with whom O’SCANNLAIN, KLEINFELD, and IKUTA, Circuit Judges, join, dissenting:
Years of consuming forensic science television shows have gone to our heads. We know the plot by heart: the hapless State has charged the wrong guy and our scientists-turned-sleuths will come up with the trial-changing evidence at the last minute. But State v. Richter isn’t the pilot for CSI: Sacramento. Real trials are rarely as gratifyingly formulaic as those seen on TV, and real defense attorneys can seldom boast the Holmesian intuition imputed to them by savvy scriptwriters. In the real world, defense attorneys must often contend with an unsympathetic bench, financial and temporal pressures, and unexpected evidentiary developments. They must also sometimes decide between various unappealing defense strategies. When we ignore these gritty realities and do not adequately analyze the specific circumstances surrounding an attorney’s performance, we inevitably fail to heed the Supreme Court’s admonition about second-guessing trial counsel. See Strickland v. Washington, 466 U.S. 668, 689 (1984).
The majority opinion is a model of the intrusive post-trial inquiry into attorney performance long rejected by the Court. It declares the blood evidence to be “the single most critical issue in the ease,” one that “any competent defense counsel would have immediately recognized.” Maj. Op. at 953. This would have come as a surprise to both the State — which had focused on other physical evidence and had not even taken a blood sample from the victim before trial — and Richter’s counsel, who knew the limited parameters of the State’s evidence. What the majority is now convinced was the “central dispute” in the case, Maj. Op. at 953, sent both sides scrambling mid-trial. The majority now proclaims that no competent counsel could fail to understand the importance of acquiring expert witnesses to testify on the blood evidence, ignoring both the circumstances of the case and counsel’s own deposition testimony establishing why he had no reason to believe that such experts would be helpful. As counsel explained, the State had built its case on other evidence and had no witnesses on its list to testify about blood evidence. More importantly, if counsel had conducted the pretrial investigation demanded by the majority, it not only would have alerted the State to counsel’s defense but might well have resulted in the discovery of additional inculpatory evidence. “Strickland does not require counsel to investigate every conceivable line of mitigating evidence no matter how unlikely the effort would be to assist the defendant,” Wiggins v. Smith, 539 U.S. 510, 533, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003), or if counsel has “reason to believe that certain investigations would be fruitless or even harmful,” Strickland, 466 U.S. at 691, 104 S.Ct. 2052. Had counsel followed the majority’s recipe, he might have cooked his own client. Counsel had no way of knowing which way the blood evidence might turn, and he had very good reasons for suspecting that it would not favor Richter. So, his strategy was to address the other physical evidence the State was presenting and then fault the State for its own failure to investigate. The Court has recognized that this is a perfectly reasonable strategy. Id.
*970Far from being “nothing novel,” Maj. Op. at 947, the majority’s holding plainly enlarges the constitutional duty of defense counsel to seek out and present expert testimony. The majority’s single-minded approach faults counsel for failing to consult with forensic experts (1) before he developed his trial strategy, Maj. Op. at 955-56; (2) before trial once he selected a strategy, Maj. Op. at 956-59; and (3) during the trial, Maj. Op. at 959. Our decision will force counsel to seek expert advice at every stage of the proceedings, even when counsel believes that it will not be helpful and will detract from the other issues counsel must confront. Furthermore, counsel must now consult an expert on every conceivable evidentiary issue, because our ability to second-guess counsel’s decisions knows no bounds. The majority, for example, criticizes counsel for failing to consult two serologists, a forensic pathologist, and a bloodstain pattern expert in part because it was possible that the State might introduce similar experts at trial.
Unlike the cases where the Supreme Court has found Strickland error, this is not a case where counsel so neglected his client that “counsel was not functioning as ‘counsel.’ ” Strickland, 466 U.S. at 687, 104 S.Ct. 2052. Throughout the proceedings, Richter’s counsel was conscientious and well-prepared. When the State came up with surprise witnesses to address the blood evidence, counsel diligently prepared for cross-examination and considered whether, in addition, he should retain an expert. He ultimately decided against it, a judgment that majority is more than willing to second-guess. It is not clear, even today, what counsel was supposed to discover. Indeed, nearly fourteen years after the trial, Richter cannot point to any new evidence that affirmatively supports his innocence.
In sum, the majority holds today that defense counsel was required to adopt a different trial strategy — the majority’s scorched-earth investigative strategy — and assume new burdens that this new strategy creates. Both counsel’s original strategy and the majority’s new tactics have some merit, and neither can solve all Richter’s problems. But the only reason the majority’s strategy prevails over counsel’s is that it gets to go last, and counsel’s strategy, having proven unsuccessful, can be second-guessed.
I respectfully dissent.
I
Mark Twain once advised never to let the facts get in the way of a good story. The Supreme Court, however, has made apparent that such cannot be the case in determining whether a trial attorney has unconstitutionally deprived a habeas petitioner, through ineffective performance, of his Sixth Amendment right to counsel. Rather, applying the standards of Strickland requires a “case-by-case examination of the evidence.” Williams v. Taylor, 529 U.S. 362, 391, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (quoting Wright v. West, 505 U.S. 277, 308, 112 S.Ct. 2482, 120 L.Ed.2d 225 (1992) (Kennedy, J., concurring in the judgment)). Because I believe the majority’s story lacks this requisite substance, I recount the facts and the case from “counsel’s perspective at the time” before expressing my opinion that defense counsel’s performance was reasonable “in light of all the circumstances.” Strickland, 466 U.S. at 689-90.
A
From all appearances this was an open- and-shut robbery/homicide. After receiving a 911 call on the morning of December 20,1994, Sacramento County Deputy Sheriff (and detective) Michael Wright, and his partner, Deputy Steve Larson, arrived at *971Joshua “Gunner” Johnson’s house at 5:06 am. Wright entered the house to find Johnson with blood covering his cheeks, shirt, hands, and right shoulder. He saw a man later identified as Patrick Klein lying unconscious on the living room couch, on top of a sleeping bag, with what appeared to be a bullet wound to his left eye. After calling for medical assistance, Wright talked with Johnson, who claimed that he had been attacked earlier that morning.
Both Johnson and Klein were transported to a nearby hospital, where Klein died from his wounds. At the hospital, the attending physician discovered that Johnson had been shot twice: once in his cheek, where the bullet severed an artery and had lodged near his spine; and once in his shoulder. The physician noted that Klein had also received two wounds: a “fairly superficial” wound to his neck, and the fatal wound to his left eye. Both bullets that struck Klein were recovered from his body — the one that entered his neck was determined to be a .22 caliber, and the one that pierced his eye was a .32 caliber. The one bullet recovered from Johnson also was determined to be a .32 caliber.
While at the hospital, Johnson identified the perpetrators as Joshua Richter and Christian Branscombe. Johnson told police that he remembered waking up shortly before 5:00 am to see Richter and Branscombe in his bedroom attempting to steal his gun safe. He sat up, but was immediately shot by Branscombe. After hearing one or more gunshots in the living room, he stood up and noticed that his Cobray M-12 pistol (usually kept on his nightstand) was missing, along with his gun safe and a hip sack containing a pager and around $6000 in cash (proceeds from his daily drug sales). After stumbling into the living room to find Klein bleeding to death on the sofa, Johnson locked the door and called 911. He then ran into the backyard to dispose of several marijuana plants and other drug paraphernalia before the police arrived.
The evidence discovered at Johnson’s house largely corroborated Johnson’s recollections. About two hours after the 911 call, the police department dispatched homicide detectives to Johnson’s residence to analyze the crime scene. Lead detective Robert Bell discovered two expended .32 caliber shell casings in Johnson’s bedroom where Johnson claimed to have been shot by Branscombe. The detectives located Johnson’s pager lying in the front yard by the porch steps — consistent with Johnson’s story that Richter and Branscombe had stolen his hip sack. The detectives uncovered an expended .32 caliber shell casing and an expended .22 caliber shell casing in the living room, resting on a pile of clothing: these confirmed Johnson’s recollection that he heard at least one more shot fired in the living room, and matched the bullets that killed Klein.
Blood was found throughout the house: large blood pools were discovered in the kitchen and in the bedroom doorway; there was a concentration of blood at the end of the couch where Klein’s head had been resting; there was blood spatter in the bedroom and near the couch in the living room; and there were blood drops throughout the house. The detectives collected a few blood samples, one from a splash of blood on the molding above the blood pool in the bedroom doorway. Because Detective Bell ascertained that the forensic evidence sufficiently supported Johnson’s story, no detailed blood or blood spatter analysis was performed at that time. The detectives, however, documented the entire crime scene with a camera and a video recorder.
Based on this evidence, the police department acquired a warrant to search the garage in which Richter had been living. The evidence found in the ensuing search further corroborated Johnson’s story and *972directly implicated Richter in the crime. In the garage, investigators discovered (among other things) Johnson’s gun safe lying haphazardly on its back, two boxes and a drawer filled with CCI Stinger .22 caliber cartridges, and a magazine loaded with live CCI Stinger cartridges. A ballistics expert studied the .22 caliber bullet recovered from Klein’s body and concluded that it was an exact match to the CCI Stingers found in the garage. The expert also connected the spent .22 caliber casing in the living room to the magazine found in the garage: after analyzing the casing, the expert determined that it had been ejected from a High Standard Sport King handgun — the expert then examined Richter’s magazine and concluded that it was a typical High Standard magazine designed specifically for a Sport King.
A warrant was issued for the arrest of Richter and Branscombe, and both were apprehended the following day. Although Richter first denied having any involvement in the crime and told the arresting officer that his truck was never at Johnson’s house on the night in question, he eventually admitted that he and Branscombe had been at Johnson’s house early on the morning of December 20th. He also admitted that Branscombe used a .32 caliber weapon to shoot Johnson twice and Klein once. Richter claimed, however, that he had nothing to do with the shootings — he said that he had stayed in his truck while Branscombe went into the house to return clothing belonging to Johnson’s roommate. Richter asserted that he heard shots and screams in the house and entered to find Branscombe “freaked out” claiming that he had only shot Johnson and Klein after they attacked him.
Richter also confessed that he and Branscombe had taken Johnson’s Cobray M-12 .380 caliber handgun with them to the Yolo overpass after the shootings, and had discarded it there along with Branscombe’s .32 caliber weapon. An initial search of that area did not reveal either weapon. Detective Bell, however, did find a crisp $100 bill a few hundred yards away from the spot Richter had designated. One day later, Richter’s counsel went back to the area and returned with the Cobray M-12, which he surrendered to the police.
Two additional statements implicated Richter. At the police station, Branscombe and Richter were interviewed regarding their involvement in the events in question. Both suspects were videotaped in a holding cell during a break in the interviews, and the recording revealed a statement by Branscombe implicating Richter in the crime. Branscombe asked Richter, “Did you tell them anything?,” and Richter responded, “I just told them that I did not kill anyone, and da da da.” Branscombe replied, “Joshua, we were going to tell them the truth.” In the meantime, the police also interviewed Lauren Sullivan, Richter’s girlfriend. Although she first refused to talk about the events in question, she eventually admitted that Richter had talked with her on the night of December 20th, approximately twelve or thirteen hours after the shootings. Sullivan never protested Richter’s guilt at that time — instead, she told police ominously that she didn’t think she should tell them about what Richter had told her. Richter’s counsel knew of both statements.1
*973B
On December 23, 1994, Richter was charged with murder (under Cal.Penal Code § 187) in the first degree (Cal.Penal Code § 190.2(a)(17)), attempted murder (Cal.Penal Code §§ 187, 664), burglary (CalPenal Code § 459), and robbery (Cal.Penal Code § 211). The trial began almost a year later, on November 14,1995, and lasted nineteen days. Richter and Branscombe were tried jointly — Branscombe as the primary perpetrator, and Richter as a knowing accomplice. The jury was instructed on felony murder. Thus, as long as the jury believed that Branscombe shot Johnson and Klein in the perpetration of a robbery or burglary, and that Richter was an accomplice, Richter was also liable for murder. See generally People v. Seaton, 26 Cal.4th 598, 110 Cal.Rptr.2d 441, 28 P.3d 175 (2001); People v. Pulido, 15 Cal.4th 713, 63 Cal.Rptr.2d 625, 936 P.2d 1235 (1997).
Defense counsel faced enormous evidentiary obstacles at trial. Every known piece of forensic evidence either directly supported Johnson’s account of the shootings or was inconclusive. The .22 caliber CCI stinger bullets discovered in his client’s garage decisively matched the bullet recovered from Klein’s dead body. The $100 bill in the Yolo overpass, the positioning of Johnson’s pager in the front yard, and the presence of Johnson’s gun safe in his client’s garage pointed strongly to his client’s guilt. Moreover, his client’s behavior after the shootings was hardly exemplary — Richter had fled the scene of the crime, thrown away evidence that might have implicated him in the murder, and upon arrest first denied ever being at the scene of the crime only to later recant this implausible story. There were even videotaped inculpatory statements that counsel would be forced to attempt to exclude from evidence. The only person who could cast doubt on his client’s guilt was his client’s girlfriend, who had previously told police that she had not talked with his client until approximately thirteen hours after the events in question, and thus could not credibly confirm Richter’s whereabouts during the relevant time period. Finally, defense counsel knew that his client’s co-defendant would not be able to testify credibly at trial — after observing Branscombe’s odd behavior during an interview, counsel even suggested to Branscombe’s attorney that an insanity defense might be productive.2
Counsel also faced significant procedural obstacles. The trial was somewhat of a hurried affair due to problems Branscombe’s attorney had maintaining his bar membership. Because of past transgressions, Branscombe’s attorney’s bar membership was about to be suspended, and he wanted to finish the trial and collect his fee before the end of the year. Richter’s counsel believed that the judge, because of this predicament, created an accelerated trial schedule that upset his plans to talk with the prosecutor and perhaps obtain a plea bargain.
*974Nevertheless, due to the State’s perfunctory investigation of the crime scene, counsel knew that the existing forensic evidence did not necessarily prove his client’s guilt beyond a reasonable doubt. Counsel knew that the homicide investigators had taken only six swabs of blood at the crime scene, had failed to do a comprehensive blood spatter analysis, and had never tested any of the blood samples they had gathered in the year prior to trial. Counsel also knew that police had never discovered the High Standard Sport King .22 caliber handgun that the ballistics analysis fingered as one of the murder weapons. Additionally, counsel knew that Johnson was not a victim likely to elicit empathy— he was a known drug dealer who had cut a deal with the State to testify against Richter in return for immunity from certain drug charges.
Understanding these limitations in the State’s case, defense counsel diligently investigated evidence that could potentially exonerate his client. He hired a private investigator, James Pihl, who worked with Branseombe’s attorney’s private investigator, John Lee, to discover evidence not collected by the State. Counsel also interviewed a number of people who he had reason to believe might shed doubt on Johnson’s story. Although this extensive investigation did not reveal much helpful evidence, counsel’s investigator was able to discover that there were two bullet holes in Johnson’s bedroom that were never analyzed by the police — one in a floorboard, and another in a wall. Counsel diligently notified the police investigators of the existence of these holes, which the State thereafter investigated. After examining the photographs and video of the crime scene, counsel was also able to pick out some anomalies in the State’s case: for example, although Johnson claimed it was dark when he was shot, counsel noticed there was no blood on the living room light switch indicating that the wounded Johnson turned it on after being shot.
C
In the State’s opening statement, the prosecutor mentioned nothing about serological, pathological, or blood spatter evidence that corroborated Johnson’s story. There was good reason for this omission: the homicide investigators had neither collected much of the blood at the crime scene (they had taken only six swabs of blood) nor analyzed it in the year prior to trial. No blood sample was obtained from the large pool of blood in the bedroom doorway. In fact, the sample that is the focus of Richter’s habeas petition was removed from the door molding a few inches above that blood pool. Before trial, the State had not even taken blood from Johnson with which to compare the blood stains in the house. Accordingly, the prosecutor focused on other physical evidence: the shell casings, the location of Johnson’s gun safe, and the CCI Stinger cartridges found at Richter’s residence.
Defense counsel’s opening statement moved to exploit the State’s evidentiary lapses. Counsel proclaimed that “evidence will ... show that the subsequent investigation of the scene was quickly done and poorly orchestrated.” He told the jury that “[t]he evidence of great importance is the lack of spatter marks [on the living room couch], and other pools of blood, and spatter marks found elsewhere in the house inconsistent with stories told by Gunner Johnson.” Counsel explained that the crime scene photographs revealed that “there are no spatter marks of blood or anything on the couch” and that “there are also no marks on the sleeping bag [on which Klein was lying].” Counsel faulted the investigating officers for accepting Johnson’s theory of the story and failing to “take swabs of the various blood marks *975that were everywhere in that house and the pools of blood, including a pool of blood in the bedroom near the doorway of the bedroom, to determine whose blood that was, Mr. Johnson, Mr. Klein’s, or another part[y’s].”
Defense counsel framed a story “very, very different” from the prosecutor’s account. He told the jury that Johnson was the aggressor, that Klein assisted him, and that Branscombe shot and killed the two in self-defense. Richter’s counsel admitted that Branscombe had a .32 that evening, and that he used it. But Klein had also been shot with the .22 semiautomatic pistol that was never recovered. Counsel suggested that the .22 was Johnson’s: his investigator had discovered a .22-caliber-sized hole in Johnson’s bedroom floor that Johnson said he had made on another occasion. Counsel again returned to the State’s sloppy investigation — there were .22 casings and shells stored in Johnson’s bedroom that the State never tested to see if they were connected with the bullet that killed Klein.
According to defense counsel’s theory, Richter and Branscombe, who had smoked marijuana with Johnson until around 2:00 am on the morning in question, returned to Johnson’s house around 4:00 am to drop off some clothing and money for one of Johnson’s roommates. Richter sat in his truck while Branscombe went up to the house. Counsel asserted that Klein let Branscombe into the house, but then immediately attacked him — grabbing Branscombe in the bedroom doorway while Johnson pulled out his M-12.3 Luckily for Branscombe, Johnson’s gun was not reliable and he only got off one shot before the gun jammed. But Johnson wasn’t nicknamed “Gunner” for no reason — according to defense counsel, Johnson immediately pulled out a .22 caliber semi-automatic handgun to finish the job. Fortunately again for Branscombe, Johnson’s talent with firearms failed to match his enthusiasm — he missed Branscombe and wounded Klein with the .22. This was just the advantage Branscombe needed. Branscombe pulled out his .32, gunned down Johnson, and then finished off Klein with a shot to the head. Richt*976er, hearing the gunshots, ran into the house to find Klein lying in a pool of blood in the bedroom doorway, Johnson twisted unnaturally on the bed, and Branscombe, “totally freaked out,” standing in Johnson’s bedroom holding a smoking .32. In a panic, Richter and Branscombe fled the scene to dispose of Johnson’s M-12 and Branscombe’s .32 at the deserted Yolo overpass.
Defense counsel argued that none of the evidence found by the police was inconsistent with Richter’s story. Counsel asserted that Johnson secretly owned a .22 caliber semi-automatic handgun from which he could have shot the CCI Stinger, that Richter and Klein had just been paid for their work at a Christmas tree lot and lost the $100 bill while disposing of the guns, and most importantly that Johnson had free run of the house for at least forty minutes after the shootings and had manipulated any other evidence not consistent with Richter’s story. Counsel avowed, for example, that Johnson had moved Klein from the bedroom doorway to the couch in order to make his story sound better. Finally, counsel claimed that Johnson’s gun safe had been at Richter’s residence for several months and that Johnson had invented this theft to implicate Richter.4
D
Counsel must have struck a nerve because the State moved quickly to shore up its case. The prosecutor, who realized for the first time that counsel would be articulating a theory of self-defense,5 hurried to have the blood evidence analyzed to determine if it could be inconsistent with that theory. On the fourth day of trial, without prior warning, the prosecutor presented Bell, one of the homicide detectives, as an expert in blood spatter. The following day, the prosecutor for the first time obtained a blood sample from Johnson and had it compared with the six samples collected from the scene. Four days later, on the ninth day of trial, the prosecutor put Jill Spriggs, a criminalist, on the stand to testify about the results of that analysis. Defense counsel had been given notice of Spriggs’s testimony only a day earlier.
The defense began presenting its case shortly thereafter. Counsel presented seven witnesses, including Richter. These witnesses focused primarily on the timing of the shooting and the location of the gun safe — they provided testimony indicating that the events in question occurred around 4:20 am, rather than 5:00 am, and argued that Johnson had been storing his gun safe at Richter’s residence. Counsel did not call any expert witnesses to the stand. Although he considered doing so at certain points in the trial, counsel eventually decided that his cross-examination of the State’s experts was sufficient to create reasonable doubt about the truth of Johnson’s claims. The adequacy of this decision is the primary thrust of Richter’s habeas petition.
II
In determining whether counsel’s tactical decisions were so erroneous as to violate Richter’s constitutional right to effective assistance of counsel, we face a truly *977daunting standard of review. Under Strickland, a habeas petitioner may only prevail on a claim of ineffective assistance of counsel if he can show deficient representation “so serious that counsel was not functioning as the ‘counsel’ guaranteed the [petitioner] by the Sixth Amendment.” Strickland, 466 U.S. at 687, 104 S.Ct. 2052. The Supreme Court in Strickland refused to create a monolithic standard of acceptable attorney performance — instead, it recognized that courts must “indulge a strong presumption that a counsel’s conduct falls within the ivide range of reasonable professional assistance.” Id. at 689, 104 S.Ct. 2052 (emphasis added). The Court noted that “[tjhere are countless ways to provide effective assistance in any given case” and observed that “[e]ven the best criminal defense attorneys would not defend a particular client in the same way.” Id. Thus, to find that counsel’s conduct was deficient, we must override the “strong[ ] presumption] that counsel’s conduct was within the wide range of reasonable assistance, and that he exercised acceptable professional judgment in all significant decisions made.” Hughes v. Borg, 898 F.2d 695, 702 (9th Cir.1990).
Even if we are persuaded that counsel has engaged in “unprofessional errors,” Strickland is not satisfied unless we can conclude that “the result of the proceeding would have been different.” Strickland, 466 U.S. at 694, 104 S.Ct. 2052. This means “that counsel’s errors [must be] so serious as to deprive the defendant of a fair trial, a trial whose result is reliable.” Id. at 687, 104 S.Ct. 2052. Of course, any newly introduced expert testimony might have changed the prosecution’s approach at trial — but this is not enough to find prejudice under Strickland. See id. at 694, 104 S.Ct. 2052(“It is not enough for the defendant to show that the errors had some conceivable effect on the outcome of the proceeding. ... [or] that the errors ‘impaired the presentation of the defense.’ ”). There is prejudice only if there is a reasonable probability that, but for the errors, the result of the proceedings would have been different.
Overlaying this inquiry is the Antiterrorism and Effective Death Penalty Act (“AEDPA”), 28 U.S.C. § 2254(d)(2). Because this case is an appeal from a state court decision, we cannot simply make the findings required by Strickland and grant the petition. Rather, we must make a further finding: that our Strickland conclusion is itself so clear that the contrary judgment by the state court was unreasonable. In other words, we face a “doubly deferential” standard of review, see Knowles v. Mirzayance, 129 S.Ct. at 1420: we may not grant Richter habeas relief unless we are convinced that the state court decision was “contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States,” or was “based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d). In cases like these, where no state court has explained its reasoning, we must conduct “an independent review of the record to determine whether the state court’s decision was objectively unreasonable.” Sass v. Cal. Bd. of Prison Terms, 461 F.3d 1123, 1127 (9th Cir.2006). “Objectively unreasonable” does not mean that we can grant the habeas petition if we merely disagree with the state court’s decision, or would have decided oppositely in the first instance. See Williams v. Taylor, 529 U.S. 362, 409, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (“[A]n unreasonable application of federal law is different from an incorrect application of federal law.”). Nor can we decide that a state court was unreasonable even if we believe it committed a clear error. Lockyer v. Andrade, 538 U.S. 63, 75, 123 S.Ct. 1166, 155 L.Ed.2d 144 (2003). *978Rather, to overturn the state court’s decision, its judgment about counsel’s performance must be so egregious that we cannot but be convinced that the court’s decision is itself outside the bounds of reasonableness.
Deference under AEDPA is especially important in this case, where Richter’s claims arise under Strickland. The Supreme Court has emphasized that “evaluating whether a rule application was unreasonable [under AEDPA] requires considering the rule’s specificity.” Yarborough v. Alvarado, 541 U.S. 652, 664, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004). “The more general the rule, the more leeway [state] courts have in reaching outcomes in case-by-case determinations.” Id. Strickland is a general rule, as the Supreme Court has often recognized. See, e.g., Mirzayance, 129 S.Ct. at 1420. The Court has made plain that “because the Strickland standard is a general standard, a state court has even more latitude to reasonably determine that a defendant has not satisfied that standard.” Id.
The majority recites these standards, but doesn’t take them to heart. The majority glosses over the circumstances in which defense counsel found himself: all the forensic evidence discovered before trial implicated his client in the crime; his client’s actions in response to the crime were indicative of guilt; he was fighting a judge and fellow defense attorney who wanted to finish the trial as quickly as possible; and he was surprised by expert witness testimony from witnesses the State had not previously identified. The majority also ignores the nature of the expert testimony at issue: despite the bombastic language of Richter’s newfound experts, the content of their testimony is not as revealing as Richter claims.
A careful review of the record demonstrates the inadequacy of the majority’s conclusions. Richter’s counsel vigorously pursued the strategy he had developed in his opening statement and diligently contested all critical evidence raised by the prosecution. In no sense, then, can we conclude that “counsel was not functioning as the ‘counsel’ guaranteed [Richter] by the Sixth Amendment.” Strickland, 466 U.S. at 687, 104 S.Ct. 2052. Even if we could conclude, with the benefit of hindsight, that defense counsel made certain mistakes in his representation of Richter, these mistakes are not the type of egregious errors that mandate a finding of deficient performance. At worst, counsel exhibited questionable judgment by relying exclusively on his own powers of cross-examination in lieu of an expert’s direct testimony. And even if counsel’s judgment was flawed, it is far from clear that counsel’s decision not to call these experts resulted in an unreliable guilty verdict. Although the expert declarations offered in support of Richter’s habeas petition might have added some new tools to the defense’s arsenal, they can hardly dent the clear, unchallenged evidence implicating Richter in Klein’s murder and the utter lack of evidence (even in Richter’s new-found expert testimony) confirming or even strongly supporting Richter’s theory of self-defense.
Finally, even if we could conclude, as a matter of first impression, that defense counsel failed to meet the standard of professional conduct mandated by Strickland, there is clearly not enough evidence in the record to conclude that a state court adjudicating oppositely has made not just an erroneous but a patently unreasonable decision. To rule otherwise is to ignore the clear congressional intent in AEDPA to curtail our second-guessing of state court habeas decisions. See Williams, 529 U.S. at 386, 120 S.Ct. 1495 (Opinion of Stevens, J., joined by Souter, Ginsburg, and Breyer, JJ.) (“[I]t seems clear that Congress in*979tended federal judges to attend with the utmost care to state-court decisions, including all of the reasons supporting their decisions, before concluding that those proceedings were infected by constitutional error sufficiently serious to warrant the issuance of the writ.”).
A
The majority first claims that defense counsel’s decision not to consult with or call as a witness a serologist was deficient and probably prejudiced Richter’s defense. In its haste to paint with broad brush strokes, the majority decides this without even a cursory discussion of the circumstances of that decision or the nature of the relevant evidence. Review of counsel’s decision and the serology evidence in context indicates that counsel’s decision not to “investigate” serology evidence before trial was reasonable and consistent with his trial strategy. Moreover, even if counsel’s failure to procure a serologist constituted deficient performance, it had a nearly negligible effect on the course of trial.
The State had to plug the holes Richter’s counsel identified in its case. It had neglected to test the blood pool in the doorway to Johnson’s bedroom to see whether it was Johnson’s blood (as the State contended6) or Klein’s blood (as the defense claimed at trial7). The only sample the State had was a swab taken from the wall just above the blood pool. At trial, the State’s expert, Jill Spriggs, testified regarding her comparative analysis of Johnson’s blood sample and the swabbed sample. Because both Johnson and Klein had Type 0 blood, their blood samples had to be differentiated by reference to an enzyme, phosphoglucomutase (“PGM”). Spriggs opined that her laboratory testing determined that Johnson’s PGM subtype was 2 +1 + and that Klein’s PGM subtype was 1 +. Spriggs claimed that because the blood found on the molding near the blood pool had a PGM subtype of 2 +1 +, Klein could reliably be excluded as the donor of that blood.
This was the first defense counsel had heard of this evidence. In fact, counsel had learned of the subject matter of Spriggs’s testimony only the day before her appearance in court. Spriggs had not been on the prosecution’s witness list, and her forensic tests were run mid-trial. After learning of Spriggs’s testimony, counsel immediately requested a continuance, which the trial court denied. At that point, counsel had to move quickly to counter this unexpected development. He called the public defender’s office and then spent the evening in a library learning about PGM subtypes while preparing for cross-examination.
During cross-examination, counsel challenged Spriggs’s conclusion that Klein could be reliably excluded as the source of the blood swabbed from the bedroom wall. In response to counsel’s questions, Spriggs acknowledged that she had not tested for cross-contamination and that she had only been able to run one test. Counsel asked Spriggs whether she could testify to a “scientific certainty” that the sample only contained one person’s blood. She answered equivocally that “it does not look like there are two people contributing to that stain.”
*980Counsel also questioned whether a commingling of Johnson’s and Klein’s blood would still produce a 2 + 1+ sample and whether a degraded sample would reliably indicate PGM subtype at all. Spriggs, in response to this line of questions, eventually opined that if commingling between a 2 +1 + sample and a 1 + sample occurred, the test would reveal a stronger 1 + result than what had occurred in this case. Counsel attempted to challenge this conclusion, questioning Spriggs about what would result if the proportion of blood contributed by the two donors was different, but he was cut off by the judge. Nevertheless, counsel was able to secure an admission from Spriggs that if Johnson and Klein’s blood were combined in the sample, the PGM subtype would show a 2 +1 + band, and that the test might not reveal a stronger 1 + result (indicating the presence of Klein’s blood) if the sample was degraded.
Now, to support his habeas petition, Richter attaches the expert opinions of two serologists who make nearly the same points that counsel presented on cross-examination.8 One expert, John Thornton, speculates that because “in the electrophoretic separation of PGM subtypes, the 1 + band in a 2 + 1+ sample appears at the same location as the 1+ band in a 1 + sample,” the Spriggs finding “could not exclude the possibility that the sample contains a mixture of 2 +1 + and 1 + blood.” The other expert, Brian Wraxall, remarks that there is a possibility that Klein could have been a contributing source because “the relative intensity between the 2 + band and the 1 + band in the [wall] sample would depend on the proportion of Mr. Klein’s blood commingled with Mr. Johnson’s blood.” Also, Wraxall points out that even a non-degraded sample might show a 2 + 1+ result of uniform intensity where Johnson’s enzyme tested as a strong 2 + band and a weaker 1 + band. Wraxall had to qualify this opinion because “the 2 + band in a 2 +1 + band is usually, hut not always, equal in intensity to the 1 + band” (emphasis added) and no evidence at trial indicated whether Johnson’s blood would create a stronger 2 + band.
Richter’s professed experts do not challenge Spriggs’s conclusion that the blood pool contained Johnson’s blood. They also cannot establish that Klein’s blood is in fact in the sample, or even that it is likely that Klein’s blood is in the sample. At best they can only challenge the conclusion that evidence shows conclusively that the swabbed sample contained Johnson’s blood only — a conclusion that Richter’s counsel effectively challenged at trial during cross-examination.
1
With all due respect to our colleagues, counsel’s performance was well within the “wide range of reasonable professional assistance.” Strickland, 466 U.S. at 689, 104 S.Ct. 2052. In no way was Richter deprived of competent counsel in the sense that “counsel was not functioning as the ‘counsel’ guaranteed the defendant by the Sixth Amendment.” Id. at 687, 104 S.Ct. 2052.
Counsel did what he could under extremely difficult circumstances. While the majority excoriates defense counsel for failing to “consult” with a serologist or other forensic expert in preparation for trial, it fails to acknowledge that the results of such “consultation” are pure speculation and have no basis in the record. *981Although the Certificate of Appealability in this case listed “[w]hether trial counsel rendered ineffective assistance in violation of the Sixth Amendment by failing to investigate, develop, and present [expert testimony]” as one of the issues to be considered on appeal, both the habeas petition filed in the district court and the evidence presented in support of that petition focus entirely on counsel’s “fail[ure] to present readily available expert testimony which directly supported his theory of the case.” (emphasis added). The expert witness declarations proffered in support of the petition do not indicate that any of Richter’s new experts would have been available to advise counsel about his trial strategy. In each declaration, the expert only avers that he “would have been available to testify to this if called as a witness at Mr. Richter’s trial.” (emphasis added). Not one expert declaration asserts a witness’s availability before trial to “consult” with defense counsel.
More importantly, nothing in the record hints that defense counsel would have been better prepared to address serology testimony at trial had he retained a serologist at the outset. Both of Richter’s serology declarations are reactive — they respond to specific points made by the State’s expert during trial. Neither declaration details any type of advice the serologists would have given defense counsel before trial, when neither counsel nor anyone else knew that the State would be presenting expert testimony on blood evidence.9 Indeed, before trial neither counsel nor the State knew if the State’s limited blood evidence was still viable — it had been collected over a year in advance and may not have been well preserved. Thus, prior to trial, a serologist could have given defense counsel only general background knowledge about serology: no serologist could have “prepared” defense counsel to address the specific weaknesses in the State’s case, which only were revealed after Spriggs’s testimony.
Likewise, it is absurd to suggest, as the majority does, that counsel would have altered his trial strategy after consulting with a serologist before trial. Counsel had good reasons to refrain from having the State’s blood samples tested, and these reasons would not have been challenged by an advisory serologist. Counsel knew that a major weakness in the State’s case was the lack of serological evidence eliminating Klein as a potential source of blood pools in the bedroom and the bedroom doorway, and he knew that the homicide detectives had never taken a sample directly from the blood pool in the bedroom doorway or tested the blood samples they did possess. Given these circumstances, counsel had two options before trial: he could ask the State to do a comparative analysis of the blood samples, or he could wait and see whether the State planned on testing the samples itself. An analysis of the possible *982outcomes of each strategy demonstrates that counsel’s decision to wait on the State’s investigation was actually quite shrewd.
If counsel had “investigated” and asked the State to test the blood samples, he very well could have (and in this case would have) aided the State in producing inculpatory evidence against his client. There were only three potential outcomes to a serology test, and only one of them would have been helpful to his client. If the test showed that the blood sample matched Johnson’s blood, counsel’s self-defense theory would have collapsed. If the test showed a match with Klein’s blood, counsel’s case would have been strengthened. Finally, if the test was inconclusive, counsel was no better off than he would have been had the test never been attempted. All the evidence counsel had at the time indicated that it was highly unlikely that a test would result in the second outcome. Not one piece of forensic evidence discovered at the crime scene corroborated counsel’s self-defense theory, and counsel had no reason to think that the blood evidence would be any different — in fact, counsel knew that Richter and Branscombe had made videotaped inculpatory statements while in police custody.10
Given the low probability that a serology test would aid his client, and the high probability that it might foreclose entirely his theory of self-defense, it was quite reasonable for counsel to forgo such testing and force the State to analyze the blood on its own initiative. In this way, counsel would gain the same benefits of a positive test and yet also retain more maneuverability at trial. If the State didn’t bother to test the blood, or found that the samples were too degraded to create a conclusive result, counsel could ridicule the State for its inadequate investigatory processes and imply that other forensic evidence at the scene was not conclusive. Moreover, if the State decided independently to test the blood samples and found potentially exculpatory evidence, it would be required by law to disclose the result to defense counsel. See Brady v. Maryland, 373 U.S. 83, 87, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963). Of course, the State might test the blood and find an adverse result— but in that case defense counsel would be no worse off than were he to have requested the test himself.
To be sure, counsel’s decision to rely on the State’s failure to present expert evidence was in some aspects riskier than the majority’s preferred strategy. If the blood evidence was exculpatory, his decision not to test the samples could have prevented the jury from having such evidence to consider. But the existence of this remote possibility is no reason to deride counsel’s decision as unreasonable. Counsel, who could talk frankly with his client and discuss whether “pursuing certain investigations would be fruitless and *983even harmful,” knew more about the risks involved with either strategy than any judges on this court. See Strickland, 466 U.S. at 691, 104 S.Ct. 2052 (“Counsel’s actions are usually based, quite properly, on informed strategic choices made by the defendant and on information supplied by the defendant.”). Thus, at best the majority can claim that counsel’s decision to wait for the State’s blood analysis was riskier than its preferred strategy — and the Constitution, unlike the majority, does not mandate risk-averse representation. See id. at 689, 104 S.Ct. 2052 (“Any such set of rules [restricting the range of legitimate decisions regarding how best to present a criminal defense] would interfere with the constitutionally protected independence of counsel and restrict the wide latitude counsel must have in making tactical decisions.”).
Counsel’s actions during trial, in response to the State’s expert testimony, were also quite reasonable. Although counsel might have retained a standby serologist to “consult” with during trial (as the majority strangely holds was obligatory), he had a list of the State’s witnesses and knew the State was not putting on any evidence about the blood pool. After learning that the State was adding a witness, counsel first sought a continuance. When the court refused, he spent time researching the subject matter to determine the evidence’s strengths and weaknesses. Even with his limited research, counsel’s questions on cross-examination were thoughtful, informed, and probing. There is no evidence of inadequate preparation or other professional misconduct in his cross-examination. Counsel realized that he was not as competent as a hypothetical expert to examine the serology evidence, and prepared a list of potential expert witnesses to contact. After his cross-examination of Spriggs, however, counsel believed that he had created the possibility of reasonable doubt so as to preclude the necessity of tracking down his own expert witness (who he had no reason to believe at the time would interpret the serology evidence any differently than the State’s expert 11).
In hindsight, the decision not to obtain a serologist even at this late stage in the trial might have been a tactical error. The prosecutor ridiculed counsel’s decision in his closing argument, proclaiming, “I am not going to worry about Jill Spriggs because, hey, her seven years as a biochemist and a criminalist ... doesn’t mean anything, because I am a lawyer .... But I am not going to pay and bring in an expert to show you.” Also, counsel missed two potential weaknesses in Spriggs’s testimony- — he failed to question her sufficiently about whether a sample containing different proportions of Johnson and Klein’s blood could produce a 2 +1 + result without a stronger 1 + band, and did not raise the possibility that Johnson’s blood might naturally produce a stronger 2 + band and a weaker 1 + band such that a combination of Johnson and Klein’s blood would create a uniform 2 +1 + result.12
Such errors may be obvious to us, but only because we know to a “scientific certainty” that Richter’s defense failed. *984Counsel had no such commodious perspective on his case. And the standard by which the majority is judging counsel in this case is so demanding that in the future no counsel, having failed to defend his client successfully in a contested trial, will be immune from our withering scrutiny.
We have previously acknowledged that the failure to call an expert does not always constitute deficient performance for purposes of Strickland. In Duncan v. Ornoski, 528 F.3d 1222 (9th Cir.2008), we recognized that “it may not be necessary in every instance to consult with or present the testimony of an expert.” 528 F.3d 1222, 1235. We concluded there that because the serology evidence was “pivotal” and because counsel clearly demonstrated a lack of expertise in serology, it was deficient for him to fail to consult with an serology expert and to rest on his cross examination of the State’s expert. Counsel in this case, however, behaved far more responsibly than counsel in Duncan. In Duncan, counsel “demonstrated his lack of expertise in serology at the outset of his cross-examination of the State’s serology expert when he told him, ‘You lost me.’ ” Id. Indeed, counsel in that case “did not even know what serology was,” and began questioning the State’s serologist about hair evidence. Id. at 1235-36.
Here, unlike in Duncan, counsel was inexperienced with blood analyses but not uninformed. He spent an entire evening studying the topic before his cross-examination and asked probing and relevant questions of the State’s expert. After his cross-examination of the State’s expert, counsel made an informed decision that a serology expert would not add significantly to his case, which from the outset focused on holes in the State’s investigation. Even nearly fourteen years after the trial, during his deposition in this case, counsel was able to talk intelligently about the serology evidence and its flaws. Moreover, the experts procured by Richter in support of his habeas petition make the same type of objections to the State’s expert testimony that counsel made on cross-examination— all deal with hypothetical situations in which a 2 + 1+ result might not conclusively exclude Klein as a potential donor to the blood sample.
Also, unlike in Duncan, we are not engaged in de novo review of the state court’s decision. Duncan arose under the pre-AEDPA regime, thus, there we “review[ed] the district court’s decision to deny habeas relief,” and petitioner’s “ineffective assistance of counsel claims,” de novo. 528 F.3d at 1232-33. Under AED-PA, we cannot simply compare this case to the facts of Duncan to determine whether to grant Richter’s petition — we must determine whether the California courts’ decisions about the serology evidence were unreasonable in light of Supreme Court holdings.
The majority stretches the general rule of Strickland beyond any case the Supreme Court has decided. If counsel in this case can be said to have unreasonably rendered deficient performance in failing to use a serologist at trial, almost no counsel in the future can satisfy Strickland without first consulting with every conceivably relevant expert before trial and then calling these experts at trial to testify on any contested forensic issue. No matter' what counsel thinks about the efficiency of paying expert witness fees to duplicate his own research on a topic, he is remiss to fail to pay them. Time pressure, financial pressure, trial strategy — it doesn’t matter: the majority essentially rules, “even when not in doubt, call an expert.” 13
*9852
Although defense counsel did not perform deficiently in deciding not to call an expert — to, as counsel put it, “sprinkle more holy water” on points made during cross-examination — even assuming deficient performance, I doubt that prejudice resulted. As the Sixth Circuit has held, “the modest difference between the jury hearing [a] theory of defense through cross-examination and hearing it through the mouth of another expert” is not enough to establish prejudice under Strickland. Tinsley v. Million, 399 F.3d 796, 806 (6th Cir.2005). Although perhaps an expert witness could have articulated the flaws in the prosecution’s serology evidence more eloquently and provided an opportunity for the defense to bolster its theory through repetition, it is clear that more is required to find that there is a “reasonable probability that ... the result of the proceeding would have been different.” Strickland, 466 U.S. at 694, 104 S.Ct. 2052.
There is no reason to think that the defense would have been able to present better expert witness testimony at trial than the expert affidavits filed in support of Richter’s habeas petition. Yet neither of these expert declarations presents significantly better objections than those articulated by defense counsel on cross-examination. Any expert testimony would have provided little new information to the jury — instead, it would have simply reinforced counsel’s cross-examination, indicating that the 2 +1 + result did not preclude a conclusion that the blood sample was a mixture of Johnson and Klein’s blood, and therefore there was a “possibility” that Klein’s blood was also in the sample. As it was, Richter’s counsel argued that Spriggs did not “demonstrate actual knowledge about what really goes on in there when there is degradation, a whole year[’]s worth of sample.” He added, “Nor did she have particular knowledge of what happened when two kinds of bloods are combined.” The majority cannot (and thus does not attempt to) explain why we should assume that the jury would have credited an expert’s enunciation of that theory when it obviously rejected defense counsel’s.
The majority simply misses the point when it claims that because Spriggs “adhered strongly” to her hypothesis that the blood came only from Johnson, Maj. Op. at 967 n.21, the lack of Richter’s newfound serology testimony was prejudicial. The State’s expert would most certainly not have broken down on the stand and dramatically admitted error even if defense counsel had introduced directly contradictory expert testimony of his own. The question thus is not whether the expert testimony would have “cast” the State’s expert testimony “into doubt,” Maj. Op. at 968 — defense counsel’s cross examination did that. The question is whether the expert testimony would have added appreciably to the defense’s attempts to undermine the State’s expert. Besides noting that the testimony came from an expert, rather than a lawyer, the majority has not put forward any evidence to support its answer of this question in the affirmative.
Moreover, the expert testimony at issue, even if introduced at trial, is not conclusive enough to significantly weaken the prosecution’s theory. The proffered testimony says only that it is possible that the blood *986sample was a mixture of the two victims’ blood — it does not indicate the probability that such a mixture would create the laboratory result in this case. At best, the declarations contest Spriggs’s conclusion that the blood sample was exclusively Johnson’s. The declarations are carefully worded to say that Spriggs’s results cannot exclude Klein as an additional source of the blood pool. The declarations do not support the theory the defense would have liked to have shown: that Klein was the source of the blood pool. For that conclusion, the PGM subtype of the blood sample would have had to have been a 1 +, and it was not. Thus, the totality of the serology evidence strongly supports the State’s theory, that the blood pool was Johnson’s. All that Richter can claim is that there is a possibility that the blood pool contained both Johnson’s and Klein’s blood.
Unfortunately for Richter, neither expert cites any evidence to support a conclusion that these theoretical possibilities are at all probable in this case. As the panel found, “[bjecause these expert reports do not foreclose the likelihood that the blood from the blood sample came exclusively from Johnson, they do not impeach Johnson’s testimony that the blood came from him alone.” Richter v. Hickman, 521 F.3d 1222, 1231 (9th Cir.2008). Especially given the significant amount of other forensic evidence supporting the prosecution’s theory, and the general implausibility of Richter’s story, see infra Part II.D, it is absurd to assume that evidence about the “possibility” of blood mixing would have convinced the jury that Richter’s recollection of the events of December 20th was nearly as compelling as Johnson’s.
3
Once we put the AEDPA overlay on top of Strickland, our task becomes even easier. Even if I am wrong, and counsel’s performance was both objectively deficient and prejudicial, the matter is surely open to debate and discussion. If so, the California courts were not unreasonable to reject Richter’s claims of ineffective assistance of counsel. Nothing in Strickland or any other decision of the Supreme Court should have alerted the California courts that they were making a colossal error in judgment, one on which reasonable jurists could not disagree. See Mirzayance, 129 S.Ct. at 1420.
B
For similar reasons, defense counsel’s decision not to consult a forensic pathologist was clearly not prejudicial. Richter’s habeas petition contains the declaration of a pathologist, Dr. Paul Herrmann, who would have been able to testify that the prosecution’s theory about the creation of the blood pool in the bedroom doorway was “scientifically unreliable.” The majority asserts that this testimony, in combination with that of the other expert witnesses, would have seriously undermined the State’s case. A closer investigation of the pathologist’s hypothetical testimony, however, reveals the fallacy of this conclusion. Indeed, much of the pathologist’s testimony is pure speculation not based on his specialized knowledge or experience and could have been excluded at trial for that reason; his only “expert” analysis does not preclude the State’s theory; and all of his observations merely parrot defense counsel’s statements at trial.
Many of the assertions in Herrmann’s declaration could have been properly excluded by the trial court as having no evidentiary value. See People v. Richardson, 43 Cal.4th 959, 77 Cal.Rptr.3d 163, 183 P.3d 1146, 1179 (2008). Herrmann opines that because nothing in Johnson’s testimony indicates that he was standing in the bedroom doorway for an amount of time necessary to create a pool of blood of the necessary magnitude, the prosecution’s *987theory to the contrary is unreliable. Additionally, Herrman concludes that the treating physician’s testimony about Johnson’s shoulder wound “does not support the State’s theory because the amount of blood likely to result from such a wound would have been minimal.” Notably, Herrmann does not specify the scientific basis for either of these bare conclusions. He does not even assert that these conclusions are based on his experience or specialized knowledge of forensic pathology. California courts have excluded expert opinions for these very failings. See, e.g., Jennings v. Palomar Pomerado Health Sys., Inc., 114 Cal.App.4th 1108, 8 Cal.Rptr.3d 363, 369 (2003) (“[A]n expert’s conclusory opinion that something did occur, when unaccompanied by a reasoned explanation illuminating how the expert employed his or her superior knowledge and training to connect the facts with the ultimate conclusion, does not assist the jury.”).
Likewise, the only opinions in the pathologist’s declaration based on his “experience” are too speculative to be admissible. Although Herrmann admits that he had access to photographs of Johnson’s wounds, he does not base his conclusions on this evidence, but on testimony offered at trial, which he did not attend. Moreover, he gets a number of material facts wrong. Herrmann claims that the investigating officer did not notice that Johnson was injured until a few minutes after arriving at the crime scene and opines that “[gjiven my experience, if Mr. Johnson had been bleeding so profusely as to have deposited the pool of blood on the floor between the living room and the bedroom, it is unlikely that Detective Wright would have been oblivious to the injury.” In offering this opinion, he ignores Detective Wright’s testimony that Johnson was soaked in blood when he arrived at the house: Johnson had “blood on both cheeks, blood covering his shirt, and blood on his hands, and ... on his right shoulder.” He also fails to account for photographic evidence that Johnson bled enough to create a significant blood pool in the kitchen during the few minutes he talked with the detectives. Finally, Herrmann’s declaration ignores Detective Wright’s testimony indicating that one reason the detective did not believe Johnson was injured was because Johnson originally claimed that he was not hurt. See Richardson, 77 Cal.Rptr.3d 163, 183 P.3d at 1179 (“[An] expert’s opinion may not be based on assumptions of fact without evidentiary support, or on speculative or conjectural factors.” (internal quotation marks and alterations omitted)).
Even if by some serendipitous occurrence these conclusory and speculative assertions were not summarily excluded by the trial court,14 they are too weak to significantly contradict the State’s theory. Herrmann’s declaration is careful to conclude that the State’s theory is “unlikely” — not that it is impossible. In fact, it *988was not only possible that Johnson bled enough to create a pool of blood in the doorway, but consistent with other evidence admitted at trial. Photos taken of the crime scene several hours later established that there was a sizable blood pool in the kitchen area created by Johnson while he talked with Detective Wright in the few minutes before the ambulance arrived — a fact that the prosecutor hammered home in his closing statement.
More importantly, the pathologist’s hypothetical testimony is simply a repetition of the defense’s overarching theory, echoed by defense counsel throughout trial, that Johnson was shot around 4:20 am rather than 5:00 am and otherwise could not have had time to dispose of his marijuana plants and create blood trails throughout the house. For example, during his cross-examination of Detective Wright, defense counsel pressed, “And it wasn’t until after you had located [Johnson] in the kitchen area that you first entertained a belief that he had, in fact, suffered a wound; is that right?” The jury had full knowledge of the defense’s theory and could do with it what it pleased. The majority proffers no reason that having a pathologist repeat what defense counsel harped on during trial would be at all consequential to the jury’s decisionmaking — especially when the pathologist’s opinion is not supported by any explicit medical or scientific basis and could have been excluded for that reason.
The majority thus illogically concludes that defense counsel rendered ineffective assistance by failing to consult with and call to the stand an expert whose testimony would have been redundant, utterly unpersuasive, discredited by readily available forensic evidence, and most likely madmissable. Indeed, it concludes that a state court adjudicating otherwise has been unreasonable. This is beyond overreaching.
C
What the majority calls the “most damaging” aspect of counsel’s performance, his decision not to consult with a blood spatter expert, is actually the most understandable. Counsel had good reason for failing to consult with a blood spatter expert — he learned only minutes before the State’s blood spatter expert’s testimony that anyone would be opining on such matters. Moreover, counsel’s failure to consult with a blood spatter expert was far less damaging than the majority admits. For these reasons, even if counsel was not flawless in his approach to the blood spatter testimony, his performance did not rise to the level of unconstitutional deficiency, and in any event, the testimony he could have elicited through his own expert would not have contradicted the State’s case.
1
Before trial, counsel had no reason to know that an expert in bloodstain pattern interpretation would be helpful to his defense. Indeed, in his deposition he clearly confirmed that although he considered consulting a serologist before trial, he did not even consider consulting a blood spatter expert until after the State’s expert had testified. Counsel’s approach was not unreasonable.15
*989Counsel was not put on notice by the pre-trial actions of the prosecutor that such an expert would be necessary. Bell, the homicide detective who would testify about blood spatter for the prosecution, was not denominated as an expert on the subject prior to trial. Counsel never received a report from the prosecutor prior to Detective Bell’s testimony giving the basis for his testimony, outlining the subject matter of his testimony, or providing any indication of his expertise in bloodstain pattern interpretation. When counsel — quite responsibly — asked the court for a report prior to Detective Bell’s testimony concerning its subject matter, the judge flatly refused.
Counsel’s opening statement neither revealed counsel’s knowledge of the importance of expert blood spatter testimony nor promised the jury that he would call a blood spatter expert. Although counsel did, in his opening statement, advise the jury of the significance of blood spatter evidence, he also indicated precisely what bloodstain pattern evidence he believed supported his theory — none of which remotely required an expert to interpret. Counsel noted in his opening statement that “[t]he evidence of great importance is the lack of spatter marks in this area, and [the lack] of other pools of blood, and spatter marks found elsewhere in the house inconsistent with stories told by Gunner Johnson.” (emphasis added). He explained further that photographs of the crime scene would show that “there are no spatter marks of blood or anything on the couch;” that there were no blood marks on the sleeping bag on which Klein was purportedly sleeping; that there was a large pool of blood in the bedroom doorway that was never analyzed by the police; and that although Johnson claimed that he turned the lights on after being shot by Branscombe, there was no blood on the light switch. Not only did counsel elicit testimony establishing each one of these propositions during the course of the trial, he reiterated their significance in his closing statement.
Thus, because counsel was hardly remiss for failing to call a blood spatter expert to testify about the significance of the absence of blood spatters or pools of blood in certain areas of the house, the only possible “promise” counsel failed to honor was his pronouncement that the pool of blood (not blood spatters) in the bedroom doorway was “inconsistent with stories told by Gunner Johnson.” But, as counsel testified in his deposition, he had no reason to know before trial that the absence of small blood spatters surrounding this pool of blood would be significant in determining its source. Indeed, although there were “spatters” surrounding the pool that might ordinarily have alerted experienced counsel to the importance of a bloodstain analysis, these spatters were in this case created by a “foot stomp” — some unknown person had stepped into the middle of the blood pool and splashed portions of it on the bedroom wall and door. This was not enough to alert an attorney of counsel’s experience to the importance of a blood spatter expert before trial.
During trial, counsel diligently responded to the prosecution’s expert’s testimony. He first objected to the testimony and asked that it be excluded, claiming that it was proffered too late. He also asked for *990a preliminary facts hearing and for a copy of Detective Bell’s qualifications regarding blood spatter, but the court denied both requests. Although counsel does not remember asking for a continuance, it is highly unlikely, given the tenor of the trial and counsel’s other experience with such requests, that one would have been granted.
Once the prosecution’s expert, Detective Bell, had testified, counsel “considered [exploring] testimony relating [to] atomized blood droppings in the area of Mr. Klein’s head on the couch relating also to whether or not this could have been caused by a sneezing, coughing, choking of the person on the telephone that was somewhat adjacent thereto.” He also considered whether it was necessary to call an expert to testify concerning the various blood spatters in Johnson’s bedroom. In both cases, he waited until after his cross-examination of the prosecution’s expert, and then determined that neither expert would be profitable to his defense. This judgment appears to have been correct — Richter has failed to obtain in support of his petition a single expert who would contradict Detective Bell’s opinion about the origin of these spatter marks, or who would indicate that defense counsel’s theory about the origin of these spatter marks is even “possible.”
Even after Detective Bell’s blood spatter testimony, counsel had no reason to know that an expert would be helpful to determine whether the blood pool in the bedroom doorway came from Klein or Johnson. Detective Bell did not testify at all about blood spatter surrounding the pool— he testified about blood patterns and spatter on Klein’s body and in other areas of the crime scene. Counsel might have been alerted to the fact that a blood spatter expert might have been helpful in general, but he had no reason to believe that blood spatter testimony would definitively indicate that either Klein or Johnson created the blood pool. Indeed, counsel had good reason to believe that a blood spatter expert would not be able to confirm his client’s story — all the evidence so far presented at trial (and generally known to counsel) suggested that pursuing such an investigation would be fruitless. See supra Part II.A at n.7.
In sum, although counsel’s response to the blood spatter expert’s testimony was unlike his more reasonable response to the serologist’s testimony — he did not independently research bloodstain pattern science nor know enough about the subject to evaluate whether his cross-examination of the State’s expert was sufficiently effective — it was not so egregious to constitute ineffective assistance of counsel. By the time the blood spatter expert had testified, the trial was well underway. Counsel had to weigh the potential value of such an expert with the cost of acquiring one mid-trial. Taking time out of his trial preparation to find and retain a blood spatter expert who he had no reason to believe would disagree with the State’s expert or be able to opine on the source of the blood pool is not plainly an error in judgment. Even if, with the benefit of hindsight, it was not a prudent decision, it was certainly not constitutionally deficient. At the very least, it was reasonable for California courts to consider counsel’s choice not deficient.
2
Even if counsel made the wrong call, in light of the existing forensic evidence and the implausibility of Richter’s story it is far from clear that Richter was prejudiced. The majority confidently proclaims that the testimony of Richter’s new-found blood spatter expert, Ken Moses, would have created reasonable doubt as to Richter’s complicity in Klein’s murder. But this proclamation ignores the unchallenged blood spatter evidence at trial that directly undermined the defense’s entire case.
*991Specifically, Detective Bell opined that several important aspects of the blood evidence found at the scene were consistent with Johnson’s recollection that Klein was shot at or around the living room couch rather than in the bedroom doorway. First, the blood flow patterns on Klein’s face indicated that he was probably not moved “up or down” after he was shot in his left eye. More specifically, the blood from Klein’s left eye wound emanated from his ocular cavity and flowed only down the left side of his face. If Klein had been moved from a different position, his head could not have been leaned (even slightly) back or forward during this movement, or otherwise the blood would have flowed down toward his chin or up toward his hairline. Second, a small area of high velocity blood spatter (spatter created by an impact of tremendous force) was visible beside the living room couch. Bell opined that because he was unaware of any other action besides a gunshot wound that could have created such high velocity spatter, either Klein or another unknown person had to have been shot on or slightly above the couch. Finally, Bell testified that because the blood pool in the doorway was “pristine” and did not exhibit any “smearing” of blood, a person shot there would have had to be “lifted straight up” from the pool, which was unlikely. Richter’s newfound expert fails to contradict any of these propositions. See Maj. Op. at 965-66 n.18 (“The most damaging evidence at trial was Detective Bell’s blood spatter evidence.”)
Thus, for the forensic evidence to be consistent with Richter’s story, the drug-addled, intoxicated, 5'10", 155 pound Johnson would have had to perform an athletic feat of nearly Olympic proportions to move Klein from the bedroom doorway to the couch.16 According to Detective Bell’s testimony, Johnson would have had to lift the 6'1", 166 pound Klein “like a baby in [his] arms with [his] head parallel with the axis of [his] body” and carry him in precisely this position across the room, over a coffee table, and onto the couch. Johnson could not have dragged Klein even a short ways from the blood pool — he had to lift him “straight up.”
More importantly, for the forensic evidence to be consistent with Richter’s story, some other person would have had to have been shot at or near the couch — a proposition that is entirely inconsistent with both Richter’s recollection of the events in question and the evidence presented at trial. Richter did not testify that he saw a third wounded person on or near the couch. No additional shell casings were discovered in the living room that did not correspond to either Johnson’s or Klein’s wounds. Neither the State nor defense counsel offered any witnesses who claimed that another unknown person was in the house on the morning of December 20th. After fifteen years, no witnesses have materialized either to admit to being in Johnson’s residence during the night in question or to report on a newly disinterred corpse with at least one bullet wound.
Given the utter lack of evidence indicating that a third person was shot at Johnson’s residence, the totality of the State’s and Richter’s blood spatter testimony leads to only one possibility: Klein was shot on or near the couch. This scenario is entirely inconsistent with Richter’s story, which required Johnson to shoot the struggling Branscombe and Klein from a position near his bed.17 See Maj. Op. at *992963 (“If Klein was killed while lying on the couch in the living room, there was no possibility that Richter’s account was correct.”).
Although Richter’s newfound expert testimony opining about the lack of small satellite spatters in the area of the doorway blood pool may have weakened Johnson’s already scattered account of the shootings, it would not have been enough to render plausible Richter’s contrary account. Richter’s blood spatter expert opines that “the blood pattern [in the bedroom doorway] was caused by the pooling of blood from a source close to or lying on the floor.” This opinion is not inconsistent with the theory that the blood pool came from Johnson — instead, it means only that the blood couldn’t have originated from Johnson while he was standing up.
Richter, even after several years of foraging for expert testimony to reinforce his theory, has been unable to discover a single expert to contradict Detective Bell’s testimony that someone had to be shot within a few feet of the couch. At trial, defense counsel argued that the high velocity blood spatter near the couch was created when Johnson sneezed or coughed blood while talking on the nearby phone. Not surprisingly, Richter’s habeas petition lacks any expert declaration confirming this assertion that Johnson could cough blood with the force of a semi-automatic. Likewise, although the majority takes it upon itself to discredit Detective Bell’s testimony concerning Johnson’s movement of Klein, Richter has not produced any expert testimony disputing Detective Bell’s conclusions in this regard.
In sum, unlike Richter’s blood spatter testimony, which only contradicts Johnson’s recollection of the events, the State’s blood spatter testimony contradicts the defense’s entire case because it establishes that Johnson could not have physically shot Klein. The State’s blood spatter testimony has never been challenged — even Richter’s new blood spatter expert cannot contradict the State’s blood spatter evidence.18 Because the State’s case was not predicated on the fact that Johnson stood up while making the blood pool in the doorway, Richter’s testimony only discredits Johnson’s account of the events. In contrast, the defense’s entire case was predicated on Johnson having the ability to shoot Klein with a .22 while Klein was struggling with Branscombe, and this theory is entirely discredited by the State’s blood spatter testimony.
Because of the unchallenged forensic evidence directly contradicting Richter’s entire case, it was thus likely correct, and certainly not unreasonable, for a state court to assume that no prejudice resulted from defense counsel’s failure to obtain the expert blood spatter testimony.
D
As the majority recognizes, Strickland requires us to “consider the totality of the evidence before the judge or jury” when making a prejudice determination, 466 U.S. at 695, 104 S.Ct. 2052. But this inquiry does not merely encompass the evidence claimed by the habeas petitioner to be infected with error — it also should include a consideration of “factual findings *993... unaffected by the errors.” Id. We must thus “[t]ak[e] the unaffected findings as a given” and ask, “taking due account of the effect of the errors on the remaining findings,” whether “the defendant has met the burden of showing that the decision reached would reasonably likely have been different absent the errors.” Id. at 696, 104 S.Ct. 2052. The Supreme Court cautioned in Strickland that “a verdict or conclusion only weakly supported by the record is more likely to have been affected by errors than one with overwhelming record support.” Id. In this case, the State’s case was solid, but not overwhelming. The unaffected factual evidence before the jury, however, provides substantial support for its guilty verdict. Indeed, Richter and Branseombe’s story not only has so many holes that it is unbelievable, but uncontradicted expert testimony about the forensic evidence found at the crime scene further implicates the duo. Certainly, given these implausibilities and forensic evidence, we cannot say that a California court finding that Richter’s hypothetical expert testimony was not prejudicial was unreasonable in so doing.
In order to believe Richter and Branscombe’s story, the jury, first had to assume that Johnson was simultaneously savvy and stupid. The jury had to believe that Johnson was intelligent enough to do the following things after being shot in the shoulder and head by Branscombe: (1) take the .22 caliber semi-automatic handgun (that only Richter claimed Johnson owned) and bury it where no subsequent investigation would unveil it; (2) move the .22 caliber casing from his bedroom to the living room to implicate Richter in Klein’s death;19 (3) find his hip sack and any remaining CCI Stinger .22 caliber ammunition and bury it with the .22 caliber gun in order to make the scene consistent with a robbery; (4) throw his pager out his front door onto the lawn to make it look like Branscombe and Richter had taken his hip sack; and (5) lift Klein carefully from the bedroom doorway and move him to the couch to make it look like he was shot in his sleep. Despite this careful scheme, the jury also had to believe that Johnson was so stupid as to (1) move Klein, rather than leaving him exactly where he was shot, and simply tell the police that Klein was coming to Johnson’s aid when he was gunned down in the bedroom doorway; (2) throw his marijuana plants and drug paraphernalia over his fence into his neighbor’s yard, in plain view, instead of hiding them with the .22 handgun, the CCI Stinger ammunition, and the hip sack; (3) wait nearly forty minutes planning his story before calling the police, only to tell them three potentially inconsistent stories about what happened;20 and (4) tell the police specifically that he did not ever see Richter holding a gun in his bedroom, when he had already manipulated the crime scene to implicate Richter.
To believe Richter’s story, the jury also had to overcome uncontradicted, unchallenged forensic evidence corroborating Johnson’s story. The prosecution’s blood spatter expert testified that the blood flow patterns on Klein’s face indicated that he *994was not shot in the bedroom doorway; there was high velocity blood spatter in the area of the couch that could only have resulted from a gunshot wound; and there was no evidence of bruising or blood spattering on Klein’s face consistent with him being shot, falling, and making an impact with the bedroom floor. In short, for Richter’s account of the shootings to be at all consistent with the State’s uncontradicted expert testimony, the jury would have had to either invent a third wounded person at the crime scene whose existence was neither asserted by the parties nor suggested by the evidence at trial, or assume without any evidentiary or scientific support that Johnson coughed blood at a velocity of approximately 1000 meters per second.
Finally, to credit Richter’s version of the events, the jury had to believe that several highly improbable events occurred. The jury had to believe that (1) Richter and Branscombe left Johnson’s house around 2:00 am to return to their workplace at the Christmas tree lot because they suddenly felt motivated to clean up their work area; (2) Richter and Branscombe returned to Johnson’s house at 4:00 am to give Johnson’s roommate his pay and clothing from the Christmas tree lot, rather than waiting until the next morning; (3) upon the duo’s return to Johnson’s residence, Johnson and Klein were so intoxicated or drugged that they attacked Branscombe (who they had just smoked marijuana with a few hours before) for no reason;21 (4) Klein, who had let Branscombe into Johnson’s residence minutes earlier, initiated an attack on Branscombe with sufficient force that Branscombe felt justified in pulling out his .32 in self-defense; (5) Johnson, who had a perfectly good .22 caliber semiautomatic handgun, would first attempt to shoot Branscombe with his Cobray M-12, which was allegedly prone to jamming; (6) the M-12 immediately jammed after Johnson fired one shot, even though it functioned normally when tested later; (7) Johnson, a firearms enthusiast possessing the element of surprise, first fired his M-12 nearly straight into his bedroom floor, then missed hitting Branscombe with his .22 semi-automatic handgun from around twenty inches away; (8) Richter coincidentally had ammunition and a magazine at his residence that perfectly matched the .22 caliber casing found at the crime scene; (9) Richter and Branscombe ran off to the Yolo overpass to immediately dispose of the weapons even though they believed they were innocent; (10) Richter and Branscombe just happened to have several $100 bills, one of which happened to fall out while they were disposing of the guns in the Yolo overpass; (11) Richter and Branscombe avoided calling the police for a day and a half after the shootings, even though they learned shortly thereafter that the police were investigating Johnson’s house; (12) Richter initially denied ever being at Johnson’s house, even though he believed he was innocent; and (13) after an exhaustive search of the Yolo overpass in the area where Richter told the police the weapons had been thrown, no weapons were found, but the very next day, defense counsel returned to the over*995pass and found only Johnson’s weapon, the M-12, and not the potentially incriminating .32 that Richter and Branscombe remembered discarding in the very same spot.
Given the utter implausibility of Richter’s story, and the general plausibility of Johnson’s, it stretches the very meaning of prejudice to claim, as the majority does, that counsel’s failure to produce expert testimony that would at best contradict the prosecution’s experts and Johnson on some minor points is enough to sustain Richter’s burden of proving prejudice. Even if we could conclude as a matter of first impression that, in the face of such odds, the unproduced expert opinions might have persuaded a jury that Richter could not be guilty beyond a reasonable doubt, it is condescending to conclude that California state courts adjudicating otherwise have made an unreasonable decision under AEDPA.
Ill
The majority today grants Richter’s request for a do-over. The temptation to grant such a request is great in a contested case of “who shot whom first” when the evidence is either circumstantial or based on the eyewitness accounts of the participants. The temptation is particularly alluring when we know what trial counsel could not: that his defense is not going to work. But our post-hoc clairvoyance cannot diminish the fact that Richter’s counsel presented a coherent, viable defense. The majority’s perceived “deficiencies” in counsel’s preparation ignore counsel’s real-time dilemmas — the trial court, the prosecution, his limited resources and, most importantly, his client. We face none of those challenges. No trial will ever be perfect; no counsel complete in his preparation. Richter’s counsel, even if not perfect, was diligent in his preparation and vigorous in his defense. I would conclude that there was no Strickland error and, a fortiori, no grounds for us to grant the writ under AEDPA.
I respectfully dissent.
. Although the videotaped statements of Branscombe and Richter were never admitted into evidence, their existence was an important aspect of the case that Richter’s defense counsel had to consider. The prosecutor attempted to introduce the videotaped statements at trial during his cross-examination of Richter, but Richter denied ever making the statements. Because the defense successfully made an oral motion to exclude the videotape and transcription of its contents from trial, these statements were thus never admitted *973into evidence. Nevertheless, these statements, as Richter acknowledged on direct appeal, were definitely inculpatory. See People v. Branscombe, 72 Cal.Rptr.2d 773, 778-79 (Cal.Ct.App.1998), review denied and ordered not to be officially published (Cal. 1998) (noting that Richter accused the prosecutor on appeal "of committing misconduct ‘by asking appellant about a statement of his non-testify-mg co-defendant which implicated appellant.’ ” (emphasis added)).
. In light of this situation, it gravely oversimplifies the case to assume, as the majority does, that the “factual disputefover the source of the blood pool in the bedroom doorway] was the single most critical issue in the case ... from the standpoint of the defense,” Maj. Op. at 953.
. Although the majority is correct that the defense's theory did not rest entirely on this admittedly "unlikely scenario,” Maj. Op. at 948 n.2, this theory was the one the defense would have preferred the jury believe. Richter's counsel was careful to give the jury multiple ways to find his client not guilty. Counsel, however, repeatedly suggested that Klein had attacked Branscombe. In his opening statement, counsel suggested that "Patrick Klein was involved in the altercation in the bedroom to the extent that he wrestled with Mr. Branscombe, and as he was wrestling with Mr. Branscombe after the first shots were fired, Gunner Johnson obtained gun number two.” According to Richter's testimony, Branscombe believed that Klein was attacking him — Richter said that Branscombe had told him "that 'They tried to kill me.' ” (emphasis added). Later, in his closing argument, defense counsel suggested once again that Klein was at least an agitator. He told the jury that although they did not have to believe that Klein was armed, they could believe Klein might have become “entangled” with Branscombe in such a way as to make Branscombe "think that [Klein] was an aggressor.” Branscombe's attorney’s closing statement made this point unambiguously: he told the jury, "I submit that there are several items of evidence in this trial that would indicate to you that they were trying to kill Chris Branscombe.” (emphasis added). There was nothing erroneous about the defense's theory of the case in this regard. If Klein did not participate in the attack on Branscombe, Branscombe had no justification to shoot him at point blank range in the head. Of course, the defense did not require the jury to believe that Klein intended to kill Branscombe — Richter's counsel also argued that Branscombe might only have been guilty of manslaughter because Klein might simply have been caught accidentally in the middle of a shootout.
. Although, the defense argued at trial that Johnson’s gun safe had always been in Richter’s residence (where Johnson had previously stored several items), and presented several witnesses testifying such, the State also called several witnesses who testified that the gun safe was at Johnson's residence during the night in question. Fingerprint evidence was inconclusive in determining its provenance: police identified only Richter’s fingerprints on the safe.
. The defense team had only settled on this strategy shortly before trial. Previously, they had debated over a voluntary intoxication or insanity defense.
. The State’s theory was essentially that Johnson created the blood pool when he stood wounded by the bedroom doorway waiting for Richter and Branscombe to drive away.
. Defense counsel’s theory was that Klein wrestled with Branscombe at or near the bedroom doorway, that Johnson shot at the struggling Branscombe from near his bed, and that Johnson missed Branscombe and hit Klein. This fortuitous accident freed Branscombe from Klein’s grip, and Branscombe then shot Klein in the head, which caused Klein to collapse in the bedroom doorway.
. I should note that, although the majority trusts Richter’s experts as authoritative, both experts filed only very short affidavits containing their preliminaiy views and indicating their availability to testify. Neither expert had access to the blood samples, and neither has actually testified or been tested by cross-examination.
. The record shows that it was not "entirely predictable” that "[t]he State would present forensic testimony” on serology or any other type of blood evidence, Maj. Op. at 957-58 n.10. Indeed, even the State did not expect to introduce such testimony before defense counsel’s opening statement, and had to scramble to collect the evidence during trial. At that juncture, the State didn’t even know whether it had usable blood samples.
On this point the majority apparently "agree[s] with the dissent that counsel’s decision not to ask the prosecution to test the blood sample ... was not unreasonable.” Maj. Op. at 956 n.9. If so, the majority has created a duty to consult with experts that counsel did not intend to call as witnesses. Retaining a serologist before trial would only have “prepared” counsel by preventing him from having to search for an expert on short notice: but if this failure to prepare is constitutionally deficient, the first prong of Strickland becomes a nullity' — no counsel can possibly retain an expert to respond to every single potential expert a prosecutor might muster.
. The Supreme Court has emphasized that "when a defendant has given counsel reason to 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.” Strickland, 466 U.S. at 691, 104 S.Ct. 2052. The Court has noted that "inquiry into counsel’s conversations with the defendant may be critical to a proper assessment of counsel’s investigative decisions.” Id. Although in this case such conversations are privileged, the circumstances of the case are such that counsel could have reasonably believed his client was involved in the shootings and that they were not in self-defense. Thus, we cannot simply assume, as the majority implicitly does, that counsel should act as he would if he believed his client was not involved in the shootings— we must recognize that "[t]he reasonableness of counsel’s actions may be determined or substantially influenced by the defendant's own statements or actions.” Id.
. As previously discussed, counsel had good reason to believe that his client was involved in the shootings. He had no reason to believe that further investigation into the serology evidence would reveal exculpatory information.
. It is important to note that although it is possible that Johnson’s blood might naturally produce a stronger 2+ band and a weaker 1 + band, none of Richter’s professed experts can confirm this hypothesis. Thus, although Richter’s experts raise more hypothetical situations than did counsel on cross-examination in which a combination of Johnson and Klein’s blood would create Spriggs's result, they are still merely hypothetical and have no basis in the actual facts of the case.
. The breadth of the majority's holding is truly astonishing. According to the majority, no counsel can reasonably make the decision to not call an expert to the stand until he has consulted with said expert and determined that the expert's testimony would not be help*985ful to his case. Because this reasoning simply ignores the fact that the decision to consult with an expert is a strategic decision in its own right, and that the Supreme Court has upheld the ability of counsel to make strategic judgments after reasonable investigation, Wiggins, 539 U.S. at 533, 123 S.Ct. 2527, it promises to open the floodgates for habeas petitions in the Western United States based on the lack of expert testimony at trial.
. The majority’s assertion that such an objection to the prejudicial effect of the expert declarations was waived before the district court, see Maj. Op. at 963 n.15 & 968 n.23, is based on a misreading of the parties’ stipulations. In support of his habeas petition before the district court, Richter submitted declarations for his experts in lieu of direct testimony, and the parties stipulated that "the previously submitted declarations of petitioners’ expert witnesses are admissible for consideration on the merits, and that [the experts], if called to testify, would testify in accordance with their declarations.” In contrast to the majority’s assertion, this stipulation narrowly waives only an objection to the admissibility of the declarations in the district court in lieu of direct testimony for the purposes of considering Richter’s habeas petition. It does not clearly waive any objection to the prejudicial effect of the content of the declarations based on their admissibility in state court at trial. Indeed, determining the admissibility of Richter’s expert declarations at trial is one aspect of a "consideration [of the testimony] on the merits.”
. Contrary to the majority’s assertions, the Sixth Amendment does not require defense counsel to "consult” with any and every potentially helpful expert in preparation for trial. Instead, Strickland makes clear that only an "unreasonable” investigation is constitutionally deficient. 466 U.S. at 690-91, 104 S.Ct. 2052 ("Strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation.”). The Supreme Court has never suggested, as the majority does, that anything less than a "scorched-earth” investigative strategy is constitutionally deficient. *989Instead, the Court has long recognized that counsel should have the ability to weigh the potential value of investigation with the cost of acquiring it. See Wiggins, 539 U.S. at 533, 123 S.Ct. 25279("Strickland does not require counsel to investigate every conceivable line of mitigating evidence no matter how unlikely the effort would be to assist the defendant.”). Where counsel has no reason to believe, and should have no reason to believe, that a certain type of expert witness would be helpful to his case, it is not unreasonable for him to fail to “consult” with the expert either before, during, or after trial.
. Johnson was not only seriously wounded, but somewhat intoxicated (his blood alcohol level was 0.024) and probably under the influence of marijuana.
. The uncontradicted testimony of a ballistics expert indicated that Klein was shot with the .22 caliber bullet from between 11 and 22 inches away. Because Johnson's bedroom *992was more than 22 inches away from the couch, Johnson could only have shot the .22 caliber bullet that wounded Klein if Klein was shot in the bedroom doorway.
. Because Richter did not present any blood spatter evidence in his habeas petition that if presented at trial would have directly contra-dieted the State’s expert, we cannot assume, as the majority does, that the jury would have been forced to choose between the experts, see Maj. Op. at 964 n.16. Instead, we must consider whether the totality of the non-contradictory blood spatter evidence would have created reasonable doubt of Richter's guilt.
. Although the State's expert testified that the location of bullet casings were not entirely indicative of the location of the shooter, he also testified that it would be highly improbable that a gun fired in the bedroom would release a casing in such a way as to bounce out the bedroom door, leap over Klein's body, and come to rest in the comer of the living room.
. Because the defense's theory rested on the assumption that Johnson deftly manipulated all the forensic evidence inconsistent with Richter's story, the fact that Johnson failed to complete the effort by telling a police a consistent story was a major problem for the defense, as the prosecutor repeatedly pointed out in his closing argument.
. The majority misunderstands the defense's theory when it claims that Johnson’s intoxication can “explain why, in his befuddled state, not recognizing Branscombe, he shot what appeared to be an intruder when he awoke and saw an unknown person unexpectedly in his house,” Maj. Op. at 965. It was quite important to the defense to suggest that Johnson and Klein were not asleep when Branscombe entered the house, so that it could be proposed that Branscombe did not enter the house without their consent. Thus, in his closing statement defense counsel proclaimed, "Christian Branscombe did not surprise anybody, did not go running in the dark and start shooting. It didn’t happen that way.” (emphasis added).