dissenting:
In Heck v. Humphrey, 512 U.S. 477, 114 S.Ct. 2364, 129 L.Ed.2d 383 (1994), the Supreme Court ruled that a proper respect for finality and consistency of judgments bars actions under 42 U.S.C. § 1983 that require “impugning” an extant conviction. Id. at 486 n. 5, 114 S.Ct. 2364. In Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), and its progeny, the *151prosecution’s constitutional obligation to disclose information that is material to the defense has been located in the truth-seeking function of a trial, and not in any “sporting theory of justice.” Id. at 90, 83 S.Ct. 1194. The majority opinion undermines both the finality premise of Heck and the truth-seeking foundation of Brady. It holds that Marcos Poventud, who secured a new trial from the State of New York based on a police officer’s failure to disclose information that might have impeached the victim’s identification of Poventud as the armed robber of a livery cab, can sue for Brady damages even though Poventud resolved the charges against him on remand by entering a guilty plea (to a lesser offense) that made clear that the eyewitness identification was sound, and that Poventud’s alibi defense at the first trial was perjury.
I respectfully dissent from this decision and write to explain why the majority’s reasoning impairs the future application of Heck and Brady in this Circuit.
I
On the evening of March 6, 1997, between Oliver Place and Marion Avenue in the Bronx, a livery cab driver, Younis Duopo, was held up at gunpoint and shot in the neck. Poventud and a co-defendant were indicted for the armed robbery and attempted murder. At Poventud’s trial, the central issue was identity: Poventud and some of his friends testified that on the date and at the time of the robbery, he was with them elsewhere, playing video games; Duopo, the victim, identified Poventud as his assailant, both pretrial from a photo array and again at the trial itself.
Rejecting Poventud’s testimony and crediting the victim’s identification, the jury convicted Poventud of attempted murder in the second degree, attempted robbery in the first degree, assault in the first degree, and criminal possession of a weapon in the first degree. He was sentenced to serve an indeterminate sentence of 10 to 20 years. The conviction and sentence were affirmed on appeal. See People v. Poventud, 300 A.D.2d 223, 224, 752 N.Y.S.2d 654 (1st Dep’t 2002).
In 2005, the New York Supreme Court, Bronx County, vacated the conviction and ordered a retrial on the ground that the prosecution had failed to disclose impeachment evidence in violation of Brady. Immediately after the hold-up, police found photo identification of Poventud’s brother, Francisco, in a wallet found in Duopo’s cab. From a photo array, Duopo selected a photograph of Francisco, which he initialed and dated. When it was ascertained that Francisco had been in prison at the time of the crime, Marcos Poventud became a suspect. The state court vacated on the ground that Brady was violated by the State’s failure to disclose Duopo’s initial identification of Francisco as the assailant. See People v. Poventud, 10 Misc.3d 337, 341, 802 N.Y.S.2d 605 (N.Y.Sup.Ct.2005). A new trial was ordered.
The vacatur afforded Poventud the opportunity to test the reliability of the identification before a jury on retrial, but he chose instead to resolve the outstanding charges by pleading guilty to the lesser included offense of attempted robbery in the third degree. At the ensuing guilty plea proceeding, Poventud admitted his armed presence at the scene and his participation in the robbery:
COURT: In this case it’s charged that on or about March 6, 1997, at approximately 8:40 in the evening, in the area of Oliver Place and Marion, M-A-R-I-ON, Avenue here in the county of the Bronx, you did attempt to steal personal property from another person by using force, in that you used a weapon in your *152attempt to steal personal property. Are those charges true?
THE DEFENDANT: Yes.
This plea colloquy thus conclusively confirmed the jury’s key findings of fact: that Duopo’s ultimate identification of Marcos Poventud was sound and that Poventud’s trial testimony (and that of his friends) was false.
Poventud was re-sentenced to one year in prison and, having already served nine years, was released.1 Soon after his release, Poventud filed this action under 42 U.S.C. § 1988 seeking money damages for alleged violation of his Brady right. He then obtained a stay of that action and filed (but later withdrew) a motion' challenging the voluntariness of his plea. On defendants’ motion for summary judgment in the § 1983 action, Judge Batts ruled that Poventud’s claims were barred by Heck, and dismissed the case. See Poventud v. City of New York, No. 07 Civ. 3998(DAB), 2012 U.S. Dist. LEXIS 30763, 2012 WL 727802 (S.D.N.Y. Mar. 6, 2012).
On Poventud’s appeal of Judge Batts’s ruling, the three-judge panel of this Court divided. See Poventud v. City of New York, 715 F.3d 57 (2d Cir.2013). The majority held that the Heck bar is subject to a gaping, unprecedented exception: namely, that any person convicted of a crime can bring a § 1983 action necessarily implying the invalidity of that conviction if he cannot currently bring a habeas petition— including any person released from prison after the service of his sentence. The dissent rejected that theory and concluded that Poventud did not benefit from any exception to Heck (if one even exists).
A majority of the active judges voted to decide in banc the scope of the Heck bar and (if necessary) any exceptions to it. The in banc majority again reverses the district court’s dismissal. In doing so, however, it abandons the panel majority’s reasoning, relying instead on a point of law that received merely passing reference in a footnote to the panel majority’s opinion. See Poventud v. City of New York, 715 F.3d 57, 61 n. 2 (2d Cir.2013) (expressing doubt that success on § 1983 claim would impugn Poventud’s guilty plea, but declining to reach the issue).
II
Poventud’s 2006 judgment was entered on his guilty plea, made in open court and with the assistance of counsel, and has not been disturbed.2 Poventud’s solemn admission of guilt, which places him at the scene of the crime, armed, with the intent to commit robbery, “quite validly removes the issue of factual guilt from the case,” Menna v. New York, 423 U.S. 61, 62 n. 2, 96 S.Ct. 241, 46 L.Ed.2d 195 (1975) (per curiam) (emphasis in original), and is admissible against Poventud for all purposes, Mitchell v. United States, 526 U.S. 314, 324, 119 S.Ct. 1307, 143 L.Ed.2d 424 (1999).
The majority opinion does not dispute that, if the success of a § 1983 claim would *153necessarily impugn a criminal conviction, the complaint must be dismissed unless the plaintiff can "prove that the conviction has been reversed on direct appeal, expunged by executive order, declared invalid by a state tribunal authorized to make such determination, or called into question by a federal court's issuance of a writ of habeas corpus." Heck, 512 U.S. at 486-87, 114 S.Ct. 2364; see also Wallace v. Kato, 549 U.S. 384, 393, 127 S.Ct. 1091, 166 L.Ed.2d 973 (2007)(noting that Heck bar applies where § 1983 claim would necessarily "impugn" an extant conviction). None of those things have happened here.
The i~ bctnc majority nevertheless holds (1) that Poventud may sue for damages pursuant to § 1983 on the theory that Poventud's Brady-based § 1983 claim does not impugn the (extant) judgment entered on his guilty plea, and (2) that the (vacated) judgment entered on his 1998 conviction was favorably terminated within the meaning of Heck. I will take these two determinations one by one, to show that Poventud's claim-as pled and as rewritten by the majority-impugns the 2006 judgment, see infra Points III, IV, V, and that the vacatur of the 1998 judgment was not a favorable termination because it culminated in the guilty plea, see infra Point VI.
III.
Because Poventud's guilty plea is central to the two-part showing I have just summarized, I begin with an overarching point: Poventud's complaint unambiguously impugns the validity of his guilty plea by asserting actual innocence. The complaint does this, moreover, both as it is pled and as it is presented on the motion for summary judgment.
The Second Amended Complaint (the "Complaint") alleges that the prosecution's evidence that Poventud was present at the crime scene was inherently unreliable, and even insufficient itself to sustain a conviction. See Second Am. Compi. ¶1136-41, 69-74, 128. The Complaint characterizes the withheld evidence as "evidence of innocence." Id. ¶ 128.
Poventud's sworn affidavit submitted in opposition to defendants' motion for summary judgment on Heck grounds declares his innocence in unequivocal terms: "I did not commit the crime. I am innocent." Aff. of Marcos Poventud, ¶ 5 (July 19, 2011). So too does his opposition briefing impugn his guilty plea directly: "Plaintiff knew that maintaining his innocence had resulted in spending nine years in prison, and bowed to the pressure to `admit' guilt because it would result in his immediate release." P1's Resp. to Defs' Rule 56.1 Statement and Statement of Additional Facts, ¶ 269. Finally, Poventud's damages theory, as set out in his summary judgment papers, is squarely premised on having served jail time notwithstanding his innocence or, at best, on having served time in excess of the one-year sentence on his 2006 conviction.
Moreover, Poventud's briefs to the three-judge panel everywhere declare his innocence and attack the reliability of his guilty plea. So do his papers on rehearing in banc: Poventud repeats the claim that his guilty plea was obtained through coercion, and thus is entitled to no credence: "[Poventud's] allocution to the `facts' consisted of answering `yes,' unsworn, to the court's summary of the allegations against him." Appellant's Br. 16-17.
Throughout this entire litigation, then, Poventud has categorically insisted that he is innocent of any participation in the Duopo robbery and that his 2006 plea was obtained through coercion. The majority opinion, for reasons easy to understand, undertakes to recast Poventud's claim as *154seeking damages only for the procedural impairment of his original trial, and posits that Poventud might just prevail if he follows the majority’s lead and seeks damages on different allegations of fact for a different loss and on a different theory. But Poventud, a savvy and counseled litigant, is the “master of the complaint,” and his allegations, submissions, and underlying theories of liability and damages should be taken at face value. Holmes Grp., Inc. v. Vornado Air Circulation Sys., Inc., 535 U.S. 826, 831, 122 S.Ct. 1889, 153 L.Ed.2d 13 (2002) (internal quotation marks omitted). Now, after two amendments to Poventud’s complaint, a summary judgment motion on Heck grounds, a thorough district court opinion, a full appeal to a three-judge panel (complete with dissent), and consideration in banc after further briefing and oral argument, it is not premature to decide this case on the consistently-argued allegations and theories of the plaintiff.
Poventud’s guilty plea — placing him at Duopo’s shooting, armed, with the intent to commit robbery' — simply “c[an]not be reconciled with the claims of his civil action,” VanGilder v. Baker, 435 F.3d 689, 692 (7th Cir.2006): Poventud swears that he “was nowhere near the crime scene at Oliver Place and Marion Avenue.” Poventud Aff. ¶ 6. Thus, Heck plainly bars this action unless Poventud can invoke some Heck exception, a proposition we foreclose.3 See infra Point VII.
IV
But let us assume that Poventud is not the master of his Complaint, and that he seeks § 1983 damages (as the majority would have it) only for the procedural flaw in his 1998 trial. The Heck bar still forestalls Poventud from going forward.
In urging otherwise, the majority maintains that neither Poventud’s extant 2006 conviction nor his vacated 1998 conviction erects a Heck bar to his pursuit of Brady damages. Specifically, the majority concludes that a damages award for a Brady violation in connection with Poventud’s 1998 conviction (after trial) would not impugn the integrity of the conviction in 2006 (based on the guilty plea). Further, the majority concludes that the vacatur of the 1998 conviction, even with a remand for retrial, was sufficient to satisfy the favorable termination predicate of Heck not*155withstanding our holding in DiBlasio v. City of New York, 102 F.3d 654 (2d Cir. 1996), because Poventud’s Brady claim does not sound in malicious prosecution and is, in any event, more focused on withheld evidence than on actual innocence.
The majority’s analysis is premised on a fundamental distortion of Brady. The goal of Brady is to advance the truth at trial and to promote a result consistent with underlying guilt or innocence; the evil of a Brady violation is that it saps confidence in the verdict and impairs the fairness of the trial in terms of its substantive outcome. “Our Court and others have long recognized that Brady violations obscure a trial’s truth-seeking function.... ” United States v. Mahaffy, 693 F.3d 113, 134 (2d Cir.2012). “The message of Brady and its progeny is that a trial is not a mere ‘sporting event’; it is a quest for truth in which the prosecutor, by virtue of his office, must seek truth even as he seeks victory.” Monroe v. Blackburn, 476 U.S. 1145, 1148, 106 S.Ct. 2261, 90 L.Ed.2d 706 (1986) (Marshall, J., dissenting from denial of certiorari). Brady was formulated to advance the search for truth, not to provide a guilty defendant with a sporting chance at acquittal; for that reason, the Brady Court expressly refused to raise a “sporting theory of justice” to “the dignity of a constitutional right.” 373 U.S. at 90, 83 S.Ct. 1194.
Accordingly, there is no Brady deprivation absent a concern that the truth-finding function of the trial has been thwarted. See infra Point V.A. That is why prosecutors have no constitutional obligation to disclose “any information that might affect the jury’s verdict”; as the Supreme Court has emphasized, such a “constitutional standard of materiality approaches the ‘sporting theory of justice’ which the Court expressly rejected in Brady.” United States v. Agurs, 427 U.S. 97, 108, 96 S.Ct. 2392, 49 L.Ed.2d 342 (1976); see also Brady, 373 U.S. at 90, 83 S.Ct. 1194.
The majority assumes that Brady is a rule of procedure detached from its ultimate goal. That leads the majority to allow § 1983 damages even though the undisclosed evidence is (as we now know) not material to innocence or the seriousness of the crime, and even though the evidence would have been helpful only to strengthen Poventud’s perjurious alibi. This error permeates the majority opinion, turning all its meticulous analysis to error and subverting Brady itself.
This is a mistake, and a serious one. The majority reconceives Brady as a device for preserving the defendant’s odds of winning an acquittal by any means, and by perjury in particular. The majority’s faulty premise thereby corrupts Brady, and diminishes it.
Y
The “truth-finding function” of Brady inheres in the elements of a Brady-based § 1983 action — namely, materiality, causation, and damages. But none of these elements can be proven without impugning Poventud’s guilty plea, and that tactic is blocked by Heck. I take up each element in turn.
A
As to materiality: the constitutional right defined by Brady and its progeny is the criminal defendant’s procedural due process right to the disclosure of “evidence that is material to his guilt or punishment.” Cone v. Bell, 556 U.S. 449, 469, 129 S.Ct. 1769, 173 L.Ed.2d 701 (2009) (citing Brady, 373 U.S. at 87, 83 S.Ct. 1194). “[E]vidence is ‘material’ within the meaning of Brady when there is a reasonable probability that, had the evidence *156been disclosed, the result of the proceeding would have been different.” Id. at 469-70, 83 S.Ct. 1194. So, to establish a Brady violation at his civil trial, Poventud must show not only that the impeachment evidence at issue is favorable to him and was undisclosed, but that it is “material either to guilt or to punishment.” Brady, 373 U.S. at 87, 83 S.Ct. 1194.
This is always a retrospective determination, as nondisclosure is material only when “the favorable evidence could reasonably be taken to put the whole case in such a different light as to undermine confidence in the verdict.” Kyles v. Whitley, 514 U.S. 419, 435, 115 S.Ct. 1555, 131 L.Ed.2d 490 (1995). Brady's materiality standard thus implements the underlying purpose of Brady itself: to “ensure that a miscarriage of justice does not occur.” See United States v. Bagley, 473 U.S. 667, 675, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985). As the Supreme Court has emphasized, “[t]he proper standard of [Brady ] materiality must reflect our overriding concern with the justice of the finding of guilt.” Agurs, 427 U.S. at 112, 96 S.Ct. 2392 (1976); see also United States v. Coppa, 267 F.3d 132, 139 (2d Cir.2001) (holding that the “essential purpose” of Brady is to “ensur[e] the reliability of [a] criminal verdict”).
Consistent with Brady's focus on the reliability of criminal judgments, a prosecutor’s failure to turn over exculpatory or impeachment evidence is a Brady violation rising to the level of constitutional error only when this failure “undermine[s] confidence in the outcome of the trial.” Bagley, 473 U.S. at 682, 105 S.Ct. 3375; see also Kyles, 514 U.S. at 434, 115 S.Ct. 1555 (citing Bagley, 473 U.S. at 678, 105 S.Ct. 3375) (vacatur required where verdict is not “worthy of confidence”). The mere failure to disclose favorable evidence is not enough, because such a rule “would impose an impossible burden on the prosecutor and would undermine the interest in the finality of judgments.” Bagley, 473 U.S. at 675 n. 7, 105 S.Ct. 3375. Moreover, if nondisclosure alone were sufficient, independent of any concern about the reliability of the ultimate outcome, Brady damages could be recovered even by a defendant who was acquitted — a proposition that several courts of appeals and district courts in our Circuit have rejected.4
Accordingly, as the Supreme Court said in Strickler v. Greene, a Brady claim is not made out by showing “any breach of the broad obligation to disclose disculpatory evidence,” because “there is never a real ‘Brady violation’ unless the nondisclosure was so serious that there is a reasonable probability that the suppressed evidence would have produced a different verdict.” 527 U.S. 263, 281, 119 S.Ct. 1936, 144 L.Ed.2d 286 (1999). Thus, “[i]f there is no *157reasonable doubt about guilt whether or not the additional evidence is considered,” the Supreme Court said in Agurs, “there is no justification for a new trial.” 427 U.S. at 112-13, 96 S.Ct. 2392.
The state court vacated Poventud’s 1998 conviction on the ground that the State’s failure to disclose Duopo’s initial identification of Poventud’s brother, Francisco, eroded confidence in the verdict. On the record before the state court, confidence was impaired because the nondisclosure had bearing on the accuracy of the critical identification made by the victim, and (reciprocally) on Poventud’s alibi defense. In short, the withheld information was material from that court’s perspective in time.
Poventud, however, will be unable to rely on the materiality finding of the state court in this § 1983 suit. There can be no estoppel because none of the defendants (the police officers, the district attorney, and the City) were parties in the criminal appeal, and no defendant here is in privity with any litigant in the criminal appeal. See Brown v. New York, 60 N.Y.2d 897, 898-99, 470 N.Y.S.2d 573, 458 N.E.2d 1250 (1983) (concluding that issue preclusion did not apply against the defendant municipality in a civil action for false arrest and assault based on dismissal of a criminal charge because the district attorney and the municipality do not “stand in sufficient relationship to apply the doctrine”); see also Stancuna v. Sherman, 563 F.Supp.2d 349, 353-54 (D.Conn.2008) (“Although the Second Circuit does not appear to have expressly so held, a number of other circuits have held that government employees in their individual capacities are not in privity with their government employer.” (collecting cases)).5
Poventud, therefore, will be required to prove by a preponderance that the nondisclosure was material, ie., that it caused a result that is wrong or unworthy of confidence. But his own guilty plea forecloses that possibility. It establishes — beyond doubt — that the undisclosed impeachment evidence could only have been used at Poventud’s trial to insinuate falsely that Younis Duopo, a truthful witness offering an accurate identification of Poventud as his robber, should not be believed. In short, the plea establishes that the supposed Brady evidence is wholly immaterial.
The plea gives the necessary assurance categorically, because the nondisclosure that justified vacatur by the state court in 2005 no longer calls into question the correct resolution of the only issue on which this nondisclosure had bearing. The victim’s identification of Poventud was sound. The failure to provide Poventud with impeachment material with which to challenge that identification, moreover, is shown to be immaterial by virtue of Poventud’s own solemn admission. Poventud cannot have it both ways: he cannot state that he is guilty, that he was present on the day in question and participated in the crime, but that he was nonetheless prejudiced at his trial by the nondisclosure of evidence that could have helped him only by suggesting that the accurate testimony of the victim should not be believed. This *158conclusion flows from the meaning and purpose of Brady.
The majority disregards Poventud’s guilty plea and seeks to focus only on the vacatur of the 1998 judgment. The Supreme Court, however, has counseled against such a blinkered approach. Thus, in Lockhart v. Fretwell, 506 U.S. 364, 113 5.Ct. 838, 122 L.Ed.2d 180 (1993), a habeas petitioner alleged ineffective assistance because his counsel failed to interpose an objection based on circuit precedent that was later overruled. Notably, the standard for ineffective assistance is the same retrospective standard that is used to assess Brady materiality: namely, whether there is a “reasonable probability” that but for the claimed error, the result would have been different. Strickland v. Washington, 466 U.S. 668, 694, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). Fretwell, like Brady, rejects the sporting chance approach to the criminal trial and focuses on the justice of the ultimate result.6 Fretwell concluded that the outcome of the proceeding was correct despite counsel’s failure to object: the petitioner had not been deprived of constitutionally effective assistance, only of “the chance to have the state court make an error in his favor.” 506 U.S. at 371, 113 S.Ct. 838 (internal quotation marks omitted). The petitioner could not premise a constitutional claim, the Court concluded, on “a windfall to which the law does not entitle him.” Id. at 370, 113 S.Ct. 838. The same retrospective look, under the same standard, yields the same result for Poventud: he would receive an impermissible windfall if afforded damages for the nondisclosure of impeachment material that he could only have used to make accurate testimony appear unreliable.
B
As to causation: Poventud must also satisfy the elements of the § 1983 action derived from the common law of torts — • specifically, causation. See Smith v. Wade, 461 U.S. 30, 34, 103 S.Ct. 1625, 75 L.Ed.2d 632 (1983). “The Supreme Court has made it crystal clear that principles of causation borrowed from tort law are relevant to civil rights actions brought under section 1983.” Warner v. Orange Cnty. Dep’t of Prob., 115 F.3d 1068, 1071 (2d Cir.1996) (alteration and internal quotation marks omitted). Poventud’s claim therefore cannot be salvaged by recasting it (as the majority does) as one potentially seeking nominal damages for no more than a violation of procedural due process. To recover money damages of even one dollar, Poventud must prove that the undisclosed material was both the factual and the proximate cause of the harm he has identified: wrongful imprisonment. And causation must be shown even on the majority’s theory that the harm Poventud suffered was the mere inconvenience of standing trial. Poventud cannot sustain this burden without challenging his guilty plea and the resulting 2006 conviction, which is barred by Heck.
First, Poventud must show that the constitutional violation that he alleges was an actual cause of his injury. In the Brady context, the causation inquiry “essentially replicates the materiality inquiry with a heightened burden of proof’: that is, “[hjaving already shown a reasonable probability that he wóuld not have been convicted but for the withholding of evidence, a plaintiff must then make the same *159showing by a preponderance of the evidence.” Drumgold v. Callahan, 707 F.3d 28, 49 (1st Cir.2013). Poventud, who cannot establish materiality without impugning his guilty plea, is likewise blocked from proving factual causation.
Second, to prevail on his Brady claim, Poventud must prove that the failure to give him impeachment material was a proximate cause of his harm, whether the harm claimed is prison, separation from family and friends, the inconvenience of sitting through his trial, or some sort of risk premium for the increased chance of conviction or a longer sentence. “[I]n all § 1983 cases, the plaintiff must prove that the defendant’s action was a proximate cause of the plaintiffs injury.” Gierlinger v. Gleason, 160 F.3d 858, 872 (2d Cir.1998).
Although proximate cause is generally a question to be determined by the trier of fact, “where the actual cause of the injury is undisputed, ... proximate cause ... is a question of law for the court.” Caraballo v. United States, 830 F.2d 19, 22 (2d Cir. 1987) (citations omitted). The proximate cause inquiry focuses on “whether a cause is a substantial factor in bringing about the harm, or whether the cause is too remotely or insignificantly related to the harm to be a legal basis for liability.” Henrietta D. v. Bloomberg, 331 F.3d 261, 278-79 (2d Cir.2003) (citations and internal quotation marks omitted).
This proximate cause determination has a moral dimension because proximate cause recognizes only those causal factors that society is prepared to hold legally responsible for a given consequence. See Anza v. Ideal Steel Supply Corp., 547 U.S. 451, 467, 126 S.Ct. 1991, 164 L.Ed.2d 720 (2006) (“That is, to recover, a plaintiff must show ... that his injury is sufficiently connected to the tort that ‘the moral judgment and practical sense of mankind [will] recognize responsibility in the domain of morals----’”) (quoting Sutherland, Law of Damages 18 (1882)); Dobbs et al., The Law of Torts § 185, at 622 (2d ed. 2011) (“[P]roximate cause is not about causation at all but about the significance of the defendant’s conduct or the appropriate scope of liability in light of moral and policy judgments about the very particular facts of the case.”).
Poventud cannot establish proximate cause in a § 1983 trial without impugning his guilty plea. That is because he must show that the State’s failure to provide him with impeachment evidence was a substantial factor in causing him injury, and a factor that renders damages appropriate as a matter of law. But as already established at some length, the undisclosed evidence here could only have been useful to Poventud in one very particular way: to support an inference that Poventud was elsewhere at the time of the crime. Poventud has now solemnly admitted that this inference is wholly false. Moreover, Poventud’s theory of proximate causation ignores the obvious point that his alleged injury was caused by his own participation in the crime. To find proximate cause on such facts would read moral judgment out of the proximate cause determination just as a finding of materiality would embrace the “sporting chance” approach to the criminal trial. Brady, 373 U.S. at 90, 83 S.Ct. 1194 (refusing to accord a trial strategy of this sort “the dignity of a constitutional right”). Accordingly, Poventud cannot prove proximate cause without impugning his guilty plea and inviting the court to disbelieve it.
C
As to damages: the majority opinion appears to be the first to hold that money damages may be awarded for an alleged Brady violation occurring at the trial of a criminal defendant who thereafter pleads *160guilty (or is convicted at retrial) of the same underlying crime (or any lesser included offense).7 Our Court has twice considered whether § 1983 damages are available in analogous cases. We affirmed the dismissal of two § 1983 complaints claiming such damages: when a vacated conviction was compromised by a subsisting guilty plea, and when an initial charge was resolved by a plea to a lesser included offense. In both cases, we deemed that result sufficiently evident that we decided the issue by summary order. See McNeill v. People of City & State of N.Y., No. 06-CV-4843, 2006 U.S. Dist. LEXIS 77085, 2006 WL 3050867, (E.D.N.Y. Oct. 24, 2006), summarily aff'd, 242 Fed.Appx. 777, 778-79 (2d Cir.2007) (“Although Appellant’s state court conviction was vacated, his subsequent guilty plea stands as a bar, under Heck, to a § 1983 action.”); Papeskov v. Brown, No. 97 Civ. 5351, 1998 WL 299892, at *5 (S.D.N.Y. June 8, 1998) (Sotomayor, J.) (“[A] plea of guilty, even to a charge lesser than that for which the plaintiff was arrested, bars a § 1983 action.”), summarily aff'd, 173 F.3d 845 (2d Cir.1999).
The only district judge in this Circuit to deal with identical facts viewed it as elementary that Heck would bar the entirety of the plaintiffs Bracfy-based § 1983 claim, however framed: “Even if plaintiff seeks damages solely for any ‘extra’ time served, it nevertheless imputes an illegitimacy to her plea and sentence. We do not see any basis for, or find any authority supporting, the separation of these two periods of imprisonment for purposes of a § 1983 action.” Stein v. Cnty. of Westchester, 410 F.Supp.2d 175, 179 (S.D.N.Y.2006) (Conner, J.). No authority for the majority’s position has materialized since Judge Conner decided Stein.
To claim damages based on imprisonment is inherently difficult given Poventud’s guilty plea to holding up Mr. Duopo. As a backup theory, the majority opinion recognizes as valid Brady claims that “did not result in concrete injury.” Maj. Op., ante, at 135. But Brady is not a pure process claim. If it were, criminal defendants could claim damages based on a monetization of the increased probability of conviction they faced by reason of the suppression of the evidence, regardless of whether the prosecution ended in acquittal or conviction. This, of course, would defy Heck, which was itself a Brady claim, as well as Amaker v. Weiner, 179 F.3d 48, 51 (2d Cir.1999), which the majority purports to follow. More importantly, this approach is incompatible with the purpose of Brady, which is to ensure confidence in the outcome of criminal proceedings in terms of guilt or innocence. Brady’s materiality standard requires a showing of prejudice that inherently looks to whether the defendant was, in fact, concretely injured.8
*161VI
Heck bars Poventud’s claim for the additional reason that there has been no favorable outcome of the 1998 conviction within the meaning of that precedent.
The 1998 judgment was neither reversed nor “expunged.” Maj. Op., ante, at 136-37. Reversal would mandate entry of an opposite judgment that dismisses the indictment, and an expungement would obliterate, wipe out and annihilate the conviction, whereas, in fact, Poventud’s 1998 conviction was “vacated” in contemplation of a retrial. That was a contingent rather than a final outcome. Vacatur is not necessarily an “outcome” if an outcome is how a proceeding comes out at the end.
As the majority concedes, vacatur of Poventud’s 1998 conviction with remand for retrial is not a final favorable termination as that term was understood at common law, because at common law a final favorable termination meant “that the proceeding cannot be brought again,” and no final favorable termination is obtained when a prosecution is “abandoned pursuant to a compromise with the accused.” Maj. Op., ante, at 131; see also id. at 136.
Moreover, the vacatur of the 1998 judgment cannot be deemed a favorable outcome under Heck without uncoupling the vaeataur from the guilty plea to which it led. It would follow from that analysis — • as Poventud conceded at oral argument and as the majority concedes — that a vacatur is a favorable outcome for Heck purposes even if it is followed by a plea to the very same offenses as the vacated conviction (rather than to a lesser included offense), and even if it is followed by conviction on the very same offenses after a retrial. That is counterintuitive. Mere vacatur can develop into a favorable outcome if the prosecution is abandoned and the charges against the defendant are dismissed. However, vacatur does not yield a favorable outcome when, as here, original charges are compromised pursuant to a plea agreement that results in a conviction for a lesser included crime. In DiBlasio, this Court expressly held that a conviction for a lesser included offense after vacatur does not constitute a favorable termination for purposes of Heck. I think most criminal defendants would agree that a vacatur leading only to retrial or a plea is, generally speaking, an outcome that can be considered a complete victory only for defense counsel.
The majority recites that it complies with the favorable termination rule of DiBlasio, but then tries to narrow the rule to the particular constitutional claim there at issue: malicious prosecution. This narrowing fails because Heck, itself premised on Brady error, drew an analogy to malicious prosecution requirements — an analogy that, as the majority recognizes, was not coincidental but “continues throughout Heck.” Maj. Op., ante, at 130. Heck explains that, under common law, “a cause of action for malicious prosecution does not accrue until the criminal proceedings have terminated in the plaintiffs favor,” and concluded that “so also a § 1983 cause of action for damages attributable to an unconstitutional conviction or sentence does not accrue until the conviction or sentence has been invalidated.” 512 U.S. at 489-90, 114 S.Ct. 2364. Those terms — “terminated in plaintiffs favor” and “invalidated”— are synonymous, rather than distinct. See DiBlasio, 102 F.3d at 659 (“If interpreted literally, this sentence would seem to mean that any time a conviction is overturned by a writ of habeas corpus there has been a final determination in favor of the accused. *162We are not convinced that this is what the [Heck] Court intended.”); cf. Heck, 512 U.S. at 493, 114 S.Ct. 2364 (Souter, J., concurring) (noting that the Heck majority “transplanted” the common law favorable termination requirement to § 1983 claims that impugn an extant conviction).
To explain away incompatible precedents, a footnote in the majority opinion suggests a division of Brady claims between (1) those involving withheld exculpatory evidence that could have proved innocence and thus “do suggest a malicious prosecution claim[,]” and (2) those that, like Poventud’s, are “less concerned with ... innocence and [that] instead focus[ ] on ‘evidence that an identifying witness was unreliable, and evidence impeaching the credibility of significant' prosecution witnesses.’ ” Maj. Op., ante, at 137 n. 20 (quoting Second Am. Compl. ¶ 128). According to the majority, the first subset requires a final favorable termination as understood by the common law and this Court in DiBlasio, and the second does not. Id.
The distinction that the majority draws does not favor its result. The majority thinks that the Complaint does not suggest malicious prosecution; but the Complaint describes a nefarious “police cover-up” leading to a “wrongful attempted murder and robbery conviction.” Second Am. Compl. ¶ 1. Thus it is alleged that the police targeted Poventud with no evidence of his guilt (much less probable cause), id. ¶¶ 14-17, 24, 35-46, purposely failed to investigate leads that would have exonerated him, and withheld evidence that would have impeached the victim’s identification.9
The majority thinks that the Complaint is “less concerned with innocence” and more concerned with witness impeachment. But the Complaint flatly alleges that the suppressed identification is “evidence of innocence.” Second Am. Compl. ¶ 128. And it is alleged to be “evidence of innocence” because, “[a]t the time of the crime, [Poventud] did not physically resemble his brother, nor did [he] resemble [his brother] as he was depicted in the old photograph identified by Duopo.” Id. ¶ 25. Thus, although the evidence that was withheld would have been useful to impeach Duopo’s credibility, it is, at bottom, evidence that Poventud was not Duopo’s assailant. These allegations, which run throughout the Complaint, certainly “sound in” malicious prosecution, though the claim is ultimately brought under Brady.10
The majority’s distinction between Bra-&/-exculpatory claims and Bra(%-impeaehment claims is, in any event, novel and unworkable. It implies that a defendant may bear a heavier Heck burden in pursuing a Brady claim if the withheld evidence is actually exculpatory than if it is merely impeaching. Even a mediocre lawyer can blend one of these leaky categories into the other. Moreover, what can justify this curious distinction other than the “sporting chance” view of Brady that has been expressly rejected by the Supreme Court?
*163The proper distinction to be drawn from Heck and its progeny is not between malicious prosecution and Brady claims, much less between Brady-exculpatory and Bradi/-impeachment claims. Rather, it is between (1) constitutional claims that impugn a conviction (whether because the plaintiff claims he is innocent or because he claims the trial’s substantive outcome cannot be trusted) and (2) constitutional claims that do not (of which excessive force claims are the most obvious example). When dealing with the former, as we do here, we should faithfully apply our favorable termination precedent, which avoids the pitfalls, inconsistencies, and surprises of the majority’s approach.11
Precedent compels us to conclude that the Heck bar blocks Poventud’s claim. Poventud’s criminal proceeding did not terminate until he pled guilty to a lesser included offense. DiBlasio, 102 F.3d at 658. Therefore, Poventud’s Brady-based § 1983 claim “does indeed call into question the validity of his conviction.” Amaker, 179 F.3d at 51.
VII
Because we conclude that Poventud’s claim necessarily implies the invalidity of his extant conviction, we reach the issues that launched this rehearing in banc: whether the Heck bar applies only to persons in custody, as the majority of the three-judge panel held; whether there are any exceptions to the Heck bar; and whether any exceptions that may exist would save Poventud’s claim. We reject the holding of the majority opinion issued by the three-judge panel, an opinion which has in any event been vacated. Assuming arguendo that there are some exceptions to Heck, we conclude that Poventud’s action could not come within them.
On the basis of self-described dicta signed by five Supreme Court Justices (three of whom are no longer on the Court), a Circuit split has opened as to whether some exceptions to Heck may be permitted. In a nutshell, these Justices posited that “a former prisoner, no longer ‘in custody,’ may bring a § 1983 action establishing the uneonstitutionality of a conviction or confinement without being bound to satisfy a favorable-termination requirement that it would be impossible as a matter of law for him to satisfy.” Spencer v. Kemna, 523 U.S. 1, 21, 118 S.Ct. 978, 140 L.Ed.2d 43 (1998) (Souter, J., concurring) (emphasis added).
Several Circuits have concluded that the Spencer concurrences cannot override Heck’s binding precedent. See, e.g., Entzi v. Redmann, 485 F.3d 998, 1003 (8th Cir. 2007); Gilles v. Davis, 427 F.3d 197, 209-10 (3d Cir.2005); Randell v. Johnson, 227 F.3d 300, 301 (5th Cir.2000) (per curiam); Figueroa v. Rivera, 147 F.3d 77, 80 (1st Cir.1998). These courts hold that Heck’s bar is absolute, heeding the Supreme Court’s admonition that, even if binding precedent “appears to rest on reasons rejected in some other line of decisions, the Court of Appeals should follow the case which directly controls, leaving to [the Supreme] Court the prerogative of overruling its own decisions.” Agostini v. Felton, 521 *164U.S. 203, 237, 117 S.Ct. 1997, 138 L.Ed.2d 391 (1997) (internal quotation marks omitted).
Other Circuits have nevertheless held that SpencePs dicta allows courts to recognize unusual and compelling circumstances in which Heck’s holding does not absolutely foreclose a claim. See, e.g., Burd v. Sessler, 702 F.3d 429, 435-36 (7th Cir. 2012); Cohen v. Longshore, 621 F.3d 1311, 1317 (10th Cir.2010); Wilson v. Johnson, 535 F.3d 262, 267-68 (4th Cir.2008); Powers v. Hamilton Cnty. Pub. Defender Comm’n, 501 F.3d 592, 603 (6th Ch\2007); Guerrero v. Gates, 442 F.3d 697, 704 (9th Cir.2006); Harden v. Pataki, 320 F.3d 1289, 1298 (11th Cir.2003).
There is no need to choose a side in this split because the narrow exception articulated by Justice Souter would be inapplicable here in any event. The motivating concern in the Spencer dicta was that circumstances beyond the control of a criminal defendant might deprive him of the opportunity to challenge a federal constitutional violation in federal court. Poventud is not such a person.
Poventud challenged his first conviction in state court and won — making it unnecessary for him to. seek federal habeas relief. At that point, Poventud had the option of defending in an untainted trial or of pleading guilty to the same crime on reduced charges and accepting a reduced sentence. He chose to plead. Poventud then had the option of filing a motion to challenge the voluntariness of his plea — • and Poventud did so, but he withdrew it prior to an evidentiary hearing. It was therefore by no means “impossible as a matter of law,” Spencer, 523 U.S. at 21, 118 S.Ct. 978 (Souter, J., concurring), for Poventud to challenge his conviction and thereby satisfy Heck’s favorable termination requirement; he simply decided not to.
On this one point, the full in banc court seems to be unanimous. The majority disclaims any occasion to “reach the broader issue on which the panel rested its decision[,]” Maj. Op., ante, at 136 n. 19, which is that the Heck bar does not survive the release of the plaintiff from custody, see Poventud, 715 F.3d at 60. The majority opinion nevertheless acknowledges that, if Poventud’s Brady claim were cast in terms of malicious prosecution, it would be barred by DiBlasio, which is of course a Heck-bar case. That could not happen if (as the majority of the three-judge panel held in this case) the Heck bar operates only so long as a § 1983 plaintiff is in jail, and is removed when he is at liberty (as Poventud is and has been). In acknowledging that their analysis “circumseribe[s] Poventud’s Brady-based § 1983 claim” in several ways, the majority acknowledges the bar to a civil claim challenging the subsisting 2006 judgment. Thus, notwithstanding that Poventud is as free as any of us, the majority’s footnote 20 reflects the holding that certain of his claims could well be “barred” by Heck.12 Maj. Op., ante, at 137 n. 20. Similarly, the majority’s footnote 22 is at pains to deny that a § 1983 judgment in favor of Poventud would impugn his 2006 conviction, see id. at 137 n. 21-a consideration that would be obviated but for the bar of Heck. The majority may say and claim passim that it does not decide whether release from custody removes the Heck bar, see also id. at 125 n. 1; id. at 127 n. 7, but the text and mandate of the majority opinion suggest otherwise. It is useless to deny gravity while falling.
*165The majority opinion thus necessarily rejects the idea that, once a criminal defendant is at liberty, Heck no longer bars § 1983 claims challenging subsisting judgments — the Heck analysis that we went in banc to reconsider.
VIII
The majority erodes Heck and corrupts Brady by adopting a deeply flawed notion of due process — due process as a “sporting chance.” This holding will have consequences, none of them salutary.
The moral force of a guilty plea will no longer “quite validly removef ] the issue of factual guilt from the case,” Menna, 423 U.S. at 62 n. 2, 96 S.Ct. 241 (emphasis omitted), but will be merely an admission to be evaded in § 1983 lawsuits impugning the results of extant state criminal proceedings. Individuals who have been fairly convicted of serious crimes will seek and receive damages for being deprived of a better opportunity for perjury, while people who are actually innocent and exonerated based on new evidence have no cause of action for damages — not to mention the victims of crime such as Mr. Duopo, shot in the neck while on the job. This case illustrates why the sporting chance theory of criminal justice that was rejected by the Brady Court is beneath the dignity of a constitutional right.
CONCLUSION
For the foregoing reasons, I would affirm the decision of the district court.
. Judge Lynch argues that the one-year sentence of the plea-bargain here is an “approximation of the result that would have come from a fair trial,” Concurring Op. of Judge Lynch, ante, at 142, as though a prosecutor would not take account of the time already served in negotiating a plea following vacatur.
. Judge Lynch is radically skeptical about the plea, even though it is the only fact that is authoritatively settled; but he treats as true (and deplores) (1) findings of the state court that were discredited by the plea, and that in any event do not bind any defendant here, see infra at 144-45 & n. 5, (2) allegations of the Complaint, which we treat as true solely as a tool of analysis, and (3) sworn statements made by Poventud for the purpose of litigation. Judge Lynch subjects only the plea itself to metaphysical tests of knowability.
. As to Judge Chin’s opinion, we are in accord with his doctrinal analysis in Point I, supra. We respectfully differ with Judge Chin, however, as to the limited reversal he proposes. Judge Chin identifies the following surviving Brady “claims”: misleading the district attorney when he gauged the strength of the case (pretrial and post-vacatur); and misleading the state court in the setting of bail. See Concurring Op. of Judge Chin, ante, at 149-50. However, Brady is a trial right, formulated to safeguard the fairness of trial outcomes; it does not require disclosure of impeachment evidence during pretrial events, however critical. See United States v. Ruiz, 536 U.S. 622, 633, 122 S.Ct. 2450, 153 L.Ed.2d 586 (2002) (holding that the failure to disclose impeachment evidence prior to a guilty plea does not amount to a Brady violation); Friedman v. Rehal, 618 F.3d 142, 154 (2d Cir.2010) (considering that Ruiz’s reasoning likely also extends to exculpatory evidence) (citing 6 LaFave, et al., Criminal Procedure § 24.3(b), at 369 (3d ed.2007)). By the same token, the Brady disclosure obligation is not (as Judge Chin assumes) a defendant’s right at the preliminary evaluation by a prosecutor or in a bail hearing.
In any event, the "claims” Judge Chin purports to uphold are not claims or causes of action; they are theories of damages posited in aid of an impossible cause of action. Whether a Brady violation occurred is a distinct inquiry from whether a particular harm — such as the denial of bail — flowed from that violation. And, as we demonstrate below, the former, critical inquiry cannot be answered in Poventud’s case without impugning his guilty plea.
. See Livers v. Schenck, 700 F.3d 340, 359 (8th Cir.2012) (stating that “there was no Brady violation because [the plaintiffs] were not convicted” where plaintiffs were acquitted); accord Morgan v. Gertz, 166 F.3d 1307, 1310 (10th Cir.1999); Cannistraci v. Kirsopp, No. 1:10-cv-980, 2012 U.S. Dist. LEXIS 68399, 2012 WL 1801733 (N.D.N.Y. May 16, 2012); Ambrose v. City of New York, 623 F.Supp.2d 454, 467-71 (S.D.N.Y.2009); see also Flores v. Satz, 137 F.3d 1275, 1278 (11th Cir.1998) (per curiam) (no Brady § 1983 claim following prosecution's determination not to prosecute); McCune v. City of Grand Rapids, 842 F.2d 903, 907 (6th Cir.1988) (no Brady violation where charges were dismissed before trial); Grenier v. Jonas, No. 1:09-CV-121, 2010 U.S. Dist. LEXIS 20658, 2010 WL 883743 (D.Vt. Mar. 5, 2010) (same); cf. Mosley v. City of Chicago, 614 F.3d 391, 397 (7th Cir.2010) (reviewing other circuits’ case law holding that "a trial that results in an acquittal can never lead to a claim for a Brady violation because the trial produced a fair result, even without the exculpatory evidence,” but not deciding the issue); Smith v. Almada, 640 F.3d 931, 941-42 (9th Cir.2011) (Gwin, D.J., specially concurring).
. Stancuna interpreted the preclusive effect of an earlier federal judgment. Although New York law determines the preclusive effect of a judgment entered in New York, see Migra v. Warren City Sch. Dist. Bd. of Educ., 465 U.S. 75, 81, 104 S.Ct. 892, 79 L.Ed.2d 56 (1984), there is "no significant difference between New York’s preclusion law and federal preclusion law,” Pike v. Freeman, 266 F.3d 78, 90 n. 14 (2d Cir.2001); see also Marvel Characters, Inc. v. Simon, 310 F.3d 280, 286 (2d Cir.2002) ("The parties agree that there is no discernible difference between federal and New York law concerning res judicata and collateral estoppel.”).
. The majority's discussion of Fretwell, tactically consigned to its footnote 16, rejects my characterization of the majority's approach as advancing a "sporting chance” theory. See Maj. Op., ante, at 134-35 n. 16. However, the footnote forthrightly lays out the majority's approach in a way that adopts and implements the sporting chance theory and demonstrates how it will operate on remand in this case. See id.
. One case approaches such a result. In Olsen v. Correiro, 189 F.3d 52 (1st Cir.1999), the plaintiff’s murder conviction was overturned due to the suppression of evidence at his first trial, and he subsequently entered a plea of nolo contendere to manslaughter. Id. at 55. The First Circuit rejected his request for incarceration-based damages, holding that this request was barred either because it went beyond “the limits of § 1983 actions” or because it was prohibited by the doctrine of proximate cause. Id. at 67-68. The Court nevertheless permitted — without much reasoning' — the recovery of $6,000, which appears to be composed of an attorney’s fee and a small punitive damages award.
. Thus, we disagree that the majority can successfully salvage Poventud's Complaint by recasting it as one for nominal damages. See Maj. Op., ante, at 135-36. Poventud's Brady "injury” could only be the frustration of his ability to bolster perjurious testimony, and thereby defeat the trial’s truth-seeking function — i.e., the “sporting chance” that is not afforded by Brady and its progeny. The line of cases cited in footnote 4, supra, supports this conclusion that nominal damages are not *161available for Brady violations (at least in this case). See supra at 156 n. 4.
. For example, the police "did not investigate [Poventud’s] alibi, but simply proceeded with processing [his] arrest." Second Am. Compl. ¶ 41. And when the gun used to shoot Duopo was linked to a later shooting, “[r]ather than take the risk that Duopo would identify Martinez [the suspect in the later shooting] and undercut their case against [Poventud], police did not show Martinez to Duopo in a photo array or a lineup." Id.
. The majority fails to suggest how district courts are to parse a complaint allegation-by-allegation to determine which Brady-based § 1983 claims are Heck-barred because they "sound in” malicious prosecution. This doctrine is high-maintenance as well as novel, unnecessary, and erroneous.
. Judge Lynch’s concurrence does little else but impugn the 2006 judgment. (Judge Calabresi and Judge Sack laid siege to it in the majority opinion of the three-judge panel, so that makes three judges to have done so.) To impugn the plea, Judge Lynch attacks the plea-bargaining process (as though pleas are not always the product of pressure), observes the imperfect reliability of eyewitnesses, puts in doubt the ability to know anything about human conduct, see Concurring Op. of Judge Lynch, ante, at 139-40, 144-45, and concedes the “legal[] validity]” of the plea only in grudging and perfunctory terms — chiefly in a footnote, id. at 141 n. 4.
. Insofar as the majority endeavors to limit Heck’s bar to malicious prosecution claims, I rely on this opinion’s discussion as to why that effort fails as a matter of doctrine and as it relates to Poventud’s claims specifically. See supra at 161-63.