ORDER OF COURT
The petition for rehearing having been denied by the panel of judges who decided the case, and the petition for rehearing en banc having been submitted to the active judges of this court and a majority of the judges not having voted that the case be heard en banc, it is ordered that the petition for rehearing and the petition for rehearing en banc be denied.
LYNCH, Chief Judge,dissenting from the denial of en banc review.
Through AEDPA, Congress has narrowly limited federal court review of state criminal convictions. Specifically, federal habeas relief is now unavailable to state prisoners unless the state court adjudication of the prisoner’s claim “resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States.” 28 U.S.C. § 2254(d)(1).
Here, a Massachusetts jury convicted petitioner Michael O’Laughlin of several charges relating to the gruesome nighttime beating of his neighbor, Annmarie Kotowski, which left her near death. The Supreme Judicial Court of Massachusetts (“SJC”) unanimously upheld that conviction, see Commonwealth v. O’Laughlin (O’Laughlin II), 446 Mass. 188, 843 N.E.2d 617 (Mass.2006), reversing the intermediate state appellate court’s holding that the evidence was insufficient, see Commonwealth v. O’Laughlin (O’Laughlin I), 63 Mass.App.Ct. 805, 830 N.E.2d 222 (Mass.App.Ct.2005). The federal district court denied O’Laughlin’s habeas petition.
Nonetheless, a panel of this court on federal habeas review found that the evidence against O’Laughlin was insufficient to support the conviction and held that the SJC’s contrary determination was an unreasonable application of clearly established federal law. See O’Laughlin v. O’Brien (O’Laughlin III), 568 F.3d 287 (1st Cir.2009).
To reach this result, the panel opinion applied a sufficiency standard that is materially different from the Supreme Court’s clearly established law under Jackson v. Virginia, 443 U.S. 307, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979), and exceeded its narrowly defined power to review state court criminal convictions under AEDPA. Because the sufficiency standard applied by the panel has consequences well beyond this case and the panel’s application of that standard to this case upsets the congressionally defined role of federal habeas re*2view in our federal system, I respectfully dissent from the denial of en banc review.
I.
Around 2:00 a.m. on November 17, 2000, Kotowski was severely beaten in her apartment by an assailant whom she could not later identify because of the memory loss she suffered as a result of the attack. The assailant nearly killed Kotowski, breaking almost every bone in her skull, as well as several in her hands. The evidence supported the prosecution’s theory that Kotowski had awakened to find her assailant in the apartment and had been beaten to prevent her from identifying her assailant. Kotowski knew O’Laughlin; indeed, she had been cool to his interest in her.
O’Laughlin had the means to perpetrate this attack. He lived only two doors down from Kotowski and, as a maintenance worker in the apartment complex, had a key to Kotowski’s apartment, had been there, and knew the apartment’s layout. That level of access was necessary to commit the crime because the police found no signs of forced entry into Kotowski’s apartment. Additionally, the police found a weapon that could have been used in the attack — a baseball bat with O’Laughlin’s name inscribed on the barrel — hidden under some leaves in the woods behind the apartment complex. Moreover, the police encountered O’Laughlin near the scene of the crime just minutes after the attack and noticed the next day that he had a cut on his face and a bruise below his left ear, which were consistent with being involved in a struggle.
O’Laughlin also had a motive. He had smoked crack cocaine in the hours before the attack and had called several drug dealers from the telephone in his apartment only minutes before the attack, desperately seeking more drugs. O’Laughlin had no money to buy more drugs that night, and he thought that Kotowski was well off from seeing her apartment. Although Kotowski’s attacker took nothing from the apartment, police found her purse on the floor near the bathroom. Kotowski testified that she typically kept her purse on the inside handle of her bedroom door. The prosecution argued that Kotowski’s attacker left without taking anything because, during the attack, he had heard Kotowski’s upstairs neighbor get to the phone and speak, and feared that the police were on the way. The police reenacted the upstairs neighbor’s telephone call and determined that a person in Kotowski’s apartment could hear that the upstairs neighbor was speaking on the phone.
O’Laughlin also demonstrated a consciousness of guilt. He appeared “uneasy and distant” when the police encountered him outside shortly after 2:00 a.m. in near-freezing temperatures wearing nothing but his boxer shorts. O’Laughlin told the police a series of lies and repeatedly shifted his story as to what he was doing on the night of the attack. When O’Laughlin encountered the police shortly after the attack, they told him they had been unable to locate unit 202, the apartment of the upstairs neighbor who had called the police. O’Laughlin, who knew that the units had recently been renumbered and that unit 202 was now unit 16, did not disclose that information to the officers. Rather, he misdirected them by changing their focus to purported animal noises he had heard coming from the dumpster, where the police then went. Finding nothing of interest in the dumpster, the police decided to leave. When the police returned to the apartment complex the next day, O’Laughlin refused to let the police swab what appeared to be a blood stain in his apartment and cleaned up the stain before allowing the police to return. And he gave a number of inconsistent, and sometimes *3inherently incredible, statements about the events of that night.
A primary defense theory was that Kotowski’s estranged husband, not the defendant, had committed the assault. The husband testified, and the jury rejected that theory.
II.
The state’s petition for rehearing en banc presents what is essentially a three part argument: (1) that the panel has articulated and applied a new sufficiency of evidence test, which is not clearly established law from the Supreme Court and is, in fact, inconsistent with Jackson; (2) that the panel decision otherwise violates AED-PA’s strictures that relief may be granted only where the state court engages in an “unreasonable application” of clearly established federal law; and (3) that the panel has, contrary to Jackson, failed to look to the totality of the evidence and has failed to draw all inferences in favor of the verdict as required. All three aspects of the state’s petition raise significant issues of law, which in my view warrant rehearing en banc.
Before addressing the merits of the state’s arguments, it is important to set forth the basic framework for our review. We owe deference to the state court proceedings on two levels. First, because O’Laughlin’s core argument is to the sufficiency of the evidence, we must uphold the jury’s verdict unless “it can be said that no rational trier of fact could find guilt beyond a reasonable doubt.” Jackson, 443 U.S. at 317, 99 S.Ct. 2781. In making this determination, we view the evidence in the light most favorable to the prosecution. Id, at 319, 99 S.Ct. 2781.
Second, we review O’Laughlin’s sufficiency challenge through the lens of federal habeas review. Under the federal habeas statute, we must presume the correctness of all factual determinations made by the state court, see 28 U.S.C. § 2254(e)(1), and may grant the writ only if the state court’s decision was “contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States,” id. § 2254(d)(1). Under this standard, we must uphold even state court decisions that are incorrect so long as they are reasonable. See Williams v. Taylor, 529 U.S. 362, 410, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (“For purposes of today’s opinion, the most important point is that an unreasonable application of federal law is different from an incorrect application of federal law.” (emphasis in original)); see also Schriro v. Landrigan, 550 U.S. 465, 473, 127 S.Ct. 1933, 167 L.Ed.2d 836 (2007) (“The question under AEDPA is not whether a federal court believes the state court’s determination was incorrect but whether that determination was unreasonable — a substantially higher threshold.”).
Thus, to prevail, O’Laughlin must make a two-fold showing: (1) that no reasonable juror could have voted to convict; and (2) that the contrary view of the state court was not only wrong but unreasonable.
A. The Proper Test for Assessing a Sufficiency Challenge
The panel’s opinion applied the Sixth Circuit’s test for sufficiency, which holds that if the jury’s verdict rested on “circumstantial evidence ... [that] amounts to only a reasonable speculation,” it must be vacated. See O’Laughlin III, 568 F.3d at 302 (quoting Newman v. Metrish, 543 F.3d 793, 796 (6th Cir.2008), petition for cert. filed, 77 U.S.L.W. 3645 (U.S. May 12, 2009) (No. 08-1401)). The panel opinion opened its evaluation of the SJC’s opinion by quoting from and implicitly adopting the Neuman test. It then said that test *4was met in this case, repeated the term “reasonable speculation,” and drew heavily on Newman, giving it four pages of discussion.
“Reasonable speculation” is not the standard the Supreme Court has articulated to assess evidentiary sufficiency under Jackson. Instead, the Jackson standard asks “whether, after viewing the evidence in the light most favorable to the prosecution, any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.” 443 U.S. at 319, 99 S.Ct. 2781 (emphasis in original). The term “reasonable speculation” appears nowhere in Jackson or in any of the Supreme Court’s sufficiency jurisprudence. Indeed, until the panel’s opinion in this case, the Sixth Circuit was the only circuit to use that standard.
The “reasonable speculation” standard is materially different from the Jackson standard in that it prohibits convictions from resting on reasonable inferences drawn from the evidence.1 This new standard strongly disfavors circumstantial evidence,2 contrary to our long-standing recognition that a criminal conviction may rest on circumstantial evidence alone. See United States v. Rodríguez-Durán, 507 F.3d 749, 758 (1st Cir.2007) (“[Cjircumstantial evidence alone may be sufficient to provide a basis for conviction.”); see also United States v. Downs-Moses, 329 F.3d 253, 261 (1st Cir.2003) (“[W]e do not favor direct evidence over circumstantial evidence, as either type of evidence may satisfactorily support a conviction.”).
This devaluation of circumstantial evidence by a federal habeas court has sweeping implications. The government relies heavily on circumstantial evidence to prove cases where direct witness testimony about the commission of a crime is unavailable. This is particularly true in domestic and acquaintance violence prosecutions, where the victims, who often have the only direct evidence of their abusers’ guilt, are left unable to testify or they recant or refuse to cooperate. See T. Linger, Prosecuting Batterers After Crawford, 91 Va. L.Rev. 747, 768 (2005) (“Recent evidence suggests that 80 to 85 percent of battered women will recant at some point.”). The panel’s “reasonable speculation” rule, which substantially discounts the power of circumstantial evidence, jeopardizes these and other prosecutions where the only evidence of the defendant’s guilt is circumstantial.
In addition to the infirmities in the substance of the “reasonable speculation” rule, the panel’s use of another circuit court’s precedent to grant habeas relief is troubling given AEDPA’s requirement that the state court’s error be measured against “clearly established Federal law, as determined by the Supreme Court of the United States.” 28 U.S.C. § 2254(d)(1) (emphasis added). Indeed, the Supreme Court has repeatedly cautioned that if it has not recognized a particular circuit court legal standard, a state court’s failure to follow that standard cannot compel the grant of habeas relief. See, e.g., Knowles v. Mirzayance, — U.S. -, -, 129 S.Ct. 1411, 1419, 173 L.Ed.2d 251 (2009) (holding that federal habeas relief is unavailable based upon the application of “a specific legal rule that has not been squarely established by this Court.”); Carey v. Mus*5ladin, 549 U.S. 70, 76-77, 127 S.Ct. 649, 166 L.Ed.2d 482 (2006) (same).
B. Limitations Under AEDPA
The panel opinion’s analysis of the reasonableness of the SJC’s opinion substantially departed from the level of deference for the determinations of the state courts that Congress and the Supreme Court have provided the federal courts under AEDPA.3 The panel opinion states that the SJC’s sufficiency determination was unreasonable because: (1) O’Laughlin’s prior contacts with Kotowski were minimal; (2) the facts here “do[] not align” with a robbery motive; (3) the prosecution’s weapon was “merely ‘consistent’” with Kotowski’s injuries; (4) the motive evidence here was “mere conjecture”; (5) O’Laughlin was not caught in a lie and did not flee the scene; and (6) the evidence of consciousness of guilt was “bare conjecture.” O’Laughlin III, 568 F.3d at 306-OS. At most, these examples amount to a disagreement as to the weight particular pieces of evidence should receive and the inferences to be drawn from them. But a state prisoner is not entitled to habeas relief under AEDPA merely because a federal court disagrees with the state court’s assessment of the evidence. See Williams, 529 U.S. at 410, 120 S.Ct. 1495 (recognizing that a state court’s determina,tion is not an unreasonable application of law merely because it is erroneous). The state court’s determination must be so incorrect as to be an unreasonable application of law, which is “a substantially higher threshold.” Schriro, 550 U.S. at 473, 127 S.Ct. 1933.
The state’s petition for rehearing argues that en banc review should be granted in this case because it is part of an emerging pattern of non-compliance with AEDPA where the federal courts of appeals have been granting habeas relief in circumstantial evidence cases. See Newman, 543 F.3d at 796; see also Brown v. Farwell, 525 F.3d 787 (9th Cir.2008), cert. granted sub nom, McDaniel v. Brown, — U.S. -, 129 S.Ct. 1038, 173 L.Ed.2d 468 (2009).
Having read the record before the federal court, I am convinced that the SJC’s holding on the sufficiency issue was not only a reasonable application of Jackson but also entirely correct under the Jackson standard. The SJC’s seventeen-page opinion shows there was a detailed review of the trial record by the justices and articulates and applies the correct constitutional standard for sufficiency of the evidence.
I start with the SJC’s finding, that O’Laughlin had a financial motive for breaking into the unit, which led to the attack on Kotowski. The SJC explained: “The defendant had purchased crack cocaine on two different occasions on the evening of November 16 and had been ingesting drugs and beer all night. He had run out of both drugs and money later in the evening and was desperately seeking to obtain more drugs.” O’Laughlin II, 843 N.E.2d at 627. In short, O’Laughlin needed money, and the prosecution put on evidence that O’Laughlin suspected Ko*6towski was well off and a potential source of funding.
The panel opinion, however, dismissed robbery as a motive because Kotowski’s assailant took nothing from the apartment. See O’Laughlin III, 568 F.3d at 306. But as the SJC explained, this motive was not negated by the fact that the assailant ultimately took nothing. It held that the attacker could well have been frightened off before he took anything by the sounds of the upstairs neighbor speaking with the police on the phone as the attack took place. See O’Laughlin II, 843 N.E.2d at 627 n. 11. This explanation is quite reasonable and has evidentiary support. Indeed, a reenactment of the police phone call demonstrated 'that a person in Kotowski’s apartment could hear the neighbor walking around upstairs and speaking on the phone. And the location where the police found Kotowski’s purse also supports this theory.
The panel opinion also attempts to dismiss the SJC’s finding that O’Laughlin had the means to commit the crime because he had a master key and there were no signs of forced entry by saying that others had a master key and that the victim could have let her assailant into the apartment. See O’Laughlin III, 568 F.3d at 302. These alternative explanations are, of course, possible. But to credit them requires that the court draw inferences against the jury’s verdict, which it may not do under Jackson. And in any event, these alternative theories in no way diminish the SJC’s finding that O’Laughlin had the means to commit the crime.
There is no need to catalog the entire array of the panel opinion’s disagreements with the SJC’s rational inferences. Instead, I will focus on the panel opinion’s primary disagreement with the SJC, which relates to the appropriate weight to give the evidence of O’Laughlin’s consciousness of guilt. The SJC characterized this as the strongest evidence of O’Laughlin’s guilt. See O’Laughlin II, 843 N.E.2d at 627-28. The panel opinion, however, found this evidence “minimally probative.” O’Laughlin III, 568 F.3d at 303.
The SJC’s conclusion that there was strong evidence of consciousness of guilt is both correct and entirely supported by a reasonable application of Jackson. O’Laughlin gave the police a series of differing accounts as to his activities that night, which a jury could easily conclude were untrue and implausible. The jury could reasonably infer the following. When the police first encountered O’Laughlin after 2:00 a.m. outside in his boxer shorts in near-freezing weather,4 O’Laughlin appeared “uneasy and distant” and lied to the police by saying that he was asleep and had been awakened by the sound of screaming. When O’Laughlin was asked where the screaming was coming from after the police were unable to locate unit 202, O’Laughlin altered his position and said he thought the screaming might have been a raccoon that had gotten stuck in the dumpster and explained that he had placed a stick in the dumpster to allow the animal to escape. This misdirection, along with O’Laughlin’s failure to inform the police of the apartment complex’s new unit numbering, distracted the police away from the victim’s apartment.
The next day, O’Laughlin gave the police a different account, saying that he had *7been asleep but was awakened by the sound of the police cruisers arriving (not screaming) and so went out in the freezing cold nearly unclothed. He later again re-characterized his actions, claiming again to have been sleeping until he was awakened by screaming from animals (foxes or raccoons) fighting.
In fact, O’Laughliris telephone records indicated that he had been on the phone frequently just minutes before the police arrived. His accounts that he had been sleeping were a lie. O’Laughlin later also lied in order to explain away those phone calls in a letter he sent to the building manager, saying that an intruder into his unit had placed those calls from his apartment while he slept. A jury could reasonable conclude that O’Laughliris shifting stories and overt lies demonstrated a strong consciousness of guilt.
Also damning are O’Laughliris efforts to remove what he described as a blood stain from the door of his closet. When police noticed the stain and requested to take a sample, O’Laughlin withdrew his consent to have his apartment searched. After the police left, O’Laughlin wiped up the stain with his finger and saliva.
It was only when the police returned to his apartment and threatened to get a search warrant that he consented to allow the police to search his apartment again. By then, he had wiped away the stain.
O’Laughliris explanation that his behavior was motivated by a concern that the police would discover his drug paraphernalia could be rationally rejected as entirely inconsistent with the fact of both consents. Moreover, the police assured O’Laughlin that he would not be prosecuted for any drug crimes if he voluntarily surrendered any contraband to the police, further reducing the credibility of O’Laughliris explanation for his reluctance.
Whether a state court decision represents an unreasonable application of clearly established federal law is a difficult judgment for a federal court to make. The need for better articulation of the legal standards for making that judgment, particularly where the record is largely circumstantial, in my view, constitutes an appropriate basis on which to grant en banc review.
C. The Panel Opinion’s Method of Analysis
More fundamentally, the panel opinion’s method of analysis sets a dangerous precedent because it draws inferences against the jury’s verdict and it takes a piecemeal approach to analyzing the evidence. That is, the panel opinion’s approach to the sufficiency question in operation amounts to a de novo assessment of the record through which the court substitutes its own independent assessment of the record for the jury’s evaluation. This sort of analysis should be foreign to an appellate court, which has access only to a cold record. And it is a particularly dangerous exercise in a case like this, where much of the prosecution’s case depended on the testimony of witnesses and an assessment of the credibility of those witnesses.
Under Jackson, an appellate court considering a sufficiency challenge is not on an equal footing with the jury. Instead, “the factfinder’s role as weigher of the evidence is preserved through a legal conclusion that upon judicial review all of the evidence is to be considered in the light most favorable to the prosecution.” Jackson, 443 U.S. at 319, 99 S.Ct. 2781 (emphasis in original).
The panel opinion, however, fails to respect the role of the jury in deciding which evidence to credit and what reasonable inferences to draw from that evidence. For example, the panel opinion rejected *8the reasonable inference that Kotowski’s assailant fled before taking anything of value from the apartment because he heard her upstairs neighbor on the telephone with the police. Instead, the panel held that “no rational juror could conclude that the placement of this phone call frightened him off given the volume and duration of Mrs. Kotowski’s screams.” O’Laughlin III, 568 F.3d at 302. Although it would be entirely permissible for a jury to resist drawing this inference, an appellate court applying the Jackson standard does not have the freedom to pick between competing reasonable inferences in this manner. See, e.g., United States v. Ortiz, 447 F.3d 28, 34 (1st Cir.2006) (recognizing that “competing inferences are not enough to disturb the jury’s verdict,” even in a case where the evidence of the defendant’s guilt was “entirely circumstantial”).
The panel opinion’s method of analysis suffers from a related defect to the extent that it dismisses pieces of evidence by treating them in isolation, instead of considering the totality of the evidence together. For example, the panel opinion resisted concluding that a reasonable jury could consider that the aluminum baseball bat that the police found in the woods twenty-five yards from Kotowski’s apartment was evidence of O’Laughlin’s guilt. The panel dismissed this because “any bat likely would have been consistent with her injuries” and because the upstairs neighbor had described sounds like wood hitting on wood. O’Laughlin III, 568 F.3d at 302. But the fact that Kotowski’s injuries were consistent with being beaten with a baseball bat is only one of the facts that would have permitted a rational juror to conclude that the bat which the police found was the assailant’s weapon. The panel’s analysis ignores several other key facts relating to the bat, which had been found in close proximity to the scene of the crime. O’Laughlin’s name was inscribed on it, and O’Laughlin was closely tied to the crime by the prosecution’s other evidence of his motive, means, and consciousness of guilt. The bat was hidden under some leaves and debris as though someone had tried to carefully conceal it. And the bat was otherwise clean, suggesting that it had not been in the woods for long and had been cleaned off.
This model of effectively performing de novo review rather than drawing inferences in favor of the verdict, if widely adopted, would have unfortunate consequences. It calls for appellate courts to make determinations that are well beyond the institutional capacity of a court to do working from a cold record. And it conflicts with the usual rules of finality, effectively giving criminal defendants multiple opportunities to make their case in the first instance. Indeed, the government is particularly prejudiced when we overreach to decide these cases on sufficiency grounds because principles of double jeopardy prevent the habeas petitioner from being retried. See Evans v. Thompson, 518 F.3d 1, 7 (1st Cir.2008) (“Sometimes double jeopardy principles mean the habeas petitioner is simply released, as is done when the evidence of guilt is insufficient.”).
III.
Sufficiency questions are inherently fact-bound, and so my disagreement with the panel opinion may appear, at first glance, to be a dispute over these facts with few implications beyond this case. But my concerns with the panel opinion run far deeper than that. In my view, the panel’s implicit adoption of the “reasonable speculation” standard impermissibly alters our circuit’s approach to sufficiency questions, substantially disadvantaging whole categories of cases- — like domestic and acquaintance violence prosecutions — -where the *9government’s evidence is largely circumstantial. And the panel’s approach sets a dangerous precedent to the extent that it deviates from the narrow role Congress envisioned for federal habeas review under AEDPA by substituting the federal court’s own independent assessment of the facts for the state court’s.
We have previously granted en banc review to address similarly important questions relating to the contours of federal habeas review, see, e.g., McCambridge v. Hall, 303 F.3d 24 (1st Cir.2002) (en banc), and I would do so again here. Thus, with the greatest respect for my colleagues, I dissent from the denial of en banc review. See United States v. Rivera, No. 95-2186, 1996 WL 338379 (1st Cir. June 18, 1996) (Lynch, J., dissenting from the denial of en banc review).
. The term "reasonable speculation" contains an inherent conflict. Recognizing that a conclusion rests on speculation suggests that the inferences upon which it is based are unreasonable.
. The dichotomy drawn by the panel between circumstantial and direct evidence is questionable. Convictions often rest on inferences which are drawn from direct testimony, as well as circumstantial evidence.
. The result of this case — a federal appeals court reversing the state’s highest court on the question of evidentiary sufficiency in a criminal prosecution — is also surprising given the SJC's historically favorable treatment of criminal defendants. Indeed, the SJC often extends criminal defendants greater protections than those required under the federal Constitution. See, e.g., Commonwealth v. Gomes, 453 Mass. 506, 903 N.E.2d 567 (2009) (limiting the police's power to pat frisk suspected drug dealers for weapons); Commonwealth v. Mavredakis, 430 Mass. 848, 725 N.E.2d 169 (2000) (holding that the police have a duty to inform a criminal defendant of his lawyer’s efforts to contact him).
. The prosecution argued that O'Laughlin came outside wearing only boxer shorts on such a cold night because he had stripped off his bloody clothes just before the police arrived. After the police left that night, O’Laughlin had several hours during which he could have disposed of any bloody clothing and any weapon before the police returned the next morning.