground that it is without merit.’”). 14 14 And here, that is precisely what the state post-conviction court did. An ineffective-assistance-of-counsel claim has two elements: The claimant must show (1) the deficiency of counsel’s performance and (2) prejudice. Strickland, 466 U.S. at 687. As to (Continued) 22 But our Circuit has adopted a more nuanced interpretation of a “claim” “adjudicated on the merits,” reasoning that a state court could not have properly adjudicated a claim if it decided on a “materially incomplete record.” Winston II, 683 F.3d at 496 (quoting Winston I, 592 F.3d at 555). 15 Thus, Valentino—like all state prisoners in Maryland, the Virginias, and the Carolinas—may evade § 2254(d)’s limitation on relief if he can show that the state post-conviction court has “unreasonably refuse[d] to permit ‘further development of the facts’ of a claim.” Id. (quoting Winston I, 592 F.3d at 555); see also Gordon, 780 F.3d at 202. In Winston I and Gordon, our Circuit assessed what it means for a state post- conviction court to deny a petitioner’s claims on a materially incomplete record. Winston I considered whether the Virginia Supreme Court improperly truncated the factual Valentino’s bloody sock, the post-conviction court denied Valentino’s ineffective- assistance claim on the latter element: Valentino “failed to establish ‘a reasonable probability that, but for counsel’s unprofessional error[], the result of the proceedings would have been different.’” J.A. 726–27 (quoting Strickland, 466 U.S. at 694). Thus, Valentino’s belated request for forensic analysis seems to bear little on whether the state court decided his Strickland claim based on its merits. 15 At least two of our sister circuits have suggested that our interpretation creates a considerable tension with § 2254(d)’s structural limitation on relief—not to mention other provisions of AEDPA that speak directly to evidentiary hearings. The First Circuit, for instance, has explained that the Supreme Court’s interpretation of § 2254(d) in Cullen would be “eviscerate[d]” if a simple request to supplement the record could avoid that limitation on relief altogether. Atkins v. Clarke, 642 F.3d 47, 49 (1st Cir. 2011); see also Garuti v. Roden, 733 F.3d 18, 23 (1st Cir. 2013); Ballinger v. Prelesnik, 709 F.3d 558, 562 (6th Cir. 2013); Cullen, 563 U.S. at 206 (Breyer, J., concurring in part and dissenting in part); Wilson v. Workman, 577 F.3d 1284, 1319 (10th Cir. 2009) (Gorsuch, J., dissenting). In any event, this panel remains bound by our Circuit’s established interpretation. So we apply it faithfully here. 23 development of an Atkins-related ineffective-assistance-of-counsel claim in a death- penalty case. 592 F.3d 535. 16 The relevant state-court proceedings centered on whether the petitioner, Leon Winston, demonstrated “significantly subaverage intellectual functioning.” Id. at 548. If so, he would qualify as “mentally retarded” under Virginia law, Atkins would prohibit his execution, and thus his counsel rendered constitutionally ineffective assistance by failing to mount an Atkins defense at trial. Id. Before the Virginia Supreme Court, Winston offered evidence of I.Q. testing and a separate Fairfax County diagnosis of “mild retardation.” Id. Based on the admitted I.Q. scores, the court found Winston’s marks too high to show significantly subaverage intellectual functioning. Id. And the court rejected the Fairfax diagnosis because it did not necessarily reflect a sufficiently low I.Q. score. Id. But after the Virginia Supreme Court rendered its decision, and while Winston’s subsequent § 2254 federal habeas application was pending in the district court, the underlying I.Q. test supporting the Fairfax diagnosis was discovered. That score, according to the petitioner, proved that he qualified as mentally retarded under Virginia law. Id. Yet, the Virginia Supreme Court had denied, without explanation, Winston’s motion for discovery and an evidentiary hearing to further develop the factual basis of his claims. Id. Thereafter, in determining that the state court was not objectively unreasonable under § 2254(d) in rejecting Winston’s Atkins-related 16 In Atkins v. Virginia, the Supreme Court held that “the execution of mentally retarded criminals” violates the Eighth Amendment’s prohibition on “cruel and unusual punishments.” 536 U.S. 304, 321 (2002). But the Court left the criteria for determining mental retardation to the states. Id. at 317. In Winston, the petitioner argued that his counsel was ineffective for failing to mount a defense under Atkins. 24 ineffective-assistance-of-counsel claim, the federal district court refused to consider the new evidence of the Fairfax County I.Q. score. On appeal, we explained that, when a “petitioner offers, for the first time in federal habeas proceedings, new, material evidence that the state court could have considered had it permitted further development of the facts, an assessment under § 2254(d) may be inappropriate.” Id. at 555. “If the record ultimately proves to be incomplete,” we elaborated, “deference to the state court’s judgment would be inappropriate because judgment on a materially incomplete record is not an adjudication on the merits for purposes of § 2254(d).” Id. at 555–56. We proceeded to find the state-court record materially incomplete. We reasoned that Winston’s newly discovered I.Q. score bore directly on the dispositive question— whether Winston was mentally retarded under Virginia law. And if the evidence showed what the petitioner alleged, Winston would have met that standard, and Atkins would have prohibited his execution. Id. at 557. Thus, we declined to apply § 2254(d) and permitted further factual development. 17 We again considered this question in Gordon, 780 F.3d 196. In that case, the petitioner sought state post-conviction relief on the ground that he instructed counsel to file an appeal shortly after his sentencing. Id. at 203. But the attorney denied that the petitioner had made that request. Id. Despite this “classic factual dispute” on a dispositive question, 17 In Winston II, we revisited Winston I based on the Supreme Court’s intervening decision in Cullen. We found that Cullen did not impact this portion of the Winston I analysis. But see Garuti, 733 F.3d at 23 (asserting that Cullen “essentially” overruled Winston I). 25 the state court refused to hold a hearing and ignored the bulk of petitioner’s allegations. Id. Gordon then filed a federal habeas corpus application, which the federal district court dismissed based on the state court’s reasoning. We considered whether the state court’s dismissal should be reviewed under § 2254(d)’s “highly deferential standard.” Id. at 202. And that depended on whether the state court’s decision was an “adjudicat[ion] on the merits.” § 2254(d). We explained that “the state court focused on one line in Gordon’s affidavit, while ignoring Gordon’s allegations in his papers that he asked [the attorney] to file an appeal.” Gordon, 780 F.3d at 203. Moreover, “at a minimum, Gordon’s affidavit clearly implicated [the attorney’s] duty to consult [about an appeal], which the state court did not address at all.” Id. at 203–04. “As a result, the state court did not adjudicate Gordon’s claim on the merits.” Id. at 204. So we again declined to apply § 2254(d)’s deferential standard. The district court here found Gordon distinguishable. Noting the complete absence of factual development in Gordon, the district court reasoned that, because Valentino’s trial developed at least some facts, the record was not materially incomplete. Valentino, 2018 WL 9814680, at *2. Yet, this all-or-nothing reasoning does not account for Winston I, where the state court relied on both a fully developed trial record to determine guilt and a subsequent proceeding focused on the issue of mental competence. Therefore, we disagree with the district court’s analysis. Although we diverge from its analysis, we agree with the district court’s conclusion that the state court adjudicated Valentino’s claims on the merits. A state court does not unreasonably truncate further factual development when it declines to order discovery of a 26 fact that it finds immaterial. In Winston I, the dispositive legal issue was whether the petitioner had significantly subaverage intellectual functioning, and the Fairfax test score was material to that issue. See 592 F.3d at 548. Yet the Virginia Supreme Court denied discovery aimed at uncovering it. And in Gordon, the dispositive legal issue concerned the effectiveness of counsel, and the petitioner’s request to file an appeal was decisive for that issue. 780 F.3d at 203–04. But again, the state court blinded itself to the evidence. In both cases, we found that the courts unreasonably truncated the development of disputed and material evidence for the legal claims. 18 In contrast, the state court here determined that the bloody-sock evidence was just not prejudicial. A successful ineffective-assistance-of-counsel claim requires showing both deficient performance and prejudice. Strickland, 466 U.S. at 687. But the state post- conviction court found the second Strickland prong lacking: That is, assuming the evidence showed what Valentino alleged, it would not have created “a reasonable probability that, but for counsel’s unprofessional error[], the result of the proceedings would have been different.” Id. at 694; see also Nicolas v. Attorney General of Maryland, 820 F.3d 124, 130–31 (4th Cir. 2016); Spicer v. Roxbury Correctional Institute, 194 F.3d 547, 559 (4th Cir. 1999); United States v. Ellis, 121 F.3d 908, 918 (4th Cir. 1997); cf. Richardson v. Branker, 668 F.3d 128, 140 (4th Cir. 2012) (requiring us to consider “what 18 We express no opinion on what it means for a state court to unreasonably truncate further factual development of a material fact issue. Because the issue raised by Valentino was found immaterial, we need not reach this question. 27 arguments or theories . . . could have [reasonably] supported[] the state court’s decision”). Far from a refusal to “adjudicate” Valentino’s claim “on the merits,” the state court’s conclusion required it to reject the value of the evidence that Valentino offered. 19 Valentino is free to litigate this conclusion’s reasonableness. But because it was an “adjudicat[ion] on the merits,” he must do so through the lens of § 2254(d). B. The state post-conviction court’s determination was not unreasonable under § 2254(d) We review de novo the district court’s assessment of Valentino’s claims under § 2254(d). Morva v. Zook, 821 F.3d 517, 524 (4th Cir. 2016). This requires us to “examine [Valentino’s] argument through the dual lens of the AEDPA standard and the standard set forth by the Supreme Court in [Strickland].” Branker, 668 F.3d at 144. The Sixth Amendment guarantees an accused the “Assistance of Counsel for his defence.” U.S. CONST. amend VI. In Strickland, the Supreme Court explained that this guarantee is designed to “ensure that criminal defendants receive a fair trial.” 466 U.S. at 689. But if the accused’s attorney makes unprofessional errors so serious as to “undermine[] the proper functioning of the adversarial process,” id. at 686, then “counsel [does] not function[] as the ‘counsel’ guaranteed the defendant by the Sixth Amendment,” 19 Valentino focuses his argument on the bloody sock. But to the extent that he bases this argument on the lack of other forensic testing, we reach the same conclusion. The state court first found that Valentino’s counsel did not perform deficiently—a sufficient reason to reject his claim that obviates the need for further discovery on what that evidence would have shown. J.A. 726–27; see also Strickland, 466 U.S. at 687. Second, the court again found that the evidence was not prejudicial. J.A. 726–27. On these facts, we are satisfied that this was an “adjudicat[ion] on the merits.” § 2254(d). 28 id. at 687. Drawing on these principles, the Supreme Court has set forth a two-part test for evaluating ineffective-assistance-of-counsel claims. First, the defendant must show that his counsel performed deficiently. Id. at 688. “Deficient performance[] requires [] showing ‘that counsel’s representation fell below an objective standard of reasonableness,’ as measured by ‘prevailing professional norms’ and in light of ‘all the circumstances’ of the representation.” Owens v. Stirling, --- F.3d ---, 2020 WL 4197742, at *9 (4th Cir. 2020) (quoting Strickland, 466 U.S. at 688). In evaluating counsel’s performance, judicial scrutiny “must be highly deferential”: “[T]o avoid the ‘distorting effects of hindsight,’” we “‘must indulge a strong presumption that counsel’s conduct falls within the wide range of reasonable professional assistance.’” Yarbrough v. Johnson, 520 F.3d 329, 337 (4th Cir. 2008) (quoting Strickland, 466 U.S. at 689). “[T]hat is, the defendant must overcome the presumption that, under the circumstances, the challenged action ‘might be considered sound trial strategy.’” Strickland, 466 U.S. at 698 (quoting Michel v. Louisiana, 350 U.S. 91, 101 (1955)); see also United States v. Tucker, 603 F.3d 260, 264 (4th Cir. 2010); Roach v. Martin, 757 F.2d 1463, 1478–79 (4th Cir. 1985). Second, the defendant must prove prejudice: “[A] reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Strickland, 466 U.S. at 694; see also Gray v. Branker, 529 F.3d 220, 234 (4th Cir. 2008). “The likelihood of a different result must be substantial, not just conceivable,” Harrington, 562 U.S. at 86, so as to “undermine confidence in the outcome” of the trial, Strickland, 466 U.S. at 694. 29 When assessing a Strickland claim through the lens of AEDPA, our review is “doubly deferential.” Knowles v. Mirzayance, 556 U.S. 111, 123 (2009); see also Woods v. Etherton, 136 S. Ct. 1149, 1151 (2016) (“[F]ederal courts are to afford ‘both the state court and the defense attorney the benefit of the doubt.’”) (quoting Burt, 571 U.S. at 15). “[W]here, as here, a state prisoner claims ineffective assistance of counsel as the basis for habeas relief, we must review the claim through the ‘highly deferential’ lens of Strickland as well.” Owens, --- F.3d at --, 2020 WL 4197742, at *9. Thus, “[t]he question ‘is not whether a federal court believes the state court’s determination’ under the Strickland standard ‘was incorrect but whether that determination was unreasonable—a substantially higher threshold.’” Knowles, 556 U.S. at 123 (quoting Schriro, 550 U.S. at 473); see also Hunt v. Lee, 291 F.3d 284, 289–90 (4th Cir. 2002). “‘This double-deference standard effectively cabins our review’ to determining ‘whether there is any reasonable argument that counsel satisfied Strickland’s deferential standard.’” Owens, --- F.3d at --, 2020 WL 4197742, at *9 (quoting Morva, 821 F.3d at 528). The lack of explanation does not mean we do not apply the deferential standard of review. When, as here, the state court has not specified the precise reasons for its decision, “the habeas petitioner must show there was no reasonable basis for the state court to deny relief.” Harrington, 562 U.S. at 98. A state court’s decision is unreasonable where it is “so lacking in justification that there was an error well understood and comprehended in existing law beyond any possibility for fairminded disagreement.” Id. at 103. “If this standard is difficult to meet, that is because it was meant to be.” Id. Section 2254(d) codifies the view that habeas corpus is a “‘guard against extreme malfunctions in 30 the state criminal justice systems,’ not a substitute for ordinary error correction through appeal.” Id. at 102–03 (quoting Jackson v. Virginia, 443 U.S. 307, 332 n.5 (1979) (Stevens, J., concurring in the judgment)). Thus, “[s]urmounting Strickland’s high bar is never an easy task. . . . [But] [e]stablishing that a state court’s application of Strickland was unreasonable under § 2254(d) is all the more difficult.” Harrington, 562 U.S. at 105. Here, we consider two conclusions: First, the state court determined that defense counsel was not deficient for failing to obtain forensic testing of the toiletries and clothes in Islam’s hotel room. Second, the state court held that Valentino was not prejudiced by the lack of forensic testing on the bloody sock. 1. The state court reasonably determined that counsel’s performance was not deficient as to the items found in the hotel room In evaluating the reasonableness of the state court’s determination on the performance prong, we follow a well-trodden path mapped out by the Supreme Court. First, we begin with the “‘strong presumption’ that counsel ‘made all significant decisions in the exercise of reasonable professional judgement.’” Cullen, 563 U.S. at 191, 196 (quoting Strickland, 466 U.S. at 689−90). This requires that we not only “give the attorneys the benefit of the doubt, but [also] affirmatively entertain the range of possible reasons that [Valentino’s] counsel may have had for proceeding as [he] did.” Id. at 196 (internal citations and quotations omitted); see also Harrington, 562 U.S. at 791 (“Strickland . . . calls for an inquiry into objective reasonableness of counsel’s performance, not counsel’s subjective state of mind.”); United States v. Vyner, 846 F.3d 1224, 1227 (D.C. Cir. 2017) (internal citations and quotations omitted) (“[T]he 31 presumption may only be rebutted through a showing that no sound strategy . . . could have supported the conduct.”). Then, we ask whether Valentino has overcome this presumption by showing that no fair-minded jurist could find one of those reasons to be sound trial strategy. See § 2254(d); Harrington, 562 U.S. at 105 (“The question is whether there is any reasonable argument that counsel satisfied Strickland’s standard.”). Valentino asserts that his counsel should have sought forensic testing of items found in Islam’s hotel room: her duplicate toiletries and her clothes. According to Valentino, this evidence “could have produced dispositive evidence of Valentino’s innocence.” Pet’r’s Br. 30. He explains that a DNA analysis of the duplicate toiletries would “have proved that a third party was in the room.” Pet’r’s Br. 30. And gunshot residue on Islam’s clothing, according to Valentino, similarly would have verified his account of events. Pet’r’s Br. 30. But Valentino’s argument paints an overly optimistic picture of what forensic testing would have “proved,” while downplaying the risk of such testing to the trial strategy his attorney pursued. If Valentino’s counsel had requested forensic testing, he would have had to live with the results. Making this kind of decision involves “assess[ing] and balancing [] perceived benefits against perceived risks”—an exercise to which we normally “‘afford . . . enormous deference.’” United States v. Terry, 366 F.3d 312, 317 (4th Cir. 2004) (quoting United States v. Kozinski, 16 F.3d 795, 813 (7th Cir. 1994)). Here, we first note that trial counsel could have found the potential upside of testing limited based on a reasonable perception of a lack of candor in Valentino’s version of events. See Strickland, 466 U.S. at 691 (“[W]hen a defendant has given counsel reason to 32 believe that pursuing certain investigations would be fruitless or even harmful, counsel’s failure to pursue those investigations may not later be challenged as unreasonable.”); see also Elmore, 661 F.3d at 857. On top of changing his story multiple times, Valentino’s theory was extrinsically contradicted and internally contradictory. Compare, e.g., J.A. 386 (sitting on the bed when a man opened the closet door), with J.A. 394 (no closet door) and J.A. 268 (assaulted “as soon as the door shut” the first time), with J.A. 304 (Valentino professing he had “no idea” whether he was assaulted the first or second time he went in the room). Based on the problems with the consistency and coherence of Valentino’s theory, his counsel may have reasonably expected that more forensic testing was unlikely to be helpful. At the same time, defense counsel may have concluded that forensic testing ran the risk of undermining the trial strategy. Valentino attacked the prosecution’s case on two fronts. On the first front, counsel painted an alternative picture of events—a picture predicated on the theory that a second person (the man that attacked Valentino) was in the room with Islam. And in addition to Valentino’s testimony, defense counsel developed additional facts and inferences on cross examination to support the presence of a third person in the hotel room. See J.A. 111–13 (admitting picture of toothbrushes and deodorants); J.A. 113 (“So, you have two deodorants and two toothbrushes in your room.”); id. (“And nobody’s there with you?”); see also J.A. 116–17. Yet, if the toiletries and clothes had been tested and contained only Islam’s DNA, not only would these arguments have been unavailable to trial counsel, but Valentino’s alternative story would have also 33 been undermined. See Harrington, 562 U.S. at 108 (“An attorney need not pursue an investigation that . . . might be harmful to the defense.”). On the second front, defense counsel cited the absence of extensive forensic testing to argue that the Commonwealth was not committed to a “full and fair investigation.” E.g. J.A. 491 (arguing that detectives were “more interested in finding evidence to convict than in giving the full and fair investigation that this case required”). We think this strategy within the realm of reasonable defense tactics. As the Supreme Court has explained, “[t]o support a defense argument that the prosecution has not proved its case[,] it sometimes is better to try to cast pervasive suspicion of doubt than to strive to prove a certainty that exonerates.” Harrington, 562 U.S. at 109; see also Bowen v. Maynard, 799 F.2d 593, 613 (10th Cir. 1986) (A “common trial tactic of defense lawyers is to discredit the caliber of the investigation.”). But again, every additional forensic test would have undermined defense counsel’s argument. Weighing a limited expected upside against significant downside risks, trial counsel may have reasonably determined that he should not seek the forensic testing. For these reasons, we think that the state post-conviction court’s deficiency determination was not beyond debate. And so § 2254(d) precludes relief on these claims. 2. The state court reasonably determined that the failure to test Valentino’s sock did not cause prejudice We next evaluate Valentino’s ineffective-assistance claim for failing to obtain DNA testing of the bloody sock. Valentino argues that “[e]vidence of Islam’s DNA on the boot or the sock would have demonstrated that the same bullet that passed through her leg struck 34 Valentino’s foot, directly contradicting her account and proving the veracity of his.” Pet’r’s Br. 27. 20 Again, the state court disagreed, finding that the failure to test the sock was not prejudicial to Valentino. Evaluating whether counsel’s performance prejudiced the defendant requires us to consider the error in light of “the totality of the evidence before the judge or jury.” Elmore, 661 F.3d at 858; see also Turner v. United States, 137 S. Ct. 1885, 1893 (2017). Depending on the nature of the evidence and the theories available at trial, “[s]ome errors will have had a pervasive effect on the inferences to be drawn from the evidence, altering the entire evidentiary picture, and some will have had an isolated, trivial effect.” Elmore, 661 F.3d at 858 (quoting Strickland, 466 U.S. at 695–96). And it is our job to determine whether the state court’s conclusion on this point “was an error well understood and comprehended in existing law beyond any possibility for fairminded disagreement.” Harrington, 562 U.S. at 103. Here, we find no such error. Note the broad evidentiary picture before the jury: Valentino’s trial boiled down to a swearing match between Islam and Valentino, as both sides argued at trial and now agree on appeal. See Pet’r’s Br. 27 (“[T]he outcome of [the trial] depended almost entirely on a credibility determination.”); Resp’t’s Br. 6 (“That is fundamentally and explicitly what the jury was asked to do: to find that Islam was more credible than Valentino.”). On one hand, the prosecution built its case on Islam’s 20 During the state-conviction proceedings, the prosecution conceded that if a bullet first passed through Islam’s leg before hitting Valentino’s foot, then that bullet could transfer Islam’s DNA to Valentino’s sock. See J.A. 603 n.4, 639. 35 testimony. Her testimony of the material events of that night was detailed, consistent, and persuasive. And it was supported by two other clients, as well as law enforcement’s investigation. On the other hand, Valentino’s story was shifting, extrinsically inconsistent, and self-conflicted. We are not surprised that the factfinder concluded that Islam was credible and Valentino was not. In the context of a credibility competition, how new evidence alters the evidentiary picture makes all the difference. Where evidence would directly contradict uncorroborated eyewitness testimony, we have held the failure to offer it to be prejudicial. See Griffin v. Warden, Maryland Correctional Adjustment Center, 970 F.2d 1355, 1359 (4th Cir. 1992); see also Guzman v. Secretary, Dep’t of Corrections, 663 F.3d 1336, 1355–56 (11th Cir. 2011); Montgomery v. Petersen, 846 F.2d 407, 413–14 (7th Cir. 1988). In contrast, where the evidence would not have undermined the prosecution’s witness(es) or where it relates only to a peripheral issue, it generally fails to set the case in a new light so as to displace our confidence in a verdict. See Turner, 137 S. Ct. at 1894 (Evidence “too distant from the main evidentiary points” fails to create a reasonable probability of a different result.); see also United States v. Robinson, 627 F.3d 941, 952 (4th Cir. 2010); Gilday v. Callahan, 59 F.3d 257, 272 (1st Cir. 1995). Valentino asserts that the evidence here would have proven his version of events and disproven Islam’s testimony. But we think there’s room for reasonable jurists to disagree. Recall the facts the jury heard: A lead object, consistent with a bullet or bullet fragment, left residue on a hole in Valentino’s boot. When the boot was tested for blood, an indicator test came back positive. Valentino’s DNA, as well as that of another person, 36 was found inside the boot. Although the bullet fragmented inside Islam’s leg, it was unlikely that those fragments could ricochet and impact Valentino’s boot. Of course, this evidence is consistent with Valentino’s theory. From these facts, the jury could already infer that a bullet fired by an assailant passed through Islam’s leg and hit Valentino. But the jury rejected this inference in favor of one of the prosecution’s theories—Valentino was never shot, he shot himself later, or the unlikely ricochet did, in fact, occur. Now assess the addition of the proffered fact that Islam’s DNA was found on Valentino’s sock. (Or that the stain on the inside of the boot also contained Islam’s DNA). We do not think this changes the evidentiary picture. This evidence is “largely cumulative of the criticisms [of Islam’s account] known at the time of the trial,” rather than substantive proof of Valentino’s innocence. United States v. Higgs, 663 F.3d 726, 743 (4th Cir. 2011) (citing United States v. Berry, 624 F.3d 1031, 1043 (9th Cir. 2010)). And the failure to present cumulative evidence is generally not prejudicial. See Wong v. Belmontes, 558 U.S. 15, 22 (2009) (“Some of the evidence was merely cumulative of the humanizing evidence [counsel] actually presented; adding it to what was already there would have made little difference.”); see also United States v. Parker, 790 F.3d 550, 558 (4th Cir. 2015) (quoting United States v. Bartko, 728 F.3d 327, 340 (4th Cir. 2013)). One way that this DNA could have reached Valentino’s sock was by a transfer of blood from Islam’s leg, to the bullet, to Valentino’s sock (as Valentino urges). But this was not the only way that the DNA could have transferred. The Commonwealth advances competing theories of secondary transfer. A secondary transfer of DNA occurs when “DNA left on one surface is inadvertently transferred to another surface.” 7 Jones on 37 Evidence § 60:10 (7th ed. 2020). 21 Valentino’s theory—in which Islam’s DNA transfers from the bullet to his sock—was one example of such a transfer. Yet, if Valentino advances one theory of secondary transfer, we think it reasonable for the state court to have considered others. In this sense, Valentino must take the bitter with the sweet. First, consider the Commonwealth’s theory in which Valentino touched Islam’s blood (or another DNA-carrying substance) and transferred it to his sock. 22 At trial, the jury learned that Islam bled profusely. Just after being shot, Islam felt “blood trickle down” her leg. J.A. 71. Officers observed her blood “all over the floor” of the hotel room. J.A. 123; see also J.A. 133 (“Entering the room outside the hallway, there was blood on the floor. Entering the room, there was blood on the kitchen floor leading into the bathroom. There was blood going into . . . the bedroom area.”); J.A. 139–42 (publishing pictures of the bloody hotel room to the jury). Her injury was severe enough that she would leave “bloody footprints . . . leading from” her room “towards . . . the window at the end of the hallway” as she left the room to see Valentino leave. J.A. 121. When police found her, Islam was covering her leg in a towel that was itself “soaked with blood.” J.A. 122. If any 21 “For example, if object X is placed on object Y, DNA originally on X may be transferred to Y, or if a person touches object X, he or she may pick up some of the DNA on X; if that person later touches object Y, he or she may transfer that DNA to object Y. This can occur with any object or surface containing someone’s DNA.” 7 Jones on Evidence § 60:10. 22 See Pet’r’s Br. 42 (“[T]he presence of any of Islam’s DNA on Valentino’s effects was easy to explain—if Valentino had Islam’s blood anywhere on his person or his possessions after he shot her, he could have transferred that blood to his sock and boot after he left the hotel.”). 38 of this blood ended up on Valentino’s person (either when he shot her or when he “hesitated to leave,” J.A. 71), he may have transferred it to his boots and sock. Second, Valentino may have picked up Islam’s blood or DNA from their direct physical contact. Valentino beat Islam. He punched her “several times,” “slung [her] to the floor,” and ripped the hair extensions from her hair. J.A. 64–65. Islam was scratching and hitting Valentino—and Valentino testified that she bit him in the forearm (and maybe also the leg). J.A. 272, 283; see J.A. 276–77. Additionally, Valentino and Islam tried to engage in sexual intercourse. As Islam explained on the stand, she and Valentino disrobed and attempted multiple forms of intimate relations using two different condoms. J.A. 54– 56, 166. The DNA test could not determine the source of the DNA, which could have come from blood, “sweat or semen.” J.A. 376. We would be unsurprised if DNA ended up on Valentino’s hands and then his clothes as he dressed. Such physical contacts provide yet another vector for Islam’s DNA to reach Valentino. Finally, Valentino may have picked up Islam’s DNA from merely touching objects in the room. Cf. Bess Stiffelman, No Longer the Gold Standard: Probabilistic Genotyping Is Changing the Nature of DNA Evidence in Criminal Trials, 24 BERKELEY J. CRIM. L. 110, 115–16 (2019) (“In short, small amounts of DNA can be easily transferred. . . . Because of this, finding someone’s DNA on an object is less significant to a determination of guilt or innocence of a suspect.”). When Valentino got undressed, he placed his clothes on the floor of a hotel room where Islam had been entertaining clients. See J.A. 54, 410–11. Again, this provided another opportunity for Islam’s DNA to transfer from the floor to his sock. 39 In sum, finding Islam’s DNA on Valentino’s sock would not shed new light on the trial or change the evidentiary picture in this case. Valentino’s speculative theory of how Islam’s DNA reached his sock is hardly determinative—on the contrary, it is a marginal addition to the evidence before the jury. And his theory as to how Islam’s DNA could have ended up on his sock is no less speculative nor more persuasive than alternatives offered by the Commonwealth. In the end, Valentino’s trial still would have been a credibility competition. As we review the record, we are left with the impression that Islam recounted a coherent and credible series of events—Valentino did not. And finding Islam’s DNA on Valentino’s sock would not have changed the contours of that contest. So we cannot say that the state court’s conclusion was erroneous beyond “any possibility for fairminded disagreement.” Harrington, 562 U.S. at 103. Thus § 2254(d) bars habeas relief, and Valentino’s petition must be dismissed. * * * AEDPA erects a high hurdle that defendants must overcome before a federal court may disrupt a state court judgement. “Under the doubly deferential judicial review that applies to a Strickland claim evaluated under the § 2254(d)(1) standard,” we do not ask whether “‘the state court’s determination’ under the Strickland standard ‘was incorrect but whether that determination was unreasonable—a substantially higher threshold.’” Knowles, 556 U.S. at 123 (quoting Schriro, 550 U.S. at 473). Performing our review as AEDPA requires, we agree with the district court that the state’s post-conviction adjudication was not unreasonable. Therefore, the judgment of the district court is 40 AFFIRMED. 41